®i{E ^. p. pm pfararg SbUT Cx.6 scji^ Date Due A THIS BOOK IS DUE ON THE DATE INDICATED BELOW AND IS SUB- JECT TO AN OVERDUE FINE AS POSTED AT THE CIRCULATION DESK. ^^59-') / y^ STUDIES IN SEEDS AND FRUITS STUDIES IN SEEDS AND FRUITS 'jin Investigation with the balance J (lA BY h!^1b!^UPPY, M.B., F.R.S.E. The old is still the true."— Austin Dobson. LONDON WILLIAMS AND NORGATE 14 HENRIETTA STREET, COVENT GARDEN, W.C. I912 LIBRARY N. C. State Ofullege I DEDICATE THIS BOOK TO THOSE WHO SO KINDLY APPRECIATED MY PREVIOUS WORK AMONGST THE PLANTS. IN GRATITUDE I HERE GIVE THEM OF MY BEST jLS^-^'oS PREFACE In the preparation of this work my indebtedness lay in many directions. The circumstances of my life enabled me to devote all my time to it, a very important condition for extensive original investigation. Then, in connection with West Indian plants, with which these pages are principally concerned, there was always in my hand Grisebach's Flora of the British West Indian Islands ^ the only accessible general work on the plants of those regions. Most of the German botanical works consulted were in English dress ; but amongst the exceptions was Nobbe's Handhich der Samenkunde^ a work essential for me to be familiar with, and I am deeply indebted to my wife for her assistance in mastering its contents. Then, again, as I rambled about on the coasts and in the inland woods of the West Indian Islands, or sat quietly working in my study at home, or lay in my cabin during the voyages to and fro across the Atlantic, ideas came floating through my mind which often took solid form and developed into lines of investigation before unsuspected. Some of them may have been echoes of my reading. For instance, from a recent re-perusal of Professor F. W. Oliver's address to the Botanical Section of the British Association in 1906, I find that 1 have unwittingly supplied answers to more than one of his suggestive queries. But there are other ideas that cannot be explained in this fashion, and they also have solidified and stand out boldly in the following pages. Amongst the sources to which I owe much, they are not the least. viii STUDIES IN SEEDS AND FRUITS Whilst arranging the results of my observations, the difficulty of establishing a nexus between them soon became apparent. One thing led to another in an irregular manner, the new line of inquiry being often determined by some accidental indication, or by some inconsistency in the results of experiments. Per- ceiving that it would not be conducive to method to follow the order of inception of the several inquiries, I devised the plan of arranging the materials to be now described. The shrinking and swelling processes of seeds were first discussed until the question of permeability or impermeability was raised so frequently that the matter had to be dealt with before further progress could be made. In its turn, the subject of permeable and impermeable seeds was treated on its own ground until the question of their hygroscopicity demanded investigation. So also the matters relating to the proportional weight of parts of fruits, and to the connection between the seed-number and the fruit-weight, were discussed until the disturbing influences of the abortion of ovules and the failure of seeds became so obvious that an inquiry into their nature was necessitated. Amongst the other difficult questions that presented them- selves in this inquiry was that relating to the unit of weight most suitable for seeds and fruits. It was soon found, how- ever, that the grain was by far the most fitting for my purpose, and it was accordingly adopted. The grain is not only one of the most ancient and one of the most extensively employed units of weight for small objects (such as precious stones), but it is Nature's primitive suggestion. Seeds of small size are in use as weights in the East at the present day ; and other persons besides myself must have been at times so circumstanced that they had to extemporise a balance and employ grains of rice as weights. This choice has enabled me to avoid the multiplicity of terms inseparable from most systems. Whether it happens that 60,000 grains go to weigh down a green coco-nut, or that 6000 Juncus seeds go to weigh down a grain, no other term of weight need be used. PREFACE ix Though in a few instances the grains have been converted into grammes, the use of percentages in stating the results in the great majority of cases will enable the reader to be largely independent of the unit of weight employed. My original plan was to include in this volume the results of my observations on the distribution of seeds by currents in the West Indian region, and through the agency of the Gulf Stream drift. However, this idea has been abandoned for at least two reasons. In the first place, such materials would have greatly added to the size of a book already large ; and, in the second place, since the subject was concerned with quite another matter, it could very well be treated in a separate volume. Though much of the work done in this direction has been put into shape, it has been decided to defer its publication ; and in the meantime I hope to considerably add to my facts relating to the occurrence of West Indian seed-drift on the Atlantic shores of Europe. This is the continuation of a study commenced by me in the Pacific about thirty years ago, and taken up from time to time in diflFerent parts of the tropics, many of the results being given in my book on Plant- Dispersal (1906), and in various papers enumerated in the list given on page xiii of that work. I would take this opportunity to ask any reader who is inter- ested in the occurrence of West Indian seeds on the west coast of Europe or on the neighbouring islands to communicate with me. Having studied the question during four winters in the home of the drift, in different parts of the West Indies, I wish to make as complete as possible the materials relating to this side of the Atlantic. These seeds are frequently stranded on the beaches of Europe, between the North Cape and the Straits of Gibraltar. Residents on the west coasts of Scandi- navia, Scotland, Ireland, England (south-west), France, Spain, and Portugal, and on the ofF-lying island groups, must often come upon these stranded seeds. The great majority of the seeds thus picked up do not come into the hands of any person who has studied the subject. Some of those who read X STUDIES IN SEEDS AND FRUITS these pages may be in a position to assist me in this inquiry, either by taking up the search themselves, or by lending me, for purposes of specific determination, any seeds in their possession, or by supplying information concerning collections of these seeds in the hands of private persons or in museums, or in giving me references to any literature (whether in the press or in scientific publications) dealing with the matter. Any materials left in my possession would subsequently, with the donor's permission, be distributed amongst the large museums. H. B. GUPPY, M.B. South Devon, January 31, 191 2. 3 RosARio," Salcombe, CONTENTS CHAl'. I. 2. 3- 4- 5- 6. 9- 10. 1 1. 12. 13- 14. 15. THE HISTORY OF THE INVESTIGATION THE THREE CONDITIONS OF THE SEED . THE IMPERMEABILITY OF SEEDS AND ITS SIGNIFICANCE PERMEABLE AND IMPERMEABLE SEEDS . THE GROUPING OF SEEDS ACCORDING TO THEIR PERMEA BILITY OR IMPERMEABILITY .... ADDITIONAL EVIDENCE ON THE CONTRAST IN BEHAVIOUR BETWEEN PERMEABLE AND IMPERMEABLE SEEDS HYGROSCOPICITY ....... A LAST WORD ON THE HYGROSCOPICITY OF SEEDS . THE REGIME OF THE SHRINKING AND SWELLING SEED THE FATE OF SEEDS AS INDICATED BY THE BALANCE A CLUE TO THE HOMOLOGIES OF FRUITS THE HOMOLOGIES OF FRUITS AS REVEALED IN THE DRYING PROCESS ........ THE DEHISCENCE OF FRUITS ..... THE PROPORTION OF PARTS IN FRLTITS . THE RELATION BETWEEN THE NUMBER OF SEEDS AND THE WEIGHT AND SIZE OF THE FRUIT 18 69 90 114 180 187 225 241 258 273 293 330 xii STUDIES IN SEEDS AND FRUITS -*-^^ PAGE X THE ABORTION OF OVULES AND THE FAILURE OF SEEDS . 344 . 368 17. SEED-COLORATION ..••'• 18. THE WEIGHT OF THE EMBRYO 4^3 19. THE REST-PERIOD OF SEEDS 4^7 20. THE COSMIC ADAPTATION OF THE SEED . . • -440 • 4^1 APPENDIX ..••••■■ 511 INDEX . • • • * STUDIES IN SEEDS AND FRUITS CHAPTER I THE HISTORY OF THE INVESTIGATION This investigation commenced as a study of the rest-period of seeds ; but its course has often been determined by small indications, the balance and the oven, aided by a sharp knife and a pocket-lens, being the only means of research employed. From the beginning it was for me a leap in the dark, since although investigators far abler than myself have written on the subject, there was little that seemed to offer a clue. Need- ing some firm ground to stand upon with reference to fruit- maturation and as regards the behaviour of the ripening seed, and of the seed entering the rest-period, I turned to the most authoritative works at my disposal, those of Goebel and Pfeffer. From the pages of the Organography of Plants I learned that the biology of the ripening fruit has hitherto scarcely received attention (English edition, 1900-5, ii. 570-571) ; and when consulting the Physiology of Plants with reference to the assump- tion of impermeability by many seeds when they enter the rest-period, I learned that the means by which the power of resistance to drying is gained and the changes which cause its loss are quite unknown (English edition, 1903, ii. 253). I suppose the reason why many have not ventured in this field is that there seemed no near prospect of obtaining tangible and serviceable results. It has in truth been for me like an I I 2 STUDIES IN SEEDS AND FRUITS exploring voyage in a little-known ocean where the lead rarely reaches the ocean's floor and " two thousand fathoms and no bottom" is a frequent record. Just as during my sojourn in the Pacific, the study of seed-buoyancy led me finally to discuss the history of the whole flora of those islands, so this work has been developed by slow degrees from my original observations on the vivipary of plants in that region. The possibility of all seeds being able to germinate on the plant presented itself to me whilst observing on the Hawaiian lava-plains the be- haviour of the seeds of Guilandina bonducella in the green pod ; and this led me to the study of the rest-period of seeds. But at first, to employ another simile, it was like a plunge into the depths of a primeval forest. My path branched ofF in a hundred different ways, tracks crossing and re-crossing and often leaving me in some tangled jungle. Instead of simplicity I found complexity, my inquiries taking me in all kinds of unexpected directions, the difficulty lying in the choice and in knowing when to retrace my steps. However, ulti- mately I emerged from the forest at a place far distant from where I entered, and have now the story of my experiences to tell. Perhaps I shall best explain the discursive character of this work and the variety of subjects handled, if I state briefly the various stages of my investigations. When observing the maturation and germination of the seeds of Guilandina bondu- cella in 1897 in Hawaii {Observations of a Naturalist in the Pacific^ ii. 191), I noted that in germinating these stone-like seeds assumed again the appearance of immaturity. The soft, moist seed from the green unopened pod and the soft, swollen seed on the eve of germination were both of them two or three times the size of the normal resting seed, and might at first sight be mistaken for each other. Surely, I argued, it would be possible, in the case of the seeds of this and other leguminous plants, by subjecting the pod on the plant to warm, humid conditions, to dispense with the rest-period altogether, and to bring about the germination on the plant of THE HISTORY OF THE INVESTIGATION 3 the large, soft, seemingly immature seeds before the drying and shrinking process that ushers in the rest-period begins. However, nearly ten years passed by before I saw my way to attacking the problem that first presented itself almost as in a dream during my sojourn in Hawaii. Here again it was from the same plant {Guilandina bonducelhi) that I obtained my clue. Whilst observing in December 1906 this plant in immature fruit on the beach of St Croix (Danish West Indies), it occurred to me that there was at least one indirect way of approaching the problem other than by carrying out an experi- ment more visionary than practicable in its nature. So I placed a number of the soft unripe seeds in wet sand, believing that under such moist conditions they would not go through the usual shrinking and hardening process. The experiment was completed at Black River, Jamaica ; and after five weeks I found that the seeds had retained their original size and consistence. The shrinking process had thus been deferred ; whilst in the case of seeds gathered at the same time and allowed to dry in the air the seed had been reduced by shrink- ing to about a third of its original weight. Just at this time, whilst studying the floating seed-drift of the Black River, I noticed that the shrinking process was con- sistently shirked by some of the river-side plants, the seeds. The mdica- after falling into the water, rapidly passing on into the germin- seTd-driftln'^ ating state whilst still afloat. They were not viviparous plants, Jamaica, but possessed seeds with soft coverings which would not pro- tect the embryo against injurious desiccation, the result being that unless the fallen seed found itself in moist conditions and germinated quickly, its chance of reproducing the plant was gone. Perhaps the most interesting of these plants were Crudya spicata and Moronobea coccinea, the one leguminous, the other guttiferous. The first is known as the Kakoon (Cacoon) tree, from the resemblance of its seeds to those of Entada scandens^ a plant bearing the same name. The second is the Hog-gum tree. It was, however, an observation on the seeds of Abrus STUDIES IN SEEDS AND FRUITS Working hypotheses. precatorius that gave direction and method to my inquiries. Noticing that the large, soft, unripe seeds of the green un- opened pod were three times the size and double the weight of the normally contracted hard seeds of the dehiscing pod, I found that a large, soft, unripe seed weighing 3 grains lost i J grains of water when drying and entering the resting state. This, I argued, would be the water that the resting seed would take up when swelling for germination ; and it thus appeared that in preparing for germination, a seed was merely resuming its original unripe condition. But it was to the phenomena of the shrinking process that my opportunities at first restricted my attention during the early part of 1 907 in Jamaica. Ob- servations on the seeds of desalpinia sepiaria, Canavalia ensi- formis^ C. gladiata^ and C. ohtus'ifolia^ led me to distinguish a critical period in the shrinking process which roughly coincided with the shrivelling of the cord or funicle, and the severing of the biological connection. When the soft, unripe, though full-sized seed was detached before the cord began to shrivel, it lost 70 or 75 per cent, of its weight in the drying and shrinking process ; but if the detachment was effected after the cord had commenced to wither, but before any drying of the seed was evident, then the subsequent loss of weight was only about 50 per cent. The result in the first case was a shrivelled seed ; in the second case a normal resting seed. Now 1 assumed that the difference between the two losses in thef drying stage, viz. 20 or 25 per cent., represented the water retained by the resting seed for the support of the embryo, and I termed it "the water of inclusion." In forming this inference I was also influenced by the results of simultaneous observations on the seeds of other plants, such as those of the Bastard Tamarind (JPithecolobium filicifolium) ; but I need not here particularise them further. My attention became then directed more especially to the observation of the ?ifrinkage of the soft, unripe, or uncontracted seeds of two leguminous climbers, Entada scandens and Mucuna urens, very favourable opportunities being afforded THE HISTORY OF THE INVESTIGATION 5 at Moneague in the interior of Jamaica. In March 1907 I there studied the maturation of the seed on the living plant and collected material which not only gave me employment at the time, but has occupied me at various intervals ever since. The results then obtained gave support to the theory of shrinkage above noted, and when I returned to England in May I adopted the water-of-inclusion theory, I then set to work to procure the germination of seeds of Entaddy Mucuna, Canavalia^ Abrus, etc., which had assumed the typical resting state after being gathered in the moist, soft condition, with their funicles beginning to shrivel ; and the results confirmed the view suggested by the earlier observa- tions that the resting seed in preparing for germination takes up the water lost in the previous shrinking stage. In other words, the swollen seed on the eve of germination resumed its condition of before the rest-period. The work was then continued on these lines, guided more by opportunities than by method. It was then argued that if the water absorbed for germination is the water previously lost in the drying process, it does not necessarily involve germination, meaning thereby the commencing growth of the embryo. It was held that if the above view was correct a swollen seed dried when on the eve of germination ought to return to its original resting weight and ought to retain its germinative powers. Both Therecipro- these inferences were established by the results of several germination, experiments discussed in the next chapter ; and I finally adopted " the reciprocal theory," as it was termed, which is to the effect that the water taken up for germination is the water lost in the previous shrinkage process. The theory of germination thus held good ; but it was very different when one came to confirm by experiment the water-of-inclusion view, which was really a theory concerned with the embryo's life in the resting seed. It is there suggested, as already pointed out, that some of the water The water of which the shrivelled seed has lost through being prematurely seed.^^ '" detached is retained in the normally contracted resting seed 50 per cent. 20 5) 15 55 15 55 6 STUDIES IN SEEDS AND FRUITS for the benefit of the embryo. From this point of view the water-contents of seeds might be thus characterised : — A. The water lost in the normal shrinking process ) and regained for germination . . . j B. The water which the seed shrinking excessively ) loses and the seed shrinking normally retains j C. The water which the normal resting seed loses | in the oven in addition to the water A and B j Waterless residue ..... Whilst A would be the water of germination, B would be the water of inclusion or the water of the rest-period, and C would be the water of combination, only to be driven off by exposure to a temperature of ioo° C. It follows from the above view that all resting seeds should possess Impervious coats which would secure the retention of the water of inclusion for the use of the embryo, and that the result of puncturing the resting seed or baring it of its coats would be a considerable loss of weight. It is also implied that the abnormally shrunken seed would lose less water in the oven than the normal resting seed. To make a long story short, I may remark that all the implications failed when put to the test. Many seeds proved to have pervious coats. They, as a rule, preserved the same weight when deprived of the protection of their coats by baring or puncturing. Lastly, the abnormally shrunken seeds lost about as much water in the oven as the normal resting seed. So my hypothesis, relating to a special supply of water for the use of the embryo in the resting seed, collapsed. But whilst putting it to the proof I had accumulated a large number of results of experiments which are utilised in other connections in the next chapter, and I chanced upon other suggestive lines of inquiry. After determining the water- percentage of seeds in the oven, I used to throw away the sample. But on one occasion a sample of the broken-up THE HISTORY OF THE INVESTIGATION 7 seeds of Guilandina bonducella was unintentionally left over- night in the pan of the balance. Out of curiosity I noticed its weight and found that it was 2 per cent, heavier than when originally placed in the oven. In other words, a sample of 100 grains, which had been reduced in weight by exposure to A clue to a a temperature of 100° C. to 92 grains, on the following morning JfJ^iJ-y ^ °^ weighed 102 grains. This result was startling, and quite a new road of investigation was opened up, occupying much of my time for three years and supplying materials for several of the chapters of this work. Under the stimulus of this discovery I made at the time a few speculative comments in my notebook, which proved to be the starting point of a theory of cosmic adaptation to which Chapter XX is devoted. Here again the seeds of Guilandina bonducella have been a source of inspiration, and I soon got to realise that I owed much to these interesting seeds. But in the meanwhile my horizon had been greatly Thewiden- extended. With the old theory gone it was evident that I hI,fi°oi^of could no longer treat the seed independently of the fruit, and the inquiry, that I could no longer ignore the facts that the seed had coverings, that the embryo in the resting seed was in all stages of development, and that the reserve of food within the seed presented great variations in amount as well as in disposition. The investigation promised to branch out in a multitude of ways, provokingly divergent in their direction. However, I continued the method of following indications and was soon hard at work again with the balance and the oven. Since the seed-coats had played a variety of parts in the experiments, one of the first of the new inquiries begun was concerned with the seed-coat relation, meaning thereby its The seed- relative weight as part of the entire seed. Great variety in coat relation, this respect soon displayed itself. At the same time I began to compare the seed-coat relation of the resting seed with that of the soft, unripe seed and of the seed swollen for germina- tion. Now commenced the separate treatment of the coats and kernel in the oven experiments, when I was surprised to 8 STUDIES IN SEEDS AND FRUITS find that the water-percentage for the coats was often greater than for the kernel. Before long it was realised that materials were gathering for a complete statement, as far as the indica- tions of the balance and the oven went, of the part played by On the way water in the economy of the seed from its immaturity through struction of i^s rest-period to the germinating stage. I was in fact on my ^Jj^ ''^s:»me of ^^y |.Q |.|^g construction of the seed's regime in passing from and swelling the unripe to the resting state and thence on to the germinating condition. Whilst engaged in this inquiry I took advantage of my oven-experiments for the determination of the water-contents of seeds to investigate further the curious fact that certain seeds after being subjected in a broken condition to a temperature of ioo° C, being thus deprived of their free water, not only regained all the lost water from the air, but in a few days were markedly heavier than in the entire state. This was found to be common with leguminous seeds possessing impervious coverings. The unexpected results of an experiment on the seeds of Entada scandens^ which belong to this type, threw fresh light on the matter. Separate samples of the coats and kernel were subjected to the oven- test, when the first experienced a loss of weight of about 15 per cent., and the second of about 10 per cent. They were then left exposed for five days on a table, together with samples of the coats and kernel which had not been heated. It was then found that in all cases, whether with the heated or unheated materials, the samples were considerably heavier than before the experiment, the weight of the coats in both cases being increased 2 or 3 per cent., and that of the kernels 3 to 5 per cent. There was, therefore, an inherent tendency in both the coats and the kernel when separated from each other to increase their weight by absorbing water from the air, a tendency unimpaired by a previous loss of all the free water through exposure to a temperature of 100° C. In other words, the seed in the broken condition held more water than in the entire state, and whether or not exposed to the heat-test. THE HISTORY OF THE INVESTIGATION 9 the ultimate result was the same. Such was the commence- ment of a long series of experiments which has extended off and on over three years. In some of the early experiments I tested the capacity for resorption possessed after heating by leaves, wood, slices of fruits, etc., and also by hydrated minerals, such as opal and chlorite. Another line of inquiry first taken up at this time was the determination of the relative weight of the embryo in other lines albuminous seeds. An investigation somewhat crudely begun ° i"q"'''y- soon branched off in many directions, the seeds of palms figuring largely in the results. Almost at the commencement of my work I had started experiments on the effect of time on the weight of seeds. They were carefully weighed and placed in paper packets, the intention being to extend the experiments over years. At this time I was feeling my way in many small inquiries, striking out blindly very often in my efforts to obtain further clues. Thus I took with me to Jamaica in the winter 1907-8 several seeds with the object of determining their changes of weight under different climatic conditions, the result being given in Chapter VII. These investigations began in October 1906, and during the first twelve months I mainly ignored the fruit ; but not altogether, since during the summer I had been periodically The author's observing some capsules of Scilla nutans to ascertain the no^"directed effect of cutting a window in the walls of the young fruit to the fruit, on the maturation of the seeds. The experiment was not deterrent and the changes not important. In a word, the seeds of the Bluebell had behaved like those of a Gymno- sperm. About the same time Lubimenko was carrying out a similar series of experiments on leguminous pods. Now began some observations on the dehiscence and drying of the capsules of Iris Pseudacorus and of Msculus Hippocastanum (Horse-chestnut). I experimented with the idea that dehiscence was the result of drying, a notion that guided many subsequent experiments, though a year passed before my error was discovered. lo STUDIES IN SEEDS AND FRUITS The winter 1907-8 spent in Jamaica was chiefly occupied in working as opportunity offered on the lines before indicated. I repeated several of the experiments made in England on the resorption of water from the air by seeds in the broken My work in Condition, so as to be assured that it was a capacity uncon- nected with climatic causes. At this time I began my observations on the Coco-nut {Cocos nucifera), ascertaining the proportional weight of all the parts — the husk, the shell, the albumen, and the embryo — and also determining their water-contents. The maturation and drying of the fruits of numerous other plants also occupied my attention, such as those of Anona^ Bauhinia, Citrus decumana (Shaddock), Datura^ Entada^ Ipom(sa^ Mahogany {Swietenid)^ Sapota, etc. I was surprised to find that the ripe capsule, the ripe legume, and the ripe berry often lost much the same amount of water when allowed to dry spontaneously, fleshy drupes like those of Prunus (tested in England) behaving in the same fashion. Then I reflected that fleshy fruits (drupes and berries) corresponded to the full-grown living legume and capsule in the moist, unopened condition, and that if we wish to find the correlative of these dehiscent fruits in the dry, opened state, we must look for it in the shrivelled currant and the dried-up apple. This raised the whole question of special adaptation in connection with seed dispersal. It was argued that if we can discern no evidence of adaptation in the shrivelled berry as regards seed distribution, we should look for none in the drying capsule and pod, and that the apparent display of method in the last-named is purely accidental. To enter more into the details of my work during my second winter in Jamaica would be to anticipate much that will be found in the succeeding chapters. I may, however, say that one branch of inquiry which was more fully developed was the hygroscopic behaviour of seeds, and that my study of the maturation of the fruits of Momordica THE HISTORY OF THE INVESTIGATION ii furnished me with some new ideas on the subject of the sequence in the genesis of the berry and capsule, two types " of fruits often associated. With the observations on Momordica my active work in Jamaica came to an end. However, whilst spending the last few days on the summit of Mount Diavolo, rambling in the forests and increasing my seed materials for future work in England, the theory of cosmic adaptation before alluded to was further elaborated. During the spring and summer of 1908 I was occupied in extending my observations on the rest-period of seeds. My work in on the swelling antecedent to germination, on the relative "£^" • weight of the embryo in albuminous seeds, and on various other subjects. In May I began a series of observations on the fruiting and seeding of the Ivy (Hedera Helix\ which have been extended to the spring of 191 1. The result has been to establish the growth of the embryo within the seed throughout the winter months, and the not infrequent germination on the plant (vivipary) in the spring. Here much assistance has been received from collections of fruits made at intervals during a winter by my sister, Mrs H. Mortimer. In the late summer and early autumn my systematic observations on the maturation, dehiscence, and drying of fruits were resumed. The fruits included those of Iris fcetidissima^ Quercus Robur (Oak), Arum maculatum^ Tamus communis^ and several other plants. In most cases the inquiry was continued during the next two or three years, the final result in the instance of the Oak being to establish a slight but normal tendency to vivipary or germina- tion on the tree. Perhaps the most important outcome of these observations a due to the was the clue to the homology of fruits supplied by my observa- fruits. °^^° tion of the shrinkage of seeds within the moist berry of Berberis. This afforded a clue for the comparison of fruits in their various stages, which was subsequently strengthened by data supplied by the seeds of Arum^ Tamus, and Passiflora, and ultimately enabled me to trace the homologies in the 12 STUDIES IN SEEDS AND FRUITS ripening and drying stages of different types of fruits, such as the legume, the capsule, and the berry, by fixing on a stage common to all. The result was to further undermine the prevailing notion of special adaptation to seed dispersal, and to show that the mechanism of a dehiscing capsule or legume, however adaptive it may appear, does not count for more in nature than the shrivelling of a berry. The winter 1908-9 was spent again in the West Indies, A sojourn in mainly in Grenada and Tobago, but including a sojourn Tobag-o,'and o^ two or three weeks in Trinidad. It was at Port of Spain Trinidad. j-j^^|. j made the acquaintance of Mr Hart, the late super- tendent of the Botanic Gardens, and I was indebted to the courtesy of Mr Evans, temporarily in charge, for the opportunity of obtaining an abundant supply of ripe palm fruits of different kinds. Mr Broadway of the Botanic Station in Tobago kindly gave me valuable information respecting Grenada, and subsequently assisted me by reply- ing by letter to numerous queries I had put to him. To Mr Anstead of the Botanic Station in Grenada I was very deeply indebted for, so to speak, giving me the run of the gardens and for other aid. The fruits and seeds of palms occupied much of my attention in Grenada ; but I made also several special studies, including one of the fruits of Barringtonia speciosa (an introduced plant). It was in Tobago that I made my first acquaintance with the " Twist Coco-nut," where the kernel lies loose within the hard shell. But one of the most important lines taken up during this winter was the study of the connection between monili- form legumes and the abortion of ovules, which opened up an interesting field of inquiry. The pods of Erythrina cor alio dendron supplied me with my first clue. This raised the question of the influence of the abortion of ovules and of the failure of young seeds on the form, size, and weight of the fruit ; and in this connection I made an extensive series of observations on the pods of Albizzia Lebbek, Entada ■ polystackya, and Leucana glauca. As often happened in other room in Grenada. THE HISTORY OF THE INVESTIGATION 13 cases, this led to another niquiry into the relation between the number of seeds and the size and weight of the fruit, which ultimately supplied me with materials for a special chapter. Whilst in Grenada I spent some weeks at the Grand Etang in the mountainous interior of the island. My attention here was occupied with many things in the surrounding forests ; but I was particularly interested in studying the habit of growth and the maturation of the fruit and seeds of Diocka reflexa, a leguminous climber, the seeds of which are amongst those stranded by the Gulf Stream on the western shores of Europe. The method of preparation of the seed for its Trans-Atlantic voyage and the opportunities it possessed of starting on its way were points of special interest for me. To give an idea of my mode of work in the West Indies My work I will describe my work-room at St George's, Grenada. . . . Hanging from nails on the walls to dry were the ripe fruits of Cassia fistula, Entada polystachya, and Hura crepitans (the Sandbox tree). On the window sill exposed to the sun were the opening fruits of Ravenala madagascariensis, displaying the beautiful blue arils of the seeds and completing the process of dehiscence which they had commenced on the tree. On the sill of another window were the large square fruits of Barringtonia speciosa in various stages of drying, all of which, together with the fruits above mentioned, were methodically weighed from time to time. On the table where I wrote were placed at one end my balance and at the other end my copper oven for the determination of the water-contents of seeds and fruits. Close beside me lay a saucer containing the seeds of Diocka reflexa from the green pod, which were silently illustrat- ing the coloration process of leguminous seeds. In other saucers around me lay a variety of seeds, all of them either under observation or destined for future experiment, such as the seeds of Barringtonia, Entada, Enterolobium, Monstera, and El^is. In a large press were two extensive series of pods of , 14 STUDIES IN SEEDS AND FRUITS Albizzia Lebbek and heucana glauca^ arranged according to the number of seeds, and from which I obtained my principal material for investigating the relation between the number of seeds and the size and weight of the fruit. In the same press were two small paper trays containing the bared seeds of C^salpinia Sappan, Erythrina corallodendron^ and EUis guineensis^ which, having been exposed to a temperature of 212° F. for two hours, were now regaining from the air the moisture lost in the oven. But perhaps the most interesting things around me were a number of seeds of Barringtonia speciosa which had been cut up with the hope of discovering the missing coty- ledons. Lying about the room were many West Indian fruits, such as those of Saccoglottis amazonica^ Carapa guianensis^ and Mauritia setigera, all from Trinidad ; whilst a box under the table contained my collection of seed-drift from the Trinidad and Tobago beaches, much of which had been brought down by the Orinoco from the interior of the neigh- bouring continent. In the spring of 1909 I brought back to England abundant material for further research in the various lines already instituted. During the summer the observations on the abortion of ovules and the failure of young seeds were con- tinued, and I began to pay systematic attention to the coloration of seeds, a subject about which many notes had been previously made. At the same time I was arranging my notes and working out the general results, during which many "lacunae" presented themselves; and it was with the filling up of these gaps, together with the working up of results, that most of the following twelve months were occupied. I may add that large seed-collections were at my disposal for this purpose. At the close of 19 10 I went to Turks Islands, which form geographically the southernmost portion of the Bahamas, and remained there three months. This locality was selected as the most suitable one in the West Indies for the study of oceanic seed-drift, or, in other words, for observing the dispersal THE HISTORY OF THE INVESTIGATION 15 of seeds by currents, a subject which I had been following up in various parts of the world since it first attracted my attention amongst the coral islets of the Solomon Islands in the early eighties, and one which I had constantly kept in view during my previous three winters in the West Indies. I had made collections in Jamaica, Tobago, and Trinidad, the large quantities of Orinoco drift washed up on the southern beaches of Trinidad proving full of interest. So I took up my abode in Grand Turk, and from there visited all the cays of the group, making a special study of the littoral plants and especially investigating the abundant stranded seed-drift, hardly any of which belongs to plants that have established themselves in those islands, or in fact in the Bahamian region generally. Whilst at Grand Turk I was fortunate enough to meet Dr Millspaugh, who was just completing the botanical survey The of the Bahamian flora which Dr Britton, himself, and other surve?ofthe botanists from the United States had been six years engaged Bahamas by _,,_,,,., 1 r 1 1 1 • • 1 botanists of m. The Turks Islands are the farthest south, and it might the United have appeared as if I had come in to spoil the finish of a great undertaking. However, our special interests lay far apart, and I could not for a moment lay claim to a fraction of the intimate knowledge possessed by Dr Millspaugh of the plants of this region. I was the astonished spectator of the applica- tion of his great experience to the flora of the little island of Grand Turk. All the knowledge of years acquired through the length and breadth of the Bahamas was brought to bear within the narrow limits of the flora of this small island. It was a lesson that I shall not readily forget. Dr Millspaugh very kindly lent me the manuscript of the Flora of the Bahamas by Dr Britton and himself ; and in return I gave him a collection of seed-drift obtained from the beaches of the group, which is now lodged in the Field Museum of Natural History, Chicago. When the American botanists publish their work they will enrich their science with the most complete account of a flora hitherto made in the West Indian region. States. 1 6 STUDIES IN SEEDS AND FRUITS This brings to an end my record of work and travel. During the preparation of this book many other " lacunae," as well as new lines of inquiry, have presented themselves. The first I have been compelled to largely ignore, whilst the new openings for investigation have been promptly blocked up. But little has been said of my study of the rest-period of seeds in this account of my work. As a matter of fact, however, it has formed the pivot of my inquiries from the beginning to the end. It started me on my work, and it is towards the clearing up of the problem concerned in its mystery that many lines of my inquiries now converge. The work- Looking back at all the data collected in this investigation, materials ""^ I am puzzled to account for their bulk. So often on returning home from my excursions it seemed that I had only made one or two observations worth recording during the day. Yet much of the work has been done by the seeds and fruits them- selves, whilst I have stood by to register results. Whilst I was engaged in weighing, or in some experiment, a score of seeds and fruits were silently at work around me. Amongst my seeds I have led a busy life, but no one has been so busy as the seed itself. The labour came when all the results had to be sifted, tabulated, and digested. The arrangement of the materials indeed occupied a good deal of serious thought and extended over a considerable time. It proved very difficult to avoid two dangers : the first that of going over old ground ; the second the assuming in an argument in an early part of the work what could only be demonstrated in a later page. Ac- cordingly I finally hit upon the method of first dealing with the shrinking and swelling of seeds until the question of the permeability or impermeability of their coats blocked the way and demanded a response. This enabled me to open up the whole matter of these qualities in seeds, and to establish a nexus in the general arrangement of the work. A serious And now for a serious word in concluding this chapter. word, q-j^g pj^j^ ^£ sending forth treatises without a trace of the personality of the worker is to me repellent. Knowledge in THE HISTORY OF THE INVESTIGATION 17 itself has but few attractions for me. That knowledge which brings one into touch with the life around one and enables the observer and the observed to tell their common story is the only thing that charms. I should feel no interest in inquiries that led one to the confines of habitable space and left one looking out on a dreary, cold, grey universe of nothingness. I would instead get quickly home to my cosy terrestrial sur- roundings and revel in thoughts that were comforting and consoling. Yet the fancy must always play a part if we wish to profit by and to make a real advance in any investigation. It would be easy to sustain the view that mere digging for facts is like digging into the ground. Under such conditions one does not see much beyond the length of one's nose. It is doubtful whether the progress of knowledge thus effected resolves itself into much more than the splitting up of phenomena, or into a process of differentiation that can only end in a relative zero. On the other hand, if at times we leave the solid ground of fact and rise into the air on the wings of fancy, we can at all events greatly extend our range of mental vision, and can mark down points for investigation which never would have come under our notice whilst adopting the mole's method of inquiry. It is the man in the air that gives the directions, and the man on the ground that does the work. By limiting our field of inquiry and excluding the play of fancy we are groping about as blindly as an army without its air-men. The dreamers figure in my mind as the leaders of the world. CHAPTER II THE THREE CONDITIONS OF THE SEED We are all familiar with the three conditions presented by the large, soft pre-resting seed, the contracted, hard resting seed, and the soft, swollen seed on the eve of germination. Yet to each of them we are apt to apply an epithet which is never altogether true and rarely altogether wrong. Thus we often speak of the pre-resting seed as immature, of the resting seed as mature, and of the swelling seed as germinating. In these connections it is necessary to remember that, as a rule, the embryo is fully developed in the soft, swollen pre-resting seed, and is quite ready, as shown in a later page, to proceed with germination, should the shrinking process be averted. If, then, the embryo is " mature " in the pre-resting seed, such an epithet can have no distinctive value for the resting seed. So again, when we speak of the germinating seed, we have to decide whether we mean the swollen seed on the eve of germination, or whether we refer to the seed with the tip of the radicle already protruding through the coats. Botanists, like Nobbe, PfefFer, Jost, and others, lay stress on the fact that absorption of water by the resting seed is not actual germination, but merely a preparation for that process ; whilst gardeners are familiar with the circumstance that seeds may swell up and not germinate. Strictly speaking, the germinating condition with the radicular tip showing Is a fourth stage. It is a stage of growth and activity within the THE THREE CONDITIONS OF THE SEED 19 seed, and cannot be compared with the three previous stages (characterised by a passive vitality), which may be fitly termed the shrinking, resting, and swelling stages. The germinating stage may never be reached, since the absorption of water that brings about the swelling of the resting seed is more a mechanical than a vital process, and may or may not terminate in the growth of the embryo, which is the essential feature of germination. The shrinking stage is characterised by loss of water, the The shrink- swelling stage by absorption of water, and the intervening swelling resting stage by a suspension, more or less complete, of the confemed drying process and by a greater or less state of passivity on the with water- part of the embryo. It is with these three stages that we are water-gain, now concerned, that of germination not coming within the field of our inquiry. The shrinking of the pre-resting seed and the swelling of the resting seed are the two processes to be now discussed. It is the shrinking process that ushers in the rest-period, and it is the swelling process that prepares the seed for germination. In the first case there is water-loss, in the second, water-gain ; and it may safely be assumed that as a general rule we are here concerned mainly with these pro- cesses, the proofs of which are discussed later on in this chapter. Here the balance becomes the instrument of investiga- Method of tion ; and it is to the changes in weight in the different '"'l""'y- processes that appeal is chiefly made. We deal at first with the entire seed, the independent behaviour of seed- coats and kernel being separately discussed in Chapter IX. The weight of the resting seed is taken as one, whilst the maximum weights of the pre-resting seed before shrinkage begins and of the swollen seed on the eve of germination are expressed in ratios. Thus, to take one of the very largest of leguminous seeds, that of Entada scandens : a seed which weighed 1004 grains in the soft, swollen condition weighed 408 grains in the contracted resting state ; and sub- sequently, when on the eve of germination, it increased ratios. 20 STUDIES IN SEEDS AND FRUITS its weight to 1012 grains by water-absorption. We thus get the result : — Large, soft, pre- r>„„f „ j Swollen seed on eve resting seed. ^^^^^"S '''^- of germination. 1004 grains. 408 grains, 10 12 grains. which expressed in ratios becomes : Shrinking ratio. Resting seed. Swelling ratio. 2-46 I 2-48 Modes of The ideal method of carrying out such observations would theYhrinSng "^turally be, as was done in the instance just given, to note and swelling the shrinkage and swelling of a single seed, that is, to take its maximum weight in the green fruit, to weigh it again after prolonged air-drying, and to weigh it once more when on the eve of germination. This involves the separation of the pre- resting seed from the parent plant ; and although one can choose the time when the cord or funicle is beginning to shrivel and the vital connections are being severed, still it is open to the objection that a seed thus detached does not dry under normal conditions. However, checks can often be found by comparing the state of the coats of a resting seed thus produced with that of a typical resting seed ; whilst any marked divergence from the average can be detected by a comparison of the swelling ratios of the two seeds. With seeds typically impermeable, a resting seed thus artificially obtained often lacks the impermeability of the outer coverings ; and its shrinkage, as indicated by the change in weight, is not so great as with the seed that has properly contracted on the plant. On the other hand, typical permeable seeds when dried under these conditions frequently shrink too much. Never- theless, normal results were at times obtained, some of which are mentioned below ; whilst the behaviour of the imperfectly shrunken seeds has offered a fruitful field of investigation. But there is another plan, and that is by comparing the average shrinkage of a number of detached pre-resting seeds with the average swelling of a number of resting seeds prepar- THE THREE CONDITIONS OF THE SEED 21 ing for germination. This method is more practicable and has been frequently employed, and, although open to the same objection as regards the premature separation of the seeds from the parent plant, it is easy, by the use of normal seeds as a check, to exclude those where the shrinkage has been irregular. A third method employed at times, especially for the shrinkage, has been to compare the average weight of a number of full-sized pre-resting seeds or of seeds swollen for germination with the average weight of a resting seed. This is exposed to the objection that seeds vary much in weight, an objection losing some of its force if a large number of seeds are used. With reference to the determination of the maximum weight of the swelling seed, it would appear difficult to ascertain where simple absorption ends and germination, or active growth within the seed, begins. In practice, however, this difficulty does not often arise, and it can usually be met by frequently weighing the swelling seed up to the beginning of germination, an approximate estimate being alone expected. By way of opening the subject I will first deal with my observations experiments on single seeds or sets of seeds, where the history °^ *^^- has been followed for each seed from its soft, swollen pre- and swelling resting condition through the rest-period on to the swelling single seeds stage terminating in germination. I possess such a continuous geeSlndi- series of observations for the seeds of four le2:uminous and one ^^^^ ^^^^ ^^^ . . 111.. water lost in malvaceous species. As given in the tables subjoined, these shrinking is data affiDrd an early indication of a principle which will figure swelling for prominently in subsequent pages, that the water lost in shrink- germination, ing is regained in swelling for germination. Thus, if a full- sized pre-resting seed weighing 100 grains shrinks to 40 grains during its drying, and after a rest-period of some months regains its original weight of 100 grains when swelling for germination, we have data directly indicating such a principle. There is much in these tables that will be elucidated as the work proceeds ; but I may here point out that the behaviour of the imperfectly shrunken seeds of Guilandina bonducella there 22 STUDIES IN SEEDS AND FRUITS noted lends a double support to this view, since it indicates that if less than the normal amount of water is lost in the shrinking process, less is also required in swelling for germina- tion. In Table A the histories of single seeds, and in Table B the history of a group of seeds, are followed in all three stages, the pre-resting, the resting, and the swollen stage preparatory for germination. A. — Showing the Weights of the same Seed in its Unripe or Pre-resting Condition, in its Resting Stage, and on the Eve of Germination. Name of plant. Weight in grains. Shrinking and swell- ing ratios. Duration of the resting stage. Entada scandens A „ B • „ c . Dioclea reflexa Poinciana regia Unripe. Resting. Swollen. 969 371 930 1004 408 I0I2 954 380 985 239 III 212 24-3 10-5 25-3 Shr. Rest. Sw. 2*62 I 2-51 2-46 I 2-48 2-51 I 2-59 2-15 I 1-91 2-31 I 2-41 3 months. 4 34 „ 4h „ B.— Showing the total Weights of a Number of Seeds of the same Species in the same three Stages. Guilandina bonducella (4 seeds). Thespesia populnea ( 1 1 seeds). 4i6'o ijS'o 449'o 607 34-9 627 2'37 I 2-55 (3-20 I 3-08) 174 I I -So 4J months. 4 >> Note. — All the seeds in these experiments germinated. Those of Guilandina bonducella did not complete the shrinking process, becoming permeable but "germinable" seeds. The ratios obtained independently for the normal impermeable seeds are in this case given in brackets. The same reciprocal relation be- tween the shrinking and swelling processes is established by independ- ent observa- tions on a number of seeds. The general indications afforded by observing individual seeds from the pre-resting stage to germination are confirmed and extended by the large amount of additional data given in the two tables subjoined. In Table A are arranged all my results on the shrinking and swelling ratios of seeds belong- ing to about ninety-six species, of which all but those in the supplementary list of palm seeds, etc., belong to the present discussion. They have for the most part been obtained THE THREE CONDITIONS OF THE SEED 23 Independently from different seeds or sets of seeds of the same species, since the opportunities of making a successful series of observations on individual seeds were rare. In Table B will be found a number of swelling ratios for other seeds as determined by the data supplied by Hoffmann and Nobbe. As supplementing my own results they are extremely valuable, since they supply some of the conspicuous deficiencies in Table A, and enable one to extend the field of inquiry in a tentative fashion over a large portion of the seed-bearing plant-world. It should of course be remembered that even with the best of observers and the best of conditions such results can only be approximations ; and it must not be forgotten that we are concerned here not merely with one process, but with all those changes concerned with the transition from the pre-resting stage to the eve of the germinating condition. The results as a whole are to be regarded here as supporting the general contention, based on the study of individual seeds, that the seed in swelling for germination is as a rule returning to the pre-resting or so-called unripe condition, that it gains in swelling what it lost in shrinking, and that the relation of the swelling to the shrinking seed is mainly reciprocal, the resting stage figuring as an interruption in the embryo's development in response to the pressure of external conditions. Before discussing these data, one may remark that there is little that is novel in the enunciation of this principle. Gardeners must often act on the tacit assumption of its reality, and scientific investigators have gone far to establish it. In our case it is particularly necessary to possess a clear conception of the mutual relation of the shrinking and swelling processes, since without the establishment of some preliminary general principle we should be unable to study with profit the mechanism of these processes as exhibited in the separate behaviour of the seed's coats and its kernel (see Chapter IX). This principle was so fully expected and accepted by Dr Dr Nobbe's Nobbe that he was satisfied with only a few experiments P''°° " 24 STUDIES IN SEEDS AND FRUITS for its demonstration. In his Handbuch der Samenkunde^ published in 1876, he shows that in the case of the seeds of the Common Bean {Faha vulgaris) the water lost in the shrink- ing process was gained back in the swelling stage. In the act of swelling, he says, the original volume of the " fresh " (or pre-resting) seed is restored (p. 71). Two of these fresh seeds were allowed to go through the shrinking process, the minimum size being reached in about ten days. Measurements were taken ; and it was ultimately found that one of the shrunken seeds when placed in water regained its original dimensions in about four days, whilst the other kept in a chamber saturated with water-vapour regained but little of its original size after five weeks. Table A. — Shrinking and Swelling Ratios of Seeds (Guppy). Note. — The ratios represent the relation in weight between the resting seed and the pre-resting and swelling seeds, the resting seed being taken as i. Thus if a seed weighed 25 grains before shrinking, lo grains when resting, and 24 grains when swollen for germination, its formula would be 2*5 i z'\, the shrinking ratio being indicated on the left, and the swelling ratio on the right. Average weight of a resting seed in Permeable, Family. Ratios. impermeable, or variable. grains. Shr. Sw. Abrus precatorius . Leguminosse i'5 2-15 2-25 Variable. Acacia Farnesiana . jj 2-0 2*13 2 "00 Adenanthera pavonina ,, 47 2-42 Impermeable. ^sculus Hippocastanum (Horse-chestnut) . Hippocastanese i30'o 2-17 Permeable. Albizzia Lebbek . LeguminosDG ^•3 2-27 Variable. Allium ursinum Liliacese O'l 2 'GO Permeable. Andira inermis Leguminosae 25-0 2*09 1 Anona muricata Anonacese 6-0 1-40 1 I '43 "^ ,, palustris ,, 4-0 I 45 I ,, Aquilegia (species) . Ranunculacese 003 1-62 1 Variable. Arenaria peploides . Caryophyllaceae 0-25 1-85 I ,, Artocarpus incisa (Bread- fruit) .... Arlocarpeje 45 '0 2*22 I Permeable. Arum maculatum Aroideae °"5 1-63 I Barringtonia speciosa Myrtacese 380-0 278 I ,, Bauhinia sp. . Leguminosre 4-0 210 1 2-20 Variable. Berberis sp. . Berberideze 0-I2 1-92 I Permeable. Bignonia sp. . Bignoniaceae 5"o 2-30 I " THE THREE CONDITIONS OF THE SEED 25 Table A. — continued. Average weight ofa resting seed in Permeable, Family. Ratios. impermeable, or variable. grains. Shr. Sw. Csesalpinia sepiaria . . LeguminosiTi 4-5 2-2S I 2*20 Variable. ,, Sappan . >> lO'O 2'10 I 2-24 Cajanus indicus )j 3-0 2-40 I 2-IO Permeable. Calliandra Saman . • 4-0 I 2-50 Variable. Canavalia ensiformis ,, 24-0 2-30 I 2-0O Permeable. ,, gladiata . >) 47 'o 260 I 2"00 Variable. , , obtusifolia )> I2-0 2-44 I 2-54 sp. . ,, iS-0 I 1-94 ^'' Canna indica . Cannacese 27 I '47 I 1-50 Cardiospermum grandi florum . Sapindacea; 3-0 170 I Permeable. Cassia fistula . Leguminosa; 4-0 276 I 2-52 Variable. ,, grand is ,j 9-0 I 211 , , marginata . jj 10 -Q I 2'10 Chrysophyilum Cainito Sapotacea; I2-0 I '45 I Permeable. Citrus decumana (Shad dock) . Aurantiaceas 4-5 1-65 I I -60 ,, Crinum sp. Liliacete 50*0 2'35 Datura Stramonium . Solanacea; 0-14 i'5o '' Dioclea reflexa . Leguminosse 90-0 1-90 [ I -80 Impermeable. • lOO'O 2 -20 ( 2-IO Permeable. Entada polystachya . >» 6-6 2-22 [ 2-29 >. )) 5'o [ 2-52 Impermeable. ,, scandens ,, 400*0 2-51 2-47 Enterolobium cyclocarpuE n M 17-0 2-30 Variable. Erythrina corallodendron jj 3'^ 2-i6 " ,, indica ,, 13-0 2-49 ,, velutina . ,, 7 '5 2-40 " Faba vulgaris (Broad Bean ) 33 'o 2*30 I '95 Permeable. Gossypium hirsutum Malvacea; i"o i-8o Guilandina bonduc . Leguminosaa 50-0 2-47 Impermeable. ,, bonducella ij 40*0 3 "20 1 3-08 '' ,, melanosperma 42-0 2-43 ,, (species of) ,, 60 -0 1 2-52 Hedera Helix . Araliaceae o"4 2 '00 I 2-12 Permeable. Hibiscus elatus Malvacese o"5 I 2'O0 ,, esculentus . ,, 0-8 I 170 " Sabdarifa . ,] 0-4 I 1-90 '' Hura crepitans Euphorbiacete 20 'o 2-29 I 2'IO j'' Ipomoea pes-capra; . Convolvulacese 3-0 3-20 I 2-45 Impermeable. „ tuba . ,, S"o 3-30 I 2-6o ,, tuberosa . ,, 25-0 2-40 1 2-45 Variable. Iris foetidissima Irideai 075 3*40 I Permeable. ,, Pseudacorus ,, 072 2-50 1 2-0O LeucKna glauca Leguminosse 0-8 2-84 I 2-62 1 Impermeable. Lonicera Periclymenum (Honeysuckle) Caprifoliacece 0-07 172 I Permeable. Luffa acutangula CucurbitaceEe I'O I 177 Momordica Charantia 3-0 1-50 1 >> 26 STUDIES IN SEEDS AND FRUITS Table A. — continued. Monstera pertusa Montrichardia arborescens Mucuna urens . Opuntia Tuna . Phaseolus multiflorus (Scarlet-runner) . Phaseolus vulgaris (P'rench Bean) , * Pisum sativum (Pea) Pithecolobium filicifolium Poinciana regia Primula veris (Primrose) Pyrus Malus (Apple) Quercus Robur (Oak) Ravenala madagascariensis Ribes grossularia (Goose' berry) . Ricinus communis (Castor Oil) . Scilla nutans . Stellaiia Holostea . Swietenia Mahogani Tamarind us indica . Tamus communis Theobroma Cacao (Cocoa Thespesia populnea. Ulex europseus Vicia sativa ,, sepium . Vigna luteola . Family. Aroidece LeguminosEe Cactea; Leguminosae PrimulaccK Rosacere CupuliferEe Musacese Ribesiaceaa EuphorbiacetE Liliacese Caryophyllacese MeliaceK Leguminosae Dioscoridese Buttneriaceae Malvaceae Leguminosae Average weight of a resting seed in grains. 35 o 90 "o 120 60 13-0 10 'o 0*012 o'40 25-0 o'og 3"5 O'l 0-45 37 20 "o 0-25 20 'o 3-0 O'll 0-31 o"35 0*70 Ratios. Shr. 2 "44 2*40 I -So 1-90 2-31 2-36 1-90 2-50 I -60 163 I '60 1-82 378 175 i"37 1-87 2-30 2-26 1-90 2-45 2-05 1-84 1-91 2"37 I 33 2"I0 2-IS 1-82 27 Permeable, impermeable, or variable. Permeable. Impermeable. Permeable. Variable. Permeable. Variable. Impermeable. Permeable. Variable. Impermeable. Supplementary List (see Note 20 of Appendix). Prunus communis (Sloe) Acrocomia lasiospatha Areca Catechu Cocos nucifera (Coco-nut) ,, plumosa Hyophorbe Verschafft Mauritia setigera Oreodoxa regia Sparganium ramosum Rosacere Palmaceae Pandanaceae I -00 1-89 I r 30*0 I "47 I 40*0 172 I 4600*0 1*89 I 6-0 I "59 I { 5-0 i-iS I 300 'O 175 I S"o 1-23 I o-o6 2*00 I V The question of perme- ability does not here arise. In all cases the seed proper is here re- ferred to. * See Note i of Appendix. The seeds were abnormally shrivelled. As with Faba vulgaris (Broad Bean) and Phaseolus multifloriis (Scarlet-runner), the seeds of Pisum sativum (Pea) shrink excessively if detached from the pod. It is therefore not possible to get good results in these cases. THE THREE CONDITIONS OF THE SEED 27 Table B. — Swelling Ratios of Seeds (Hoffmann and Nobbe). Swelling ratios of seeds adapted from the results obtained by Hoffmann and Nobbe, and given in the latter's Handbuch der Samenkunde, p. 119, etc. (The weight of the resting seed is here taken as i, as in the previous table.) Hoff- Nobbe. Page refer- mann. ences, etc. Triticum vulgare (Wheat) Gramineoe I "45 5 1-600 p. 119. Hordeum vulgare (Barley) 1-482 Secale cereale (Rye) I "577 ,, Avena sativa (Oats) . 1-598 ,, Zea Mays (Maize) . 1-440 1-398 ,, Panicum miliaceum (Millet) 1-250 ,, Fagopyrum esculentum (Buck- wheat) Polygonacece 1-469 ,, Ervum lens (Lentils) Leguminosse »"934 ,, Pisum sativum (Peas) ,, 2 -068 Ua) 1-9601 \{b) 1-710/ pp. 119, 122. Phaseolus sp. . . . ) (Weisse Bohnen, White Beans) j ,, 1-921 P. multiflorus? Phaseolus vulgaris . >, /(«) 2-1751 \{b) 2-007 pp. 119, 123, 124. Faba vulgaris (Broad Beans) ,, 2-040 2-570 pp. 109, 119. Vicia, Lathyrus (Vetches) ,, 1754 p. 119. Medicago sp. sp. (Lucern) ,, 1-560 V-878 p. 120. Trifolium repens (White Clover) ,, 2-267 1-890 ,, ,, pratense (Red Clover) ,, 2-175 2-053 ,, Papaver sp. sp. (Mohn, Poppy) PapaveraceK 1-910 P. somniferum, etc, pp. 80, 96, 120, Brassica Napus varieties (Raps) Cruciferce 1-510 1-483 418, 431- Raphanus sativus chinensis (Oel- rettig) ,, I -080 '•595 pp. 36, 120. Camelina sp. sp. (Leindotter, Cameline) .... ,, I -600 pp. 120, 359. Cannabis saliva (Hemp) . Cannabinea; I 439 ... pp. 90, 120. Helianthus annuus (Sunflower) . Compositae 1-565 pp. 120, 518. (Weisse Riibe) . 1-625 ■1-518 p. 120. (Zuckerriihe) . . . . 2-205 ,, Pinus austriaca Coniferse ■1-358 " Note. — According to Siewert, quoted on p. 120, the Lupines {Lupimis) have a swelling ratio of 2-00 to 2-30. With regard to Table B, it is to be observed that the original swelling results were stated as percentages of the weight of the resting seed. These have been converted into the ratios employed for my own results in Table A, where the shrinking and swelling processes are combined in one simple formula, the resting seed being taken as i. To have treated 28 STUDIES IN SEEDS AND FRUITS the swelling process as a thing apart would have been to ignore its all-important reciprocal relation to the shrinking process. This conversion is easy. Thus, whilst Nobbe states the swelling capacity of Wheat at 60 per cent., and Hoffmann puts that of Broad Bean (Faba vulgaris) at 104 per cent., I should state them as i"6o and 2*04 respectively, the resting seed being I'oo. Although there is a distinction to be drawn, as will be subsequently pointed out, between the water required for germination and the water necessary to saturate a seed, seeds under ordinary swelling experiments are apt to strike a rough average of their own by germinating, so that such experiments frequently prove to be germination experiments, in which one has to fix a somewhat arbitrary limit indicating where swelling ends and germination begins. This was in fact an almost invariable rule in my own experiments ; but by placing the seed in its earliest swelling stage in damp moss, excessive estimates were probably avoided. I do not gather that either Hoffmann or Nobbe attached much weight to the distinction between the swelling needed for germination and the swelling involved in saturation. Indeed, the latter expressly states (pp. 119, 120) that the kernels require as a rule to be thoroughly soaked before germination begins. In his care- fully guarded .recorded experiments the seeds were either immersed in water or kept moist by pouring water over them, methods that seem likely to produce excessive estimates. Yet, except in the case of Faha vulgaris^ his results as a rule come near to those obtained by Hoffmann for the same species ; and in spite of the difference in our methods my estimates for Pisum sativum (Pea), Phaseolus vulgaris (French Bean), and Phaseolus multiflorus (Scarlet-runner), are not far separated from those of Nobbe. This will be seen in the comparison made in Note i of the Appendix. As I have said before, the seeds assert themselves in ordinary experiments, and, disregarding divergent conditions, strike out a rough average result for all. For these reasons, therefore, we may, I think, claim that the THE THREE CONDITIONS OF THE SEED 29 results given in Tables A and B for the swelling ratios are fairly representative of the relative values of the swelling processes which precede germination in a state of nature. It will have been inferred that if the seed in swelling for Intherecip- germination takes up the water lost in shrinkirtg, the processes actSofthe must be essentially mechanical in their nature. That the shrinking 11 • £ J • -11 -1 ^"" swelhng- swelling ot a seed is essentially a mechanical process was processes is established by Dr Nobbe in a variety of experiments. Thus, IsIeJTtfaHy'''' he found that Clover seeds swelled with much the same UJ-Se"^^^' readiness in ordinary pure water as in water that had been previously either oxygenated, or carbonated, or chlorinated (pp. 102, 103) ; whilst he showed that the absorbing capacity is largely independent of the retention of the germinating powers, since seeds of Lady's-fingers {Anthyllis vulneraria), with a low germinative value (8 per cent.), swelled almost as freely as seeds of which nearly all (86 per cent.) were able to ger- minate (p. 114). The swelling of a seed, as he asserts on p. loi, is merely a mechanical process preparatory for germination. In the course of my own observations with the balance, there presented themselves a number of other proofs of the mechanical and reciprocal character of the shrinking and swelling processes. These are merely summarised or illustrated in the remarks immediately following, references being there made to where a detailed treatment of the points raised will be found. That the increase in weight of a seed in preparing for Additional germination is essentially due to absorption of water is ^[echlnfcS^ indicated: nature of the (i) By the fact that when a seed on the eve of germination p^cesT is dried in air at the ordinary temperature it returns approximately to its original weight as a resting seed ; (2) By the circumstance that the weight-relations between the kernel and its coats and between the embryo (in albuminous seeds) and the other parts of the seed 30 STUDIES IN SEEDS AND FRUITS been dried 1 the (3) are much the same in the seed that has after swelling for germination as they are resting seed ; In the fitness of the embryo in many pre-resting seeds to pass on at once to germination without the intervention of the resting stage. The results of a number of experiments on leguminous seeds indicate that when a seed on the eve of germination is dried under ordinary air conditions it returns approximately to its original weight. In illustration there are given below five examples selected from the table in Note 2a of the Appendix, where numerous other results, all pointing to the same conclusion, will be found, together with a discussion of the general nature of such experiments. All the results refer to seeds that subsequently germinated. Examples of the Effect of Drying under Ordinary Air Condi- tions ON Leguminous Seeds that are ready to Germinate (taken from Note 2 of the Appendix). Weight in grains. Gain ( + ) or or Swollen impermeable. Resting for After loss ( - ) seed. germi- nation. drying. Guilandina bonducella . Impermeable 33*35 97-30 35"35 + 6-0 per cent. Entada scandens . J, 312-50 728-00 318-70 -f- 2 -o , , Poinciana regia . Variable 870 19-50 8-70 0-0 ,, Phaseolus vulgaris (French Bean) . Permeable 13-50 24-60 13*35 -i-i ,, Faba vulgaris (Broad Bean) " 41 50 86-30 40*60 -2 "2 ,, Note. — A plant with both permeable and impermeable seeds is characterised as ' ' variable. " Although these samples support the view of the mechanical nature of the swelling process, they present curious divergencies in their behaviour when dried ; and the same may be said of the other results incorporated in the table in Note 2a. It appears, therefore, that as compared with their weight in the THE THREE CONE)ITIONS OF THE SEED 31 resting state some seeds on being dried in air after reaching the point of germination are heavier, others are lighter, and others remain unchanged. Various disturbing causes that would be likely to come into play in the course of the experi- ment here suggest themselves ; but they can usually be eliminated ; and, as shown in Note 2 a, none can account for the great contrast in the behaviour of the seeds of Guilandina bonducella and Faba vulgaris^ where in the one case there is a gain of about 6 per cent., and in the other a loss of 2 per cent. However, as will be seen in the table, as well as in the illustrations given above, these marked differences are displayed by two distinct types of seeds, the impermeable as represented by Guilandina bonducella^ and the permeable as exemplified by Faba vulgaris. A good deal of this contrast, therefore, lies behind the distinction between permeable and impermeable seeds ; and its significance will become evident only after a detailed consideration of those two types of seeds. The second additional proof that the swelling of a seed for Second, that germination is essentially concerned with absorption of water relations of might seem to be included in the first ; but there we were a°bumaf'^and concerned with the seed in its entirety, whilst here we are embryo are ... • , • -T-I1 r 1- • 1 • 1 1 much the deahng with its parts. Ihe proof lies in showing that the same in the absolute weight of parts that obtains in the resting seed is in Iftersw^ell- the main preserved in the seed that has been dried after swelling ingforger- r • • T- -11 • T J 1 1 • mmationas tor germination. Vo\xr illustrations supplied by leguminous they are in seeds are here given ; but for full details on this subject seed^.^^ ^^ reference must be made to Note 3 of the Appendix. I am here giving the results of observations on single seeds, and these are compared with the average weights of the parts in resting seeds of about the same size. The parts of the swollen seed were separated in the wet state and allowed to dry in an ordinary room. The small changes in weight that actually occur have a significance which is alluded to in a later page ; but they are not such as to materially affect the general conclusion to be drawn from the comparisons. The first two 32 STUDIES IN SEEDS AND FRUITS seeds are exalbumlnous, and here one is concerned only with the coats and the kernel. The last two are albuminous, and we have, therefore, in their cases to distinguish between the albumen and the embryo. Comparison of the Weights of the Coats, Kernel, and Embryo (in two cases) in Resting Leguminous Seeds, and in the same Seeds when dried after swelling for Germination. Resting seed Swollen Dried average. seed. swollen seed. Mucuna urens (exalbu- f coats minous) . . .[ kernel 237 grains 64*4 „ 34*2 grains 136-1 „ 23-1 grains 66-9 ,, 88-1 ,, 170-3 „ 90*0 ,, Faba vulgaris, Broad / coats Bean (exalbuminous) \ kernel 5-6 „ 34*4 „ 11*2 ,, 68-8 ,, 57 ,. 33'3 M 40*0 ,, 8o-o ,, 39"o ., ^t^, "'" (^'b-jalblen minous) . . .\ embryo 4*95 » 2'3S ,, 270 >. 9*43 ., 7-36 „ 6-21 ,, 4-6o „ 2-90 „ 2-50 M 10 "00 ,, 23-00 ,, 10-00 ,, Cassia marginata (albu- i ^°^^^ 2-84 „ 5*69 >. 1-27 ,. 5-90 M 1270 ,, 2-8o „ 2-70 „ 5-65 „ 1-23 „ 9-80 ,, 21-40 ,, 9-58 „ Third, that the embryo in many pre- resting seeds is able to pass on at once to ger- mination. Nature often supplies evidence of the readiness of seeds to "jump "the rest-period, the shrinking and swelling processes being then dispensed with. This is what we would expect if the shrinking of the soft pre-resting seed and the swelling of the hardened resting seed are essentially concerned with the loss and absorption of water. But in thus appealing to the potential vivipary of seeds we do so only in a very general sense, since numerous other influences may come into play. Though typically the resting stage represents a complete interruption in the embryo's development, this is not always so. A treatment of this complicated subject will be found in a later chapter ; and it is there shown that in the case of THE THREE CONDITIONS OF THE SEED 33 the seeds of plants like Arenaria peploides, Vicia sepium. Iris Pseudacorus, etc., which under normal circumstances enter into the typical resting state, it is possible, by keeping the soft, uncontracted pre-resting seed in warm, moist conditions, to induce germination, thus dispensing altogether with the shrinking and resting stages. Before making further reference to the ratios for the Neither the shrinking and swelling of seeds, it should be pointed out p^e?resting that as a rule neither the full-grown, uncontracted pre-resting sled^"°'^n^^ seeds nor the resting seeds swollen and ready for germination forgermina- c . -n 1 1 ji • 1 • tion are in are in a state or saturation. Both markedly increase their a state of weight when placed in water. The distinction between the ^**"™*'°"' amount of water required for germination and the larger amount needed for saturation is dealt with in a later page of this chapter. Here I will more particularly allude to the behaviour of the soft pre-resting seed in this respect. According to the principle that the swelling seed gains what the shrinking seed loses, we should infer that the behaviour of the large pre-resting seed and of the swollen seed on the eve of germination would be the same. This proves to be the case. When in Jamaica, I found that full-sized soft seeds from the moist green pods of Guilandina honducella gained about 20 per cent, in weight when placed in water ; whilst, in the failures of my germination experiments, when the seeds were kept in wet moss, the weight of the swollen seed was often correspondingly in excess of the normal weight for germination. So in England with seeds of Faba vulgaris (Broad Bean) and Phaseolus multiflorus (Scarlet-runner), I obtained similar results. Here, full-grown soft seeds from the green pod, that is to say, seeds that had not yet begun to shrink, added at least lo per cent, to their weight after lying in water for half a day. Then, again, resting seeds of Faha vidgaris^ that under ordinary con- ditions would have germinated when their weight had increased by 90 or 100 per cent., did not germinate at all when allowed to rernain in water, but kept adding to their weight by 3 34 STUDIES IN SEEDS AND FRUITS absorbing more water, until they had reached their saturation point of about 120 per cent. A closer con- We are now in a position to consider more closely the the shrinking shrinking and swelling ratios before tabulated. There are rattos^^"'"^ dealt with in these tables the results of observations on the seeds of more than 100 plants, four-fifths of which, as given in Table A, are from my own observations, whilst the rest, as included in Table B, are from the observations of Hoffmann and Nobbe. They belong to 38 families and comprise about 80 genera, of which rather over one-third are leguminous. In three or four cases only, viz. Pisum sativum^ Faba vulgaris, Phaseolus vulgaris, and perhaps P. multiflorus, are the same plants referred to in both tables. In the list containing my own results, seeds alone are dealt with, seed-like indehiscent fruits being excluded ; but in Table B we find also the " grains " of Cereals and a few seed-like indehiscent fruits, such as those of Buckwheat ; but 1 do not apprehend that the swelling ratios will be very materially affected. Dr Nobbe himself did not regard this disturbing cause as concerning the validity of his comparison (p. 112) ; and from my own observations on fruits to be subsequently discussed I would infer that, at all events with the grains of Cereals, which comprise most of the seed-like fruits in Table B, the effect of the coverings would be rather to lessen than to increase the contrast which evidently exists between the swelling capacities of the seeds of Cereals and the seeds of other plants. The views of Dr Nobbe sums up very briefly the results obtained by Hoffmann and himself concerning the swelling capacities of seeds (p. 120). Leguminous seeds, he infers, possess the highest capacity for absorbing water, whilst the lowest is possessed by oily and resinous seeds and by the grains of Cereals. This inference receives a general support from the results in Table A, and it will be sufficient at present to take the two extreme cases in illustration, the seeds of Ricinus communis absorbing one-third of their weight of water before THE THREE CONDITIONS OF THE SEED ^5 germination, whilst those of the leguminous Guilandina bonducella treble their weight. But this cannot take us very- far. A host of questions present themselves as we run our eyes down the lists ; and when we endeavour to answer them ofF-hand, a long vista of undetermined influences opens up. It would be easy to discover differences and to formulate distinctions ; but they would have little or no meaning now ; and it would be futile at present to base any general contrast, such as between families or between genera, on these data. (As already remarked, indehiscent fruits of the type found in palms and in genera like Prunus and Sparganium do not come within the limits of this discussion, but are referred to in Note 20 of the Appendix.) We have yet to appreciate the value and to estimate the The diffi- significance of such distinctions. Behind the varying behaviour subject°^*^^ of the shrinking and the swelling seed lie the seed's life- history and the cumulative effect of a multitude of inter- relations as between the seed and the embryo, between the kernel and its coverings, between the coverings and the fruit, and, through the fruit, between the seed and the parent and between the plant and its environment. It is, therefore, obvious that we can only with some security ignore the past when we have reason to believe that the seeds are akin in their history ; and that is why it will be wise at present to mainly confine our discussion of principles to the Leguminosae. But even here it will soon be evident that the risks increase as the affinities grow less, and the safest road will often lie in the study of the varying behaviour of seeds of the same species. Proceeding on these lines, we will inquire into the constancy The con- of these ratios in the same species, a necessary preliminary shr"nkin^ *^^ consideration, since' such estimates would lose much of their ^nd swelling , r . • r 1 1 . ratios in the value tor comparative purposes it the normal range is great, same species. But few of the " shrinking " results admit of bein^ stated in this fashion, as it was my wont in most cases to weigh 36 STUDIES IN SEEDS AND FRUITS a number of seeds together. Only the very large seeds were treated individually ; and I here give the shrinking ratios for three full-sized pre-resting seeds of Entada scandens (all of which subsequently germinated), the weight of the resting seed being taken as i. They were 2*46, 2*51, and 2*62. My method of determining the swelling ratios was better fitted for ascertaining their range in seeds of the same species, and the sample of results given in the tabic below is sufficient to bring out their relative constancy. The great contrast between the shrinking and swelling ratios in different species. Species. Adenanthera pavonina . Cassia grandis Entada scandens Faba vulgaris (Broad Bean) Guilandina bonducella . Mucuna urens Poinciana regia Canna indica . Family. Leguminosce Cannacece Number of seeds tested. Range of the swelling ratios. 2*30 2*04 2 '2 I I '95 2-8o 1-93 2-34 I '43 2 '60 2*22 2-59 2 -08 3'26 2'12 2 '44 1-52 Average ratio. 2-42 2-13 2-42 2 'ox 3-08 2-05 2-32 The next point to notice is the great contrast which the seeds of different plants display in their combined shrink- ing and swelling capacity. Relatively to their weight some seeds shrink and swell three or four times as much as others. We have seen that as a general rule the water lost in the shrinking is regained in the swelling, the one counter- balancing the other, so that it would be legitimate to estimate the missing ratio where only one value has been found. It will, however, be more convenient at first in dealing with the great range of the capacities to speak only of the swelling ratio, remembering of course its reciprocal character. In the tables the swelling capacity is stated as a ratio of the weight of the resting seed taken as i, the reason being that the swelling is only one-half of a reciprocal process which is centred in the resting seed. To avoid, therefore, the in- THE THREE CONDITIONS OF THE SEED 37 convenience of stating the values of the shrinking and swelling in different terms, a simple reciprocal ratio was invented, as described in a previous page. Thus, to take the seeds of Canna indica^ where the pre-resting seed loses 66 per cent, of its weight in entering the resting stage, and where the resting seed adds 50 per cent, to its weight in swelling for germination, the use of percentages for the two results would be clumsy and inconvenient. But stated in this manner : Pre-resting seed. Resting seed. Swelling seed. 1-5 i-o 1-5 we at once get a clear view of the problem. Now, however, when we are dealing particularly with the The employ- swelling capacity, the usual method of stating the increase as centages in a percentage of the weight of the resting seed will be adopted. ^th"he°'^^^ The conversion of the ratio into a percentage is simple enough, swelling the swelling ratio of 1-5 for the seeds of Canna indica being equivalent to an increase of weight of 50 per cent. We see accordingly that with many Leguminosae, where the seed swollen for germination is more than double the weight of the resting seed, the increase amounts to more than 100 per cent. Thus with Abrus precatorius^ where the swelling ratio, taking the resting seed as i, is 2*05, the actual increase in weight is 105 per cent. Looking at the extremes of the range of the swelling The range of capacity of seeds in general, we find amongst the hundred capacity/^ and odd plants in the tables two groups that we can handle ^^ ^I^aX fairly well, one where the absorbing capacity is not over extreme 60 per cent., the other where it is at least twice as much, ' reaching 120 per cent, and over. In the first or "minimum " group are to be included a number of plants in Table B, such as all the Cereals, species of the cruciferous genera Brasska and Camelina, Cannabis sativa, Pinus austriaca, and one or two others, together with several in Table A of the genera Anona, Canna, Citrus, Datura, Ricinus, Theobroma (Cacao), etc., belonging in both cases to a variety of families, but, 38 STUDIES IN SEEDS AND FRUITS if we except the species of Pithecolobium^ in no instance to the Leguminosae. On the other hand, of the seeds of plants to be placed in the " maximum " group, where the increase of weight is 1 20 per cent, or more, quite two-thirds are leguminous ; and of these we may cite Adenanthera pavonina^ Canavalia obtusifolia. Cassia fistula^ and species of Entada^ Enterolobium, Erythrina^ Guilandina, Leucana^ Poinciana, etc. Belonging to a variety of other families are the seeds of Barringtonia speciosa, Bignonia^ Ipomcea pes-capr^^ Prifnula, Swietenia (Mahogany), etc., which are for the most part placed here provisionally, since only occasionally, as with the species of Ipomcea, has the germinative capacity been tested for individual seeds of which the shrink- ing has been observed. Several of these probably belong to a type of seeds where the shrinking of detached seeds under experiment is considerably in excess of what takes place in nature. If we were able to interrogate nature more closely, we should probably find that the " maximum " group would be almost entirely leguminous, thus justifying the original view of Nobbe, which is referred to a few pages back. We can now turn from the range of extreme groups to that of different species. Here we can cite as extreme cases : (a) On the " minimum " side, the seeds of Ricinus communis, which require to absorb only 33 per cent, of their weight in order to germinate, and the grains of Millet {Panicmn miliaceum), which only increase their weight by 25 per cent, when swelling by water- absorption ; {b) On the " maximum " side, the different leguminous seeds that increase their weight from 150 to rather over 200 per cent, in taking up water for germina- tion, such as Canavalia obtusifolia, Entada scandens, Guilandina bonducella, and Leucana glauca. There is this much to be said concerning all cases where the amount of water absorbed for germination is very small, THE THREE CONDITIONS OF THE SEED 39 let us say below 20 per cent, of the weight of the resting seed, that such seeds are on the borderland of vivipary, where they dispense with shrinking and swelling altogether. This con- sideration links itself with some curious reflections that are dealt with subsequently. It is possible that we may have here an explanation of the very low swelling capacity ascribed by Hoffmann to a species of Raphanus [Oelrettig)^ nyl. 8 per cent., which is equivalent to a swelling ratio of roS. However, the ground will have to be cleared in many The causes directions before we can expect to discover the significance of varial:ion^in such contrasts in the swelling capacities as presented in the tables, capadty^^"^ A tentative use of the method of exclusion may perhaps assist us in getting into the right road ; and we will now endeavour to ascertain if there is any connection between these contrasts and certain conspicuous differences in the seeds. In the first place, there is the distinction in size. This can be at once dis- (a) Distinc- , . , , . .... IT . tions in size missed, smce the ordinary variation in the swelling capacity and weight. becomes almost a negligible quantity when we reflect on the great differences in size and weight between seeds. Thus the difference in weight between a seed of Primula veris (y^o^h of a grain) and a seed of Entada scandens (400 grains) is as between I and 40,000 ; yet the difference in their swelling ratios would only be as between 2-3 and 2-5. Then, again, it is possible that the distinction between albuminous and exalbuminous seeds may supply a clue to the variation in swelling. Exalbuminous seeds are most character- istic of the Leguminosae ; but plants of the same family with albuminous seeds are fairly represented in the tables. The (6) Albumin- albuminous leguminous seeds there included have the embryo albuminous more or less fully developed, the large, flat cotyledons being ^^^^^• nearly as long and as broad as the kernel. We are therefore employing seeds where the resting period has been imposed at much the same stage of development of the embryo. It will be seen from the comparison made below that such differences do not explain the great variation in the swelling capacity dis- played by seeds of leguminous plants. 40 STUDIES IN SEEDS AND FRUITS Comparison of the Swelling Capacities of Albuminous and Ex- albuminous Seeds of Species of Leguminos^, the Resting Seed being taken as i. Albuminous. Exalbuminous. Bauhinia sp. . . . 2*20 Cassia fistula . . .2*52 ,, grandis . . . 2'ii ,, marginata . . 2*10 Poinciana regia . . .2*37 Cassalpinia sepiaria . . 2*20 Erythrina indica . .2*49 Tamarindus indica . . 2'i5 Abrus precatorius . .2-05 Entada polystachya . . 2*29 Since the proportional weight of the seed-coats varies (c) Differ- greatly between different species, a subject discussed in proportional Chapter IX, it is possible that this may determine the varia- ^^lf^*^°t*^^ tion in the swelling capacity. A cursory reference to the results there tabulated will make it apparent that this is not the case. This is sufficiently indicated by the comparison made below, from which it is to be inferred that in leguminous seeds with similar swelling capacities, the coverings may con- stitute as little as 24 per cent., and as much as 61 per cent, of the entire weight. Comparison between the Swelling Capacities of the Seeds of Four Leguminous Plants and their Seed-coat Ratios. Swelling ratio. Seed-coat ratio, the weight of entire seed taken as loo. Acacia Farnesiana . Bauhinia sp. . C?esalpinia sepiaria . Mucuna urens . 2-0 2'I 2*2 2*0 6o-8 23;5 61-4 26-3 A noticeable feature in Table A is the small shrinking ((/) The seeds ratio of the seeds of moist or pulpy indehiscent fruits of the legume?*" berry type. I was rarely successful in getting them to germinate ; but in the successful cases of Jnona murkata (the Soursop) and of Citrus decumana (the Shaddock), it will be seen that the swelling ratio is similarly small, and no doubt this rule THE THREE CONDITIONS OF THE SEED 41 generally applies. That there is a real distinction between the seeds of berries and pods in this respect is shown in the comparison made below ; but it will not be possible here to do more than allude to it in passing. It will be futile to attempt to discuss it until the relations between fruits and seeds have been studied, a subject treated with detail in a subsequent chapter. Comparison of the Shrinking and Swelling Ratios (the Resting Seed being taken as i) of the Seeds of Berries or Berry- like Fruits with those of the Seeds of Legumes. Berries. Legumes. Shr. Sw. Shr. Sw. Anona muricata (Soursop) 1-40 I 1-43 Abrus precatorius . 2-15 2*05 ,, palustris I -45 I Bauhinia sp. . 2"IO 2-20 Arum macu latum . 1-63 I Csesalpinia sepiaria . 2-25 2-20 Citrus decumana (Shad- Entada scandens 2-51 2 "47 dock) .... I "65 I I "60 Faba vulgaris (Broad Lonicera Periclymenum Bean). 2-30 2-00 (Honeysuckle) 172 I Guilandina bonducella . 3-20 3-o8 Ribes grossularia (Goose- Phaseolus multiflorus berry) . 1-63 I (Scarlet-runner) . I 90 Tamus communis . 175 I Ulex europseus 2-30 Theobroma Cacao (Cocoa) 1-37 I Vicia sativa . 2-26 Chrysophyllum Cainito (Star Apple) I -45 I Only very moist baccate fruits are here named, the seeds of relatively dry berries losing rather more in the shrinking process, such as those of Hedera Helix (Ivy) and Berberis^ where the shrinking ratios are 2"00 and i'92 respectively. Bearing this in mind, it would appear from the foregoing comparison that whilst the seed of a moist berry when the fruit dries up loses on the average about a third of its weight, the unripe seed of the leguminous green pod during the shrinking pro- cess loses as a rule rather more, and sometimes much more than half its weight on entering the resting state. Yet considerations such as these do not carry us very far There is in dealing with the problem of the varying swelling ratios of JJgge di^tinc- seeds, since it is apparent from Table B that there are whole g*J^"f^-^^'^ °°* groups of plants with dry-looking fruits, such as the Cereals 42 STUDIES IN SEEDS AND FRUITS and the Cruciferae, where the seeds possess swelHng ratios quite as low as those of baccate fruits, and to these groups may be added often plants with capsular fruits. There would, there- fore, seem to be some common influence that brings not only the seed of the berry, but the seed of the cruciferous pod, of the capsule, and of dry-looking indehiscent fruits, into contrast with the seed of the legume. We will now appeal to the varying behaviour of seeds within the limits of a species, with the hope of discerning in their diff^erences some clue to the origin of the greater contrasts between the seeds of plants that stand apart from each other. We have before seen that the ordinary range of the swelling ratios in the same plant is small ; and inquiry has led me to believe that we shall not profit much by the study of small differences. At times, however, there are deviations that lie quite outside the usual range of the swelling ratios. The cases of excessive swelling, which are probably just as apt to occur in nature as they are in our experiments, are often readily explained by the unusual prolongation of the swelling period in normal seeds, which allows the seed to take up more water than is actually needed. Both permeable and impermeable seeds are liable at times to absorb much more freely than is usual. Under such conditions germination usually fails ; but occasionally it takes place, and we obtain swelling ratios far in excess of the average. Then, again, we have cases of excessive swelling, also ending at times in germination, where there has been abnormal shrinking in the drying of the moist pre-resting seed ; and since the exceptional loss of water has to be made up in the preparation for germination, accord- ing to the compensatory relation between the shrinking and swelling processes before established, we find this expressed in the increase of the swelling ratio. In these cases we are concerned only with permeable seeds, as will be explained in a later page. On the other hand, we have instances of swelling ratios con- siderably below the average and equally outside the ordinary THE THREE CONDITIONS OF THE SEED 43 range. This happens in the case of impermeable seeds where the shrinking process has been incomplete, and considerably- less water is required for germination than in the instance of the typical resting seed. I will first take those cases of excessive swelling displayed First, to the at times by normal seeds, and will begin with permeable seeds, swelling to which indeed earlier investigators seem to have mainly con- ^mesby^** fined their inquiries. Dr Nobbe inferred that as a rule seeds normalseeds. require to be well soaked for germination, and that those cases where complete soaking is not necessary are exceptional (pp. 118, 119). But he remarked that the minimum amount of water needed to start the germinating process was a subject (a) By per- r r ■ • / N T • • r . • .1 • J- ^ meable seeds, tor future niquiry (p. 120), It is, in tact, in this direction that later investigators have worked, and it is now possible to distinguish between (a) The minimum amount of water required for starting germination, (/>) The average amount that seeds absorb under natural conditions in swelling for germination, (c) The amount of water required for saturation. The quantity of water that a seed absorbs before germinat- ing under natural conditions, or in ordinary germination experiments, is considerably above the minimum amount needed, and markedly below the amount requisite for the seed's saturation. A rough indication that leguminous seeds germinate before they are saturated is found in the fact that normally seeds germinate long before they rupture their coats, which is the sign of the seeds being thoroughly soaked with water. (I refer to ruptures taking place away from the hilum.) Minimum results can, of course, be only obtained in the laboratory. By an ingenious course of experiments, referred to in Note 4 of the Appendix, Victor Jodin established a " germinative minimum " for Peas {Pisum sativum), and in his paper he quotes similar estimates made by Van Tieghem for " Feves " (I suppose. Broad Beans, Faba vulgaris). The results, which are stated in different fashions by the various investi- * 44 STUDIES IN SEEDS AND FRUITS gators, have all been converted into percentages of the weight of the resting seed carrying its normal water-contents. Comparison of the Minimum Amount of Water requisite for Germination, and of the Average Quantity absorbed under Natural Germinating Conditions, with the Maximum Amount that a Seed can take up, i.e. the Amount needed for Saturation. (Stated as percentages of the weight of the normal resting seed.) Pisum sativum. Faba vulgaris. Minimum for germination , Average for germination Maximum (for saturation) 67j;7iN 91 G ; 107 H ; 96 N no G 74 T. 95 G ; 104 H 120 G ; 118 T G = Guppy. H = Hoffmann. J::=Jodin. N = Nobbe. T = Van Tieghem. I have endeavoured here to give the general run of the results. The critical results are of course those made by the same investigator on the same set of seeds. Thus we notice particularly in the case of " Feves " (assumed above to be Faha vulgaris) Van Tieghem's estimate of 74 per cent, as the minimum amount of water required for germination and 1 1 8 per cent, as the maximum amount that a seed can absorb. So again my own results for Faha vulgaris (Broad Beans) were very definite, 95 per cent, being the average for germination and 120 per cent, the amount needed for saturation. The general indications above afforded for seeds of the types of the Pea {Pisum) and the Bean {Faba) clearly show that whilst the minimum amount of water required for germination is about 70 per cent, of the weight of the resting seed, a con- siderably larger quantity of water is needed for saturation, viz. no to 120 per cent. Between these two extremes lie the average amounts of water required for germination under ordinary conditions, as determined from the observations of Hoffmann, Nobbe, and myself. The seeds of Pisum sativum and Faba vulgaris are typical permeable seeds. Similar indications were presented in many of my experiments on impermeable seeds of the same family. THE THREE CONDITIONS OF THE SEED 45 Thus, a seed of Entada scandens weighing 400 grains in the resting state will be ready to germinate under normal condi- tions when it has increased its weight by water-absorption to from 950 to 1000 grains. But if the conditions are unfavour- able and it fails to germinate, it will continue absorbing water until it reaches a condition of saturation, when its weight will be about 1 150 or 1 160 grains. Thus : Water required stated as a percentage of the weight of the resting seed. Swelling ratios, taking the resting seed as i. For germination. For saturation. For germination. For saturation. Entada scandens . 140 to 150 % 190% 2-4 to 2-5 2-9 In other experiments seeds often continued to absorb water long after they had failed to germinate. They may occasion- ally germinate when the swelling ratio has far exceeded the normal limits, but only in an imperfect and belated fashion. One or two more instances will serve to illustrate these points. A seed of Albizzia Lebbek, 1 grains in weight, only requires to increase its weight to 4*5 grains in order to germinate ; but when it has failed to germinate it will continue absorbing water without rupturing its coat until nearly 6 grains in weight, the limits of the normal swelling ratio, 2*25, being long passed. So, again, a permeable seed of Entada polystachya^ weighing 7 grains, germinates usually after its weight has been increased by the absorption of water to about 16 grains, its swelling ratio being 2-3 ; but when it fails it continues to swell and reaches a weight of 21 grains and over, thus tripling its original weight to no purpose. Convolvulaceous seeds, possessing in the resting stage a hard, dry albumen, are especially liable to take up more water than is needed for germination. In its transi- tion into the mucilaginous state, the albumen absorbs a very large amount of water. The normal swelling ratio for a seed of Ipomcea pes-capra is about 2-5, but seeds may germinate tardily 46 STUDIES IN SEEDS AND FRUITS after their weight has been increased threefold, and seeds that fail to germinate sometimes acquire 3^ times their original weight. When we come to deal with the hygroscopicity of seeds we shall be able to state precisely the lower value of the seed's hydratation, that is to say, below the minimum required for germination. This is the " hygroscopic maximum " (see Chapter VII). Coming to the second kind of excessive swelling, we deal here with seeds abnormally shrunken, when the excess of water absorbed in the swelling process compensates for the excess lost in the shrinking stage. But it should be noted that we are not here concerned with seeds so much shrunken that they have lost their vitality, a fate that may befall permeable and impermeable seeds alike, but with seeds that still retain their germinating powers. Thus it comes about that permeable seeds alone illustrate this type of excessive absorption, since defective shrinking acts in different ways on the swelling capacity of seeds, according to their permeable or impermeable character in the resting state. If permeable, the shrinking is too great, and the swelling is also excessive. If impermeable, the shrinking is deficient and the swelling ratio is much reduced. This is in accord with the compensatory principle before established, that what the shrinking seed loses the swelling seed gains. But it is important to notice that we are here again brought face to face with the distinction between permeable and impermeable seeds. Permeable seeds that are allowed to dry when detached from the plant in the full-sized moist condition generally shrink too much ; and as a rule they fail to germinate, absorb- ing much more water in proportion to their weight than in the case of the normal resting seed. Greatly shrivelled seeds are, as is manifest, imperfectly developed, so that they scarcely call for our attention. However, to illustrate this subject, I give below my observations on the seeds of Canavalia ensiformis in three conditions : excessively shrivelled, moderately shrivelled, and normal. THE THREE CONDITIONS OF THE SEED 47 Comparison of the Shrinking and Swelling Ratios of the Seeds OF Canavalia ensiformis under Different Conditions. Condition of seed. Weight in grains. Ratios, taking the resting seed as i. Gerniinative capacity. Excessively shrivelled . Moderately ,, Normally contracted . 44-1 15-0 39-0 6o-o 25-5 597 48-3 23-3 46-6 Shr. Sw. 2*94 I 2'6o 2-36 I 2-34 2-07 I 2-00 Lost. Usually retained. Normal, Those cases where seeds of the same species at times display much diminished swelling capacities, as expressed in swelling ratios considerably below the average and outside the ordinary range of variation, now claim our attention. Such cases are peculiar to impermeable seeds, and occur when the shrinking process is incomplete, less water being required for germination, since less than the normal amount is lost in the shrinkage. This is a point of considerable importance, and one on which much of my work has been concentrated, since it supplies the key to one of the principal positions maintained in these pages. We get here our first glimpse at the signifi- cance of impermeability, and we perceive how it comes about that both permeable and impermeable seeds may be found in the same species. Very interesting indications in this direction are sup- plied by the seeds of Guilandina bonducella, and I will briefly state the general results of numerous observations made in Jamaica. When we open the full-grown green pods of this plant we find two kinds of soft green pre-resting seeds : (ii) Yellowish green soft seeds, about 100 grains in weight and 25 to 27 millimetres in diameter, which represent the maximum development in size and weight of the pre-resting seed before the shrinking process commences. Third, to the great diminu- tion of the swelling capacity caused hy the deficient shrinkage of impermeable seeds as illustrated, (a) by Gui- landina bonducella, 48 STUDIES IN SEEDS AND FRUITS (b) Dark olive-green soft seeds, rather firmer than those above, 94 or 95 grahis in weight, 23 to 25 mm. in diameter, and representing the pre-resting seed in the earliest stage of shrinking. If we open other pods that are commencing to wither we find the seeds in various stages of contraction ; and finally we come upon the dried-up brown dehiscing pod containing hard, grey impermeable seeds very much smaller than those of the a and b stages, and weighing 33 or 34 grains. Such is Nature's method, provided we do not interfere with it. But if we remove from the green unopened pod seeds in the a and b stages and allow them to dry on the table, we get the following results. The fully formed pre-resting seed of the a stage shrinks excessively, and when that process is complete we find a greatly shrivelled seed, weighing 27 or 28 grains, which takes up water readily, yet fails to germinate. But if we take one of the olive-green b seeds, where shrinking has already commenced, and allow it to dry on the table under the same conditions, we notice that instead of shrinking exces- sively its shrinkage as compared with that of the normal resting seed is deficient, and that instead of weighing about t^t^ grains, as it would have done if it had been left undisturbed in the pod, its final weight is as much as 42 grains. Such a seed, though permeable and hygroscopic, retains its germinative capacity. Now comes the critical part of the experiment. If we put the greatly shrivelled a seed in the conditions for germinating, we find that whilst it swells excessively by absorbing water and increases its weight by about 230 per cent., it fails to germinate. On the other hand, the b seed, which has shrunk deficiently, absorbs water easily, increases its weight by about 150 per cent., and germinates healthily. Very different is the behaviour of the normal resting seed. Its impervious outer coat has to be filed through before it can take up water for germination ; and before that stage is reached it increases its weight by about 200 per cent. But the important point as indicated in the results tabulated THE THREE CONDITIONS OF THE SEED 49 below is that the two germinable seeds, the permeable, deficiently shrunken seed and the normal, impermeable seed' when swelling for germination, take up much the same amount of water that they surrendered in the shrinking process. In other words, the needs for germination are satisfied by the seed's regaining the water previously lost, the permeable seed taking least and the impermeable seed most. This is well brought out in the ratios given in the table ; and thus we see how, without impairing the germinative powers, deficient shrinkage leads to decrease in the swelling capacity and to the loss of impermeability. Table showing the Different Effects of Excessive, Deficient, AND Normal Shrinkage on Seeds that are typically Imper- meable IN the Resting State, as illustrated by the Results of Experiments on the Seeds of Guilandina bonducella. Condition of the resting seed. History of the seed. Weight in grains of the pre-resting, resting, and swollen seed. Shrinking and swelling ratios, the resting seed as I. Effect on the seed. A. Excessively shrunken B. Deficiently shrunken C. Normally contracted Detached from the green pod before shrinking had begun Detached from the green pod in the early stage of shrinking Shrinking pro- cess carried out in the pod on the plant Pre- resting. 100 100 100 Rest- ing. 28-5 42-0 33*3 Swollen for germi- nation. 92 105 100 Shr. w. 3-50 I 3-30 2-38 I 2*50 3 "oo I 3 -00 Permeable and hygroscopic, germinative powers lost. Permeable and hygroscopic, germinative powers re- tained. Impermeable and normal in all re- spects. The shrmking ratio for the normal resting seed was mainly obtained by comparing the average_ weights of seeds in different stages of contraction on the same plant. The weight of the resting seed averages nearly 40 grains ; but for convenience in statmg the ratios the weight of 33 grains is employed in the table and in the text. The important lesson of the seeds of Guilandina bonducella in this matter is that decrease in the swelling capacity is 5C STUDIES IN SEEDS AND FRUITS associated with the loss of impermeability, and that these results are primarily due to the deficient shrinkage of the pre-resting seed. In my experiments other impermeable seeds behaved like those of Guilandina honducella. When they had lost their impervious character they swelled less and in consequence required less water for germination. The loss of imper- meability was in fact associated with marked diminution in the swelling ratios. A good example of this result was afforded in the case of the seeds of Dioclea reflexa, gathered by me in the woods of the Grand Etang in Grenada, their mode of occur- rence being described in Chapter V. Here the origin of the loss of impermeability could be readily traced, and the permeable seeds were easily recognised by their larger size, darker hue, softer coverings, and by other indications of deficient shrinkage. The outcome of a number of observations was as follows : — The permeable seeds of Dioclea reflexa in swelling for germi- nation increase their weight by about 80 per cent ; whilst the impermeable seeds just double their weight. This difference is not so great as in other impermeable seeds, since the seed- coverings act somewhat irregularly, as described in Chapter IX. If the share taken by the coats in the swelling process is eliminated, the contrast between the swelling capacities of these two types of seeds is made more evident, the kernel of the permeable seed showing an increase of weight of about 80 per cent., and that of the impermeable seed of about 130 per cent. The swelling mechanism of these seeds is discussed in Chapter IX. The behaviour of the seeds of Guilandina honducella and Dioclea reflexa must be typical of many other leguminous seeds with impervious coats. Any influence that impedes the shrinking of the soft seed of the green pod tends to prevent the acquirement of impermeability. One may cite in this connection the seeds of Casalpinia sepiaria and Casalpinia Sappany which are often impermeable when allowed to ripen on the plant, but permeable if they have been prematurely detached from the pod, as described in Chapter V. THE THREE CONDITIONS OF THE SEED 51 Sometimes these deviations from the normal behaviour of a seed become fixed ; and Nature tlien facilitates our inquiries by presenting in the same species two types of seeds which are distinguished not only in size and colour, but also by their different degrees of impermeability. Such seeds have also two corresponding degrees of swelling capacity, the permeable seeds requiring much less water than the impermeable seeds. Entada polystachya^ as observed by me in Grenada, is a case of this kind. Here we find two types of seeds differing from each other in almost all the critical points that distinguish permeable and impermeable seeds, and equally capable of reproducing the plant. As indicated in the table below, they differ in colour, size, and weight, as well as in their swelling capacity, the permeable seed increasing its weight by about 124 per cent, before germination, whilst the impermeable seed requires more water and adds 1 50 per cent, to its weight. (c) Entada polystachya, which dis- plays the same prin- ciple in its two types of seeds. Comparison of the two Types of Seeds produced by Entada polystachya Type. Colour. Length and breadth. Resting . weight. Shrinking and swelling. Weight in grains. Ratios. A. Large and permeable B. Small and impermeable Yellowish brown Dark brown 15 X 12 mm. 13x11 mm. 6-6 grs. 5 -o grs. Unripe. ^-;- Swollen. 14-8 6-6 14-8 12-5 5-0 12-5 Shr. Sw. 2-24 I 2-24 2*50 I 2*50 Different hypotheses present themselves in explanation of this relation between the swelling capacity and the permeability of a seed. For example, it may be suggested that it is merely a matter concerned with tj;ie water-contents or hydratation of a seed, a view that would accept, without explaining, the impli- cation of these experiments, that impermeable seeds contain less water than permeable seeds. To form an opinion now would be to prejudge a matter which will prove to be far more way. The question of permea- bility and im- permeability blocks the 52 STUDIES IN SEEDS AND FRUITS complex than it at first appears. This question of the ability or inability of seeds to absorb water through their coats has been constantly arising in this chapter in connection with other seed -capacities ; and in the particular subject we have just been considering it is manifestly impossible to make further progress until we investigate the nature of the differences associated with the distinction between a permeable and an impermeable seed. At present, there- fore, the question of permeabihty and impermeability blocks the way. We will accordingly proceed to its discussion in the next three chapters. SUMMARY (i) We deal here with the three conditions presented by the large, soft pre-resting seed, the hardened, contracted resting seed, and the soft, swollen seed about to germinate (p. i8). (2) This involves the study of the shrinking and swelling of the seed, processes which are in the main concerned with w^ater-loss and w^ater- gain (p. 19). (3) The balance is the instrument of this investigation, and the modes of thus determining the shrinking and swelling ratios are described (4) The indications of the single seed, when its history has been follovv^ed in all three conditions, prove that the water lost in shrinking is regained in swelling for germination, and that the swollen seed represents the return to the pre-resting or so-called unripe state (p. 21). (5) The same reciprocal relation between the shrinkmg and swelling processes is established by independent observations of the three conditions on a large number of different seeds (p. 22). (6) The principle that the water lost in the shrinking process is gained back in the swelling stage was accepted by Dr Nobbe in his Handbuch der Samenkunde^ 1876 (p. 23). (7) Tables of the shrinking and swelling ratios are given, the first containing the results of the author's observations, the second those of Hoffmann and Nobbe (pp. 24-27). (8) It is then pointed out that the essentially mechanical nature of the shrinking and swelling processes is involved in their reciprocal character (p. 29). THE THREE CONDITIONS OF THE SEED S3 (9) Additional proofs of the mechanical nature of the swelling process are indicated as follows : — (rt) By the fact that when a seed on the eve of germination is dried, it returns approximately to its original weight as a resting seed. The results of a number of experiments on leguminous seeds are given, and it is shown that seeds fall short of or exceed the original weight according as they were in the first place permeable or impermeable (p. 30) ; (/>) By the fact that the weight-relations of coats, kernel, albumen, and embryo are much the same in the seed dried after swell- ing for germination as they are in the resting seed ; illus- trated by examples (p. 31) ; (t) In the ability of the embryo in many pre-resting seeds to pass at once to germination without the intervention of the resting stage, such embryos being potentially viviparous (P-. 32)- ( I o) Neither the large, soft pre-resting seeds nor the seeds swollen for germination arc in a condition of saturation (p. 33). (11) The shrinking and swelling ratios are then more closely con- sidered (p. 34). (12) The views of Dr Nobbe (p. 34). (13) The difficulties of the subject, and the necessity of confining the discussion of principles to one family, namely, the Leguminosae (P-35)- (14) The constancy of the shrinking and swelling ratios of normal resting seeds of the same species is then shown (p. 35). (15) But the contrast, as exemplified by the swelling ratios, is great between different species, whether in groups or in individuals. Thus, the Cereals with a swelling capacity of not over 60 per cent., and the Leguminosae with a capacity ranging between 90 and 200 per cent., represent the minimum and maximum groups. That leguminous seeds possess the highest capacity for absorbing water when preparing for germination was stated long since by Nobbe. The contrast between individual species is displayed by the seeds of Ricinus communis and of Guilandina bonducella^ the first adding only one-third to their weight, and the last trebling their weight when swelling for germin- ation. Seeds that possess a very low swelling capacity, for instance, below 20 per cent., are probably on the borderland of vivipary (pp. 36-38). (16) The reasons of the great range in the swelling capacities of different seeds are then considered, that is to say, why the seeds of some plants absorb much water and others very little in swelling for germina- tion. After endeavouring to ascertain whether there is any connection between this great variation, and certain conspicuous differences in 54 STUDIES IN SEEDS AND FRUITS seeds, such as distinctions in size and weight (p. 39), the albuminous or exalbuminous character of seeds (p. 39), dift'erertces in the propor- tional weight of the seed-coats (p. 40), the difference in the types of fruits, as between berries and legumes (p. 41), etc., it is inferred that there is much which these distinctions will not explain (p. 41). (17) Appeal is then made to the unusual deviations in the swelling capacity of seeds of the same species, with the hope of finding a clue to the origin of the great contrasts in this respect presented by the seeds of different plants (p. 42). (18) In the first place, cases of excessive swelling in normal resting seeds are discussed, both of the permeable and impermeable types ; and it is pointed out that it is possible to distinguish between [a) The minimum amount of water required for germination as shown in laboratory experiments ; [b) The average amount of water that under natural conditions seeds absorb when sweUing for germination ; [c) The maximum amount of water that seeds can absorb to produce saturation, the seed continuing to take up water long after it has failed to germinate (pp. 43-45). (19) In the next place, reference is made to the case of excessive swell- ing in abnormally shrunken resting seeds where the absorbing process is compensatory, the unusual loss of water in the shrinking process being thus supplied. But such seeds, whether in the normal condition permeable or impermeable, have lost their germinative powers, and as such have no concern for us here. Seeds that shrivel much, absorb much, but do not as a rule propagate the species. This is usually the fate of seeds that are removed from the green pod in the soft, full- grown pre-resting state before shrinking has begun. But with imper- meable seeds it sometimes happens that such seeds shrink less when allowed to go through the shrinking process detached from the plant than when left undisturbed in the pod. If we wait until shrinking has just begun before detaching the seed from the plant, this deficiency in the shrinkage is the rule. Such seeds retain their germinative powers, but their shrinking has been deficient ; and, according to the compensatory principle that what the shrinking seed loses the swelling seed gains, their capacity of absorbing water for germination is propor- tionately reduced. They are larger and heavier than the normal impermeable seeds, and take up water easily (p. 46). (20) When, therefore, we come to appeal to cases of unusual diminu- tion of the swelling capacity in resting seeds, we find that examples are only supplied by abnormal impermeable seeds. Here the shrinking process is incomplete, and in consequence less water is required for germination than in the normal resting seed. This unexpected contrast between the behaviour of permeable and impermeable seeds when THE THREE CONDITIONS OF THE SEED SS they are detached in the soft pre-resting state from the green pod, and allowed to go through the shrinking process removed from the parent, brings us face to face with the significance of the impermeability of seeds. A great deal lies behind the fact that under such conditions with a permeable seed the shrinking is too great and the subsequent swelling for germination excessive, whilst with an impermeable seed the shrinking may be deficient and the swelling ratio much reduced (21) Illustrations of the behaviour of impermeable seeds in this respect are afi:orded by those of Gu'ilandina bonducella and Dioclea reflexa^ where deficient shrinkage of the large, moist pre-resting seeds leads to a corresponding decrease in the normal swelling capacity and to the loss of impermeability ; and by those of Entada polystachya^ which displays in its two types of normal seeds, permeable and impermeable, a small shrinking capacity and a small swelling ratio for the first, and a large shrinking capacity and a large swelling ratio for the second (pp. 47-51). (22) Experiments on seeds of the impermeable type in their soft pre- resting condition therefore indicate that a diminution of the shrinkage prevents the acquirement of impermeability and at the same time lessens the amount of water required for germination. This strange relation between the amount of water a seed absorbs in preparing for germination and the permeability or impermeability of the seed- coverings renders necessary an investigation of the distinctive characters associated with these two types of seeds. The question of permeability and impermeability, therefore, blocks the way (p. 52). CHAPTER III THE IMPERMEABILITY OF SEEDS AND ITS SIGNIFICANCE The impermeability of seeds has occupied the attention of many able investigators, but usually in connection with some other character. They have not, however, always been in agreement as to the meaning to be attached to the term " impermeable," a fundamental difference which is reflected in their occasionally inconsistent conclusions. In this work a seed is regarded as impermeable only when it is able to resist the penetration of water during an immersion of weeks or months. Nature supplies abundant evidence of the impermeable character of certain seeds in the floating drift of ponds and rivers and of the ocean currents ; and many inquirers in their observations and experiments on seed -buoyancy have dealt indirectly with this subject ; but as it would be out of place to allude to their results here, I would refer the reader for a detailed treatment of the matter to my book on Plant Dispersal. It should, however, be remarked that imperme- ability may be equally a quality of the seed that sinks and of the seed that floats, its connection with buoyancy being only of an indirect character. I will at once proceed to deal with some of the aspects of the subject, on which recent investigations have thrown light. The fre- The frequency of impermeability in the case of seeds of certain permeability." families, especially among the Leguminosae, was established by Nobbe a generation ago in the pages of his work on 56 THE IMPERMEABILITY OF SEEDS 57 seeds, Handbuch der Samenkunde ; and he it was who pointed out that even with plants where all the seeds seemed to be permeable, there might be a small residuum of 2 or 3 per cent, that continued to resist the penetration of water even after some months' immersion (pp. 112, 113). But this botanist made no special study of the subject, except in so far as it was in relation to other seed-characters. It is to a recent Italian investigator, Dr Guiseppe Gola, Itsinvesti- that we are indebted for an extensive series of studies of the or Gola. nature and prevalence of this quality. His memoir, which is entitled Ricerche sulla bio login e sulla fisiologia de Semi a tegumento impermeabile^ and which was published by the Royal Academy of Sciences of Turin in 1905, will come as a surprise to many who, like myself, were unaware that im- permeable seeds are so frequently to be met with in the plant-world. Although not able to follow him in all his conclusions, I have found in his data the materials for a solid foundation on which to begin in these pages a discussion of the subject of impermeability. After ascertaining by a preliminary inquiry that this character was exhibited most markedly by the seeds of Leguminosae and then by those of Cistaceae and Malvaceae, Dr Gola decided to limit his researches to those three families. His method was well adapted to test the capacity of seeds in this respect, since they were kept immersed in water for periods of between thirty and eighty days. When swelling occurred, it took place generally in from three to five days, and less usually in about ten days. Impermeable seeds remained unaffected at the end of the trials. The seeds of about 300 species of plants were exposed to this test, of which about 260 (belonging to 45 genera) were leguminous, whilst nearly 30 belonged to the Cistaceae, and 10 were malvaceous. Temperate genera greatly predominate in his list of results, not merely numerically, but also in the number of species tested. Including the Acacias, we may here mention the genera Astragalus^ Cytisus, Genista^ Lathyrus^ Lotus, Medicago, Melilotus, 58 STUDIES IN SEEDS AND FRUITS Trifolium, Vtcia^ etc. As a rule between 50 and 100 seeds of each species were experimented on, and the percentage of permeable seeds is recorded in a special table. On tabulating the results given by Dr Gola for the Leguminosas and Cistaceae in the manner shown below, I was surprised to learn how common it was to find permeable and impermeable seeds in the same species. Tabulation of thk Results for the Leguminos^ and Cistaceae GIVEN BV Dr Gola in his Table showing the Percentage of Permeable Seeds in the same Species. Number of species. Species with no seeds permeable. Species with all seeds permeable. Species with both permeable and im- permeable seeds. Number. Per- centage. Number. Per- centage. Number, 1 Per- centage. LeguminosK Cistaceae , 260 27 '7 7 40 203 27 78 100 Permeable We see here that 78 per cent, of the species of Leguminosas meabk^seeds possessed both permeable and impermeable seeds, those with commonly ^jj seeds impermeable or with all seeds permeable amounting same species, to 7 and 1 5 per cent, respectively. The proportions of each kind of seed in a species were in most cases fairly constant, the results for two different samples being as a rule not far apart. Within the limits of a genus the species exhibited great contrasts in their proportions of permeable and im- permeable seeds. Almost all the leguminous genera that are best represented in Dr Gola's table show variations ranging between the two extremes : [a) of species with all seeds permeable, and {h) of species with all or nearly all seeds impermeable, none containing exclusively species possessing one type of seed. Not one of the leguminous genera which are here represented by ten or more species could be designated as exclusively characterised by permeable or impermeable seeds. THE IMPERMEABILITY OF SEEDS 59 All are variable or " mixed " in this respect. It would thus appear from the results of the investigations of Dr Gola that impermeability with the Leguminosae is not usually a distinction between species, and hardly ever between genera. But although the leguminous genera best represented in the table of Dr Gola are " mixed " in the sense that not one of them can be described as exclusively possessing species with one type of seed, they may differ in degree considerably from each other. The average proportions of permeable and impermeable seeds for the species of the three genera. Acacia^ Astragalus^ and Lathyrus^ are given below ; and from the tabulated results it appears that in order of impermeability Acacia stands first, with an average of 83 per cent, for a species, and Lathyrus last, with an average of 20 per cent. Impermea- bility not a generic char- acter, though more charac- teristic of some genera than of others. Average Number of Permeable and Impermeable Seeds in the Species of three Leguminous Genera. (Tabulated from data given in Dr Gola's Memoir.) Number of species tested. Average percentage for a species. Permeable seeds. Impermeable seeds. Acacia Astragalus Lathyrus . 20 43 20 17 II I' 65 20 In the same way, with regard to species of Leguminosae, it would seem that although as a rule, that is in 80 per cent. of the plants, we cannot employ the presence or absence of impermeability in the seeds as a specific distinction, yet species differ very much in the proportion of impermeable seeds that they possess. Such, then, are the preliminary conceptions with reference to impermeability that one forms from the data obtained by Dr Gola. These conceptions will be considerably extended 6o STUDIES IN SEEDS AND FRUITS as one follows the trend of still more recent investigations. However, I pass over for a time the results of the brilliant researches of Paul Becquerel, carried out in the laboratory of the Sorbonne between 1904 and 1907, because in as far as they deal with matters discussed in these pages they are primarily concerned with the impermeability of seeds to air, and will be more appropriately referred to in the chapter on the hygroscopicity of seeds. In following the history of the investigation of this subject from the data at my disposal, the next advance on solid ground seems to be that supplied by the results obtained by Mr Crocker. If Dr Gola opened our eyes as to the frequency of impermeability in the seed-world, Mr. Crocker has done much to enlighten us as to its biological significance. In different papers published in the Botanical Gazette (Uni- versity of Chicago Press, 1906-9) he expresses the opinion that " delayed germination," or, as we might term it, " seed- longevity," is more generally due to seed -coat characters "limiting or entirely excluding water or oxygen - supply " than to embryo characters or the "so-called dormancy of protoplasm." His results go far to establish the view put forward by Nobbe and Hanlein in 1877, and quoted by him, that " delayed germination is due in many cases to the impermeability of the seed-coat to water." One of his papers (" Role of Seed-coats in Delayed Germination," October 1906) is concerned especially with this point ; but he repeats the general conclusion formed, as above given, in a later paper ("Germination of Seeds of Water-plants," November 1907), and in a short article in the same publication in January 1909. It may be remarked before concluding this reference that Mr Crocker regards impermeability to water as much more con- ducive to seed-longevity than imperviousness to oxygen {thid.^ October 1906) ; and this is the point that is of most interest to us in the present stage of this discussion. " The coats that exclude water " (he writes) " are undoubtedly much better adapted to securing a long delay." THE IMPERMEABILITY OF SEEDS 6i It is to Professor Ewart of the University of Melbourne The investi- that we are indebted for a most important contribution to our Prof^Ewart. knowledge of seed-longevity and of the impermeability of the seed-coverings. If the author had had the leisure to extend the explanatory portion of his memoir, it would have ranked as one of the most extensive records of investigations con- cerned with seeds since De Candolle published his Geographic botanique rather over half a century ago, Dr Gola and Mr Crocker, as far as the dates of publication are concerned, were his predecessors in the field ; but it would be probably more correct to say that they were contemporaries in their labours. Professor Ewart, like Dr Gola, discusses the distribution of impermeability in seeds, and, like Mr Crocker, he considers the relation between longevity and impermeability. But on this point the two investigators in America and Australia seem to diverge widely, Mr Crocker holding that '' delayed ger- mination is generally related to seed-coat characters rather than to the so-called dormancy of protoplasm," whilst Professor Ewart considers that longevity depends not on the seed-coats, but on the staying power of the protoplasm. The difference, however, is more apparent than real, since Professor Ewart evidently regards the impermeable covering as an adaptation for ensuring the long life of the seed in the soil ; and at all events it would seem likely there is enough common ground on which to base a theory that would reconcile both the views. In this memoir on " The Longevity of Seeds," which was published in the Proceedings of the Royal Society of Victoria for 1908, he deals with the seeds of about 2500 species of plants. In addition to the results of his own work, which involved the employment of nearly 3000 tests. Professor Ewart incorporates in his table, which in itself occupies 176 pages, all the previous records relating to the subject that he could find. Yet, in spite of the great importance of this contribution to knowledge, one is conscious of missing much in the 22 pages that are alone devoted to the summarising of results, 62 STUDIES IN SEEDS AND FRUITS since a very large amount of interesting, if subsidiary, matter finds no expression either in the table or in the text. Professor Ewart arrives at the conclusion that " macrobiotic " seeds (as he terms seeds that may last from 15 to over 100 years) are characterised by " more or less impermeable coats, and are restricted to a few natural orders, of which the Leguminosae greatly surpass all others, whilst Malvaceae and Myrtaceae come next in importance." The general trend of the curves, he remarks, indicates that the extreme duration of vitality, probable for any known seed, lies between 150 and 250 years. The number of species in the list of plants characterised by macrobiotic seeds amounts to 180, which is less than i per cent, of the total number of species experimented upon ; and of these about 75 per cent, are leguminous. Sixty per cent, or 30 out of the 50 species of Acacia included in the general table possess these long-lived seeds ; and since impermeability and longevity are so closely associated in this genus, we may recall here the high average of impermeability assigned above to seeds of species of Acacia from data supplied in Dr Gola's tabulated results. There is naturally no attempt here to deal with the general question of impermeability, except in its relation to longevity ; and the method of stating the results in the table rarely allows me to draw such inferences for myself with security. Still, some general principles of much importance are indicated in the text and in the table ; and they may be here alluded to. In the first place, as was previously implied by Mr Crocker, the impermeability of seeds to water exhibits itself as an adaptation for ensuring the long life of the seed in the soil. It is not essential for securing the longevity of seeds in air. Professor Ewart points out that some seeds with readily per- meable coats, such as those of species of Phaseolus and Triticum^ may retain their vitality for many years in dry air. This being the case, it follows that the experiments, based as they chiefly are on seeds that have been kept in dry air for a varying number of years, are mainly concerned with the THE IMPERMEABILITY OF SEEDS 63 longevity of seeds under those conditions. The question whether impermeable seeds will preserve their vitality longer in the soil seems to be as yet unanswered. Although, as The question quoted in this paper, Duvel has shown that in the case of permeable " ordinary seeds the advantage lies with seeds dried in air, he preser^^^^' did not determine this point for the hard seeds that did not ^^eir vitality , . . 1 -r> r T- longer in the germinate in his experiments, apparently, as rrotessor Ewart soil than in observes, because he did not employ methods that would fully test their germinative capacity. Commenting on the results of the extensive experiments of Duvel and Waldron to deter- mine the length of time weed-seeds must be buried in order to lose their vitality, Crocker points out that " vitality tests of this kind, that neglect the effect of the seed-coats, are tests of the condition of the seed-coats rather than tests of the real vitality of the embryos themselves " {Botanical Gazette^ October 1906). In this connection the testimony of Professor Pammel of Iowa is significant. As concerning the seeds of a number of weeds, he found that the percentage of seeds germinating was lower and the dormant period longer if the seeds were kept during the winter in paper packages than if they were placed in sand and exposed to the climatic conditions of an ordinary winter {Brit. Assoc. ^ 19095 "Nature," October 28, 1909). It is nevertheless evident from the Australian observations that even the most resistant of seeds tend to lose their im- permeability in the course of years when kept in dry air. With the loss of this quality the seed is not necessarily deprived of its power of retaining its vitality for a longer period, since typically permeable seeds, as has been already remarked, may preserve their germinative capacities for many years. It, how- ever, is deprived of the power of prolonging its existence in the soil ; and in consequence it either germinates or dies. How long a seed with impermeable coverings could remain alive when buried deeply in the soil, it would be extremely difficult to determine with accuracy. But it is apparent from Professor Ewart's own observations that under the conditions of undis- turbed primeval forest in Australia they might lie buried in the 64 STUDIES IN SEEDS AND FRUITS soil for very long periods without losing their impermeability. Under such conditions he found that all Acacia seeds found below the surface possessed impermeable coats and required special treatment to produce swelling and germination. Whether or not the seeds always retain their vitality when they preserve their impermeability is another matter, since longevity, as Professor Ewart observes, may not depend primarily on the impermeability of the seed-coats, but on a peculiar inherent property of the protoplasm, the duration of which under the soil is secured by the impermeable coverings. Seed-longevity would seem therefore to be determined by two independent eventualities, the limit of the impermeability of the coats and the limit of the staying power of the protoplasm of the kernel or embryo ; and the question arises as to which lasts the longest. Amongst the results of Professor Ewart's experiments it is easy to find cases where impermeability has sur- vived its utility ; but it would be hazardous to assert that this is the usual course of events under the soil. This method of stat- ing the problem seems to be the best way of reconciling the views of Mr Crocker in America and Professor Ewart in Australia. An important outcome of these two series of investigations is that the issues can be narrowed, thus permitting one to dis- tinguish between the extrinsic and the intrinsic in the results of experiments. Results applicable to the behaviour of the seed in air are in a sense extrinsic, since such are not the usual conditions under which Nature tests its longevity. Those that can be brought into sorpe kind of relation with the seed as it occurs naturally in the soil are likely to be the most instructive. Two questions, it would seem, have shaped themselves whilst considering these results. The first is : Under which conditions would an imper- meable seed retain its vitality longest, in the air or in the soil } The second is : Which has the greatest staying power, the impermeability of the seed-coats, or the germinative capacity of the kernel } Notwithstanding the evidence before us, the answer to both of them is indeterminate. THE IMPERMEABILITY OF SEEDS Gs There are, however, one or two points to which further reference might be made. Thus, to take the first query, it may be replied that even in the case of the most impermeable seeds the effect of being kept in the dry air of a room for many years would be undoubtedly to favour the development of small cracks in the outer covering, thus converting them by degrees into permeable seeds. My observations on the seeds of Entada scandens (Chapter X) will go to show how this could be brought about. We can thus perceive how much less likely it is that impermeable seeds exposed without protection to the weather could withstand year after year the alternating conditions of heat and cold, of sun and shade, of drought and humidity, which in one form or another they would experience whatever their situation. In an elaborate series of experiments, in which he reproduced the extreme changes between moist and dry conditions and between heat and cold in various shapes, such as an ordinary climate would present, Dr Gola found that the seeds lost their impermeability. To come to a matter of my own observation, it is doubtful whether any of the numerous impermeable seeds washed up on tropical beaches could withstand for many years exposure to the sun and rain. All of them would show sooner or later signs of wear and tear in the injuries to the outer coats. On the other hand, buried in the soil, the seed would be more or less safeguarded against the risks to which a seed lying on the ground would be exposed. But the degree of protection would vary with the depth below the surface, so that the seed deepest down, as shown by Duvel, Crocker, and Ewart, would have the longer life. At the same time new dangers might arise ; but on the whole it seems likely that under such favouring conditions as characterise the typical Australian forests, the buried seed might retain its imper- meability for much longer periods than when kept dry in cupboards or in botanical museums. But this raises again the second question whether the germinative capacity would be similarly retained. Of this it 5 66 STUDIES IN SEEDS AND FRUITS may be said that we shall probably be never quite secure in our interpretation of Nature's experiments in this direction. But this insecurity does not invalidate the testimony altogether ; and it scarcely seems prudent to ignore altogether the accumu- lation of evidence respecting the high antiquity of "germinable " seeds found in ancient graves or when an old soil is disturbed. With regard to the little value of negative evidence in such an inquiry, I point out in the next chapter that we can never be certain that the failure is due to the incapacity of the germinative powers and not to the method employed. Both Crocker and Ewart are emphatic on the point that it would require more evidence than was deemed necessary by the earlier investigators to convince us that the cause of the failure to respond by germination to the call of their experi- ments lay always with the seeds. I venture to think, and here I am supported by a considerable amount of evidence given in the succeeding chapters, that the truest test of the potential vitality of an impermeable seed is to be found in the constancy of its weight under all ordinary conditions and under the influence of time. If a seed with sound coats gained nothing in weight after weeks of immersion in water, made no response to the varying hygrometric changes of the atmosphere, and preserved the same weight for a number of years, I would presume its germinative soundness, whatever its previous history or whatever its attested antiquity. Not the least important part of Professor Ewart's memoir is the account given by Miss White in an appendix of the results of her investigation of the structure of coats of imper- meable seeds. After making a microscopical examination of the coats of nearly seventy species of impermeable seeds which had been the subject of Professor Ewart's inquiry, she formed the following conclusion : — " As a general rule in small and medium-sized seeds the cuticle is well developed, and repre- sents the impermeable part of the seed-coat ; whilst in the case of large seeds, such as those of Adansonia Gregorii, Mucuna gigantea^ Wistaria Maideniana^ and Guilandina honducella^ the THE IMPERMEABILITY OF SEEDS 67 cuticle is extremely unimportant and inconspicuous. In these seeds the extreme resistance which they exhibit appears to be located in the palisade cells.". . . The circumstance that the seat of the chief resistance to the penetration of water may lie in large seeds in the deeper tissues may explain how Dr Gola comes to consider that this is the rule for impermeable seeds. Miss White's investigations, however, establish the fact that the seat of impermeability lies for most seeds in the structure- less cuticle, a conclusion previously indicated, but on less extended grounds, by the inquiries of Nobbe and by those of Bergtheil and Day. SUMMARY (i) The frequency of impermeability in seeds of certain families, especially of the Leguminosa;, which was first indicated by Nobbe a generation ago, has within the last few years been established by Gola, an Italian investigator. His results bring out the facts that, though more typical of some genera than of others, impermeability is not a generic character, and that it is rarely even a specific character, since both permeable and impermeable seeds are commonly found in the same species (p. 57). (2) Another of the inferences of Nobbe that delayed germination is due in many cases to the impermeability of the seed-coverings to water has been confirmed and extended by the recent researches of Crocker in America. After this point, questions affecting seed-impermeability usually resolve themselves into matters concerning seed-longevity. Crocker rejects the idea that a seed's long life is as a rule to be attributed to embryo characters or to the dormancy of protoplasm (p. 60). (3) However, Ewart in Australia, after very extensive researches, arrived at the conclusion that the longevity of seeds depends not on their coverings, but on the persistence of the protoplasmic constitution of the embryo or kernel ; and he views impermeability of the coats as an adaptation for ensuring the long life of the seed under soil-conditions (p. 61). (4) Of the 2500 species (more or less) with which Ewart deals, either directly or through the observations of others, rather less than i per cent, are long-lived seeds that will retain their germinative capacity from fifteen to one hundred years and over. Of these " macrobiotic " 68 STUDIES IN SEEDS AND FRUITS seeds all possess more or less impermeable coats, and three-fourths are leguminous (p. 62). (5) It is suggested in this chapter that seed-longevity should be regarded as determined by two factors, represented in the imper- meability of the coats and in the persistence of the protoplasmic constitution of the embryo-kernel (p. 63). (6) It is also suggested that, accepting impermeability as an adaptation to soil-conditions, we should leave to the future investigator these two points to determine : [a) whether the impermeable seed would retain its germinative capacity longer in the soil than in the air ; [b) as to the relative durability of the impermeability of the seed-coats and the germinative capacity (p. 64). (7) Recent investigators lay stress on the fact that negative results obtained by earlier investigators in testing the persistence of the germinative powers of hard seeds were more probably due to their inacquaintance with the right methods of procuring germination than to failure on the part of the seeds (p. 66). (8) Whilst considering that a combination of the theories of Crocker and Ewart would present the best working hypothesis, the author is inclined to the view that the most practical tests of the potential vitality of an impermeable seed are to be found in the constancy of its weight under all ordinary conditions and in the lapse of years. He would presume the germinative capacity of such a seed, whatever its antiquity, provided, as has just been implied, that its coats are sound, that it absorbs no water, and that it makes no response by alterations in its weight to the varying hygrometric states of the air. The author also does not regard it as prudent to ignore altogether the accumulation of evidence respecting the great age of "germinable" seeds found in ancient graves or when an old soil is disturbed (p. 66). (9) Lastly, the opinion of Nobbe and of later investigators that the seat of impermeability lies in the outer coverings of the seed has been confirmed by the results of the recent researches of Miss J. White, who places it in the case of small seeds in the cuticle and with large seeds often in the outer palisade cells (p. 66). CHAPTER IV PERMEABLE AND IMPERMEABLE SEEDS The observations and experiments on the results of which the four following chapters are based cover the period of 1906 to 191 1. They were practically completed, and the greater part of the results elaborated and written out, before the works of other investigators had been consulted. In this condition they have been in the main reproduced in these pages, as I thought it best that they should tell their own story, my original purpose having been to make an in- dependent study of the impermeability of seeds without being influenced by the ideas of others. That has been done ; but in the final summing up of my own results 1 have been guided in the estimate of their value and in the drawing of my conclusions by the results obtained and the opinions formed by other inquirers. In order to introduce the subject and to give method to Comparison the arrangement of the results of a large number of observa- ofGuiiandina tions and experiments, I will take the very divergent behaviour, and^cana- as revealed by the balance, of the seeds of two leguminous vaiiaensi- plants, Guilandina bonducella and Canavalia ensiformis. The first named has a very hard grey seed of the size and form of a marble, weighing usually about 40 grains, and possessing very thick coats. The second has a thin-skinned white seed, about 20 millimetres long, weighing 20 to 25 grains, and typical of a large number of leguminous plants. (See Note 5 of the Appendix.) 69 70 STUDIES IN SEEDS AND FRUITS On placing these two seeds in water we obtain very- different results. That of Guilandina honducella absorbs no water and preserves its original weight after an immersion of many months or even years. On the other hand, the seed of Canavalia ensiformis begins to swell in a few hours, and within twenty-four hours has doubled its weight. One seed, therefore, is impermeable or waterproof, whilst the other is permeable. But this difference in behaviour is associated with a difference in other qualities. If we weigh the seeds daily for a week or two, employing a quartz pebble as a standard of comparison, we observe that the seed of Guilandina honducella behaves exactly like the pebble and keeps its weight to within a small fraction of a grain. The seed of Canavalia ensiformis, on the con- trary, varies considerably in response to the daily changes in the atmospheric humidity, the amplitude of its variations amounting to 2 or 3 per cent, of its average weight. One seed, therefore, behaves hygroscopically, and the other does not. As might have been expected, it is the impermeable seed of Guilandina honducella that is non-hygroscopic, whilst with the seed of Canavalia ensiformis permeability and hygroscopicity go together. When extending the weighing observations over twelve months, we find the same features of difference displayed. Whilst the Guilandina seed maintains its weight unchanged, the Canavalia seed continues to exhibit the same hygroscopic variations. If we were to represent these results in a diagram, we should denote the behaviour of the seed of Canavalia ensiformis and that of the Guilandina seed by a horizontal line. The line of the last named would be even, but that of the Canavalia seed would display numerous zig-zag irregularities, marking the hygro- scopic responses of the seed. It is essential to understand that we are here dealing with seeds that have completed their spontaneous drying in air. Where the shrinking and drying process is unfinished quite other influences come into play. PERMEABLE AND IMPERMEABLE SEEDS 71 Up to this point we have been dealing with the seed in The influence its coats. But if we remove these coverings we find another coats on the singular contrast in the behaviour of these two seeds. In thesfed's" the instance of Canavalia ensiformis we discover that it makes weight, no essential difference whether we employ the seed in its coverings, or puncture it through its coats, or deprive it of them altogether. In any case the same average weight is maintained, the baring of the kernel or puncturing of the coats merely resulting in a small increase of the hygroscopic range. Whatever may be the function of the coats of a permeable seed, they do not prevent it from responding from («) in the day to day to the variations in the atmospheric humidity, permeable though they may regulate the process. This would seem ^^^ ' to be true of the large majority of similar seeds, and it follows naturally from the permeable character and the hygrometric behaviour of the seed-coverings. (See Note 6 of the Appendix.) In the table below I have given the results of simultaneous observations on the bared, punctured, and entire seeds of Canavalia ensiformis collected from the plant at the same time. The table explains itself, except that one may add that the hygroscopic or hygrometric variation is the range of the changes in weight exhibited in the course of two or three weeks stated as a percentage of the total weight. Comparison of the Behaviour of the Seeds of Canavalia ENSIFORMIS when BARED, PUNCTURED, AND ENTIRE. Condition of seed. Hygroscopic variation. Loss of weight in two years. Entire in its coats .... Punctured through its coats Bared of its coats .... 2 '5 per cent. 3'o ,, 4 -0-4 "5 per cent. Nil. If we repeat these experiments with the seeds of Guilandina (6) in the bonducella^ and either remove the hard, impermeable, shell-like impermeable coats or pierce them with a file, we obtain results of quite a ^^^^' 72 STUDIES IN SEEDS AND FRUITS different nature. A sample of the bared kernels weighing lOO grains immediately after the removal of their shells will be found after a period of four or five days to have increased its weight to iii or 112 grains. The gain in weight begins as soon as the hard coats are removed ; and thus my materials became sensibly heavier whilst in prepara- tion for the balance. In one case, for instance, a sample of 500 grains weighed 503 grains after an hour occupied in preparing it for an experiment. This increase in weight is maintained, although in a diminished degree, for a long period. In fact, the bared kernel never returns to the weight it possessed when enclosed in its impermeable cover- ings. As is shown in the table below, after a period of a year and more, it is still 3 or 4 per cent, in excess of its original weight. Results of the Exposure to Air of the bared Kernels of GUILANDINA BONDUCELLA. (WeIGHT IN GRAINS OF SIX KERNELS AT VARIOUS Periods.) Immedi- ately after being bared of their shells. After 5 days. After 20 days. After months. After 6 months. After 16 months. After 20 months. After 26 months. 100 113-4 iiro 107-3 107-0 103-0 io3"3 104-4 The explanation, of course, is simple. The kernel when bared, being in a state of ultra-dryness, supplies its deficiency by absorbing water from the air. In so doing it has changed its nature and now responds to the hygrometric variations of the weather, behaving in fact like the kernel of a permeable seed. We have here, then, the disclosure of another striking character distinguishing impermeable seeds, such as those of Guilandina bonducella^ from permeable seeds, like those of Canavalia ensiformis. This curious quaHty has been exhibited in varying degrees PERMEABLE AND IMPERMEABLE SEEDS 73 by the bared kernels of nearly all impermeable seeds that have been subjected to this test. The dry friable kernels when coarsely broken up contrast greatly in their appearance with the relatively moist and compact materials of permeable seeds ; and hence it could be presumed without further inquiry that absorption of water-vapour from the air is the cause of the subsequent increase in weight. But this capacity in seeds of becoming considerably heavier when exposed to the air than when locked up in their impermeable coats pre-supposes a condition of ultra-dryness within the seed itself. We should thus expect that the seeds of Guilandina bonducella, as types of impermeable seeds, would contain much less water than typical permeable seeds, such as those of Canavalia ensiformis. We have accordingly to appeal to the evidence of the oven An appeal to in order to interpret the indications of the balance ; and in the of the oven, table now to be given are to be found the results of exposing these two kinds of seeds to a temperature of ioo° to 105° C. for a period of from one and a half to two hours. Water-contents of a Typical Permeable and a Typical Im- permeable Leguminous Seed as ascertained by Exposure to A Temperature of 100° to 105° C. for i| to 2 hours. Character of seed. Number of experiments. Average water - contents. Range of result. Canavalia ensiformis . Guilandina bonducella . Permeable Impermeable 4 5 16 per cent. 8 ,, 14 to 18 percent. 6 to 10 ,, Note. — These results represent the combined water-contents of kernel and coats, have omitted decimal fractional parts, as these values are dealt with in later chapters. These results establish the ultra-dryness of the seed of Guilandina bonducella enclosed in its impermeable coverings ; and we recognise in the absorption of water-vapour from the air by the bared seed an attempt to assume the condition of a permeable seed. As far as their water-contents are concerned, 74 STUDIES IN SEEDS AND FRUITS permeable hygroscopic seeds are, relatively speaking, in a state of saturation, or perhaps it would be more correct to say in a state of equilibrium, with regard to the moisture of the air. On the other hand, impermeable seeds like those of Guilandina honducella stand in no such relation to the atmosphere, and preserve their abnormally dry condi- tion independently of any atmospheric changes. When, by the removal of their coverings, such seeds have been deprived of their power of resisting the permeation of water, either as vapour or as liquid, they rapidly supply their deficiency by absorbing it from the air. Roughly speaking, the amount of water regained from the air by the bared seed of Guilandina honducella represents the deficiency in its water-contents, as compared with the bared seed of Canavalia ensiformis^ or, in other words, the price of its impermeability. As regards their water-contents and other characters diagnostic of permeable seeds, the seeds of Canavalia ensiformis may be placed with our edible leguminous seeds, such as Peas, Broad Beans, Scarlet-runners (Pisum sativum, Faba vulgaris, Phaseolus multiflorus), which usually contain 15 or 16 per cent. of water. But the low percentage of water in the seeds of Guilandina honducella appears quite abnormal when compared with the data given in the ordinary tables of the analyses of seeds used as food, though representative of impermeable seeds. Rice, Maize, Wheat, and other cereal grains, for the most part permeable, contain from 12 to 15 per cent. of water, whilst the flour yielded by them holds 1 1 or 12 per cent. Up to this point the indications appear to be sufficiently plain, though the subject gives promise of much complexity. But now comes another curious fact. Whilst the coats of a permeable seed like that of Canavalia ensiformis behave hygro- scopically when removed from the seed, neither increasing nor decreasing their previous average weight, it is very different with the impermeable seed. PERMEABLE AND IMPERMEABLE SEEDS 75 The detached shell-like coverings of the seed of Guilandina bonducella possess the same quality, of ultra-dryness and display the same absorptive capacity in air as the bared kernel, though to a less degree. They exhibit the relation between the water- contents which we might have expected, the larger water-per- centage of the seed-coverings being associated with a smaller absorptive capacity in the air as compared with the kernel. These qualities are well brought out in the tabulated results given below of an experiment in Grenada in which the shell and the kernel of several seeds were equally divided in each case between two samples, so that the air exposure and oven- tests were applied to truly mixed samples. It may here be added that, as in the case of the kernels, the detached shell re- tains its excess for a long period, though in a diminishing ratio. In one experiment, after a lapse of twenty-one months it still weighed 6 per cent, heavier than when first removed. The shell or covering of the seed of Guilandina bonducella has the same quality of ultra-dryness and the same absorptive capacity as the kernel. Comparison of the Absorptive Capacities in Air with the Water- contents OF freshly bared Kernels and of the Detached Seed-shells or Coverings of Guilandina bonducella, the Samples being truly mixed, as above described. Gain in weight after exposure to the air for 4 days. Original water- contents as determined in the oven. Seed-shells . Kernels 12*0 per cent. l6-2 „ 7 "6 per cent. 4*2 We are thus brought face to face with the curious cir- cumstance that if we break open one of the seeds of Guilandina bonducella and allow it to remain in this condition for a few days it will increase its weight on the average by 1 1 or 12 per cent. It is essential to break through the seed-shell, it being immaterial whether the shell is in a few or in many pieces, or whether the kernel is left whole or in fragments. 76 STUDIES IN SEEDS AND FRUITS The same effect is produced by puncturing or filing through the shell, though, as shown in the results tabulated below, the change is much more gradual. Here the total increase of weight is the result of the combined absorptive capacities of the kernel and its coverings. Results of Filing into the Shell or Hard Covering of the Seeds of Guilandina bonducella, stated as a Percentage of the Original Weight of the Seeds (34 to 35 Grains). ^ HlN . 1 ■5 i ■5 «• ^ 1 to* c S }> c "S a '& '^ ■) (F) „ (near) glabra 55 (F) „ melanosperma 55 (S) Ipomcea dissecta Convolvulaceas (S) „ pes-capr^ 55 (F) „ tuba 55 (F) Leucaena glauca Leguminosae (S) Mucuna urens 55 (F) Sophora tomentosa 55 (F) Strongylodon lucidum 55 (F) Ulex europasus 55 (S) Vigna luteola 55 (F) 2\rofe.— The capital letters in brackets have the following significations relating to buoyancy in sea-water : — F = Known to be dispersed by sea- currents, the proportion of buoyant seeds varying from as much as 80 or 90 per cent, in Guilandina bonducella to as little as 10 per cent, in Dioclea 7-eftexa. However, seeds vary much in this respect in different localities. From my observation of the living plant in Fiji and Ecuador I formed the conclusion that quite half of the seeds of Entada scandens have no initial buoyancy ; whereas of fresh seeds obtained from the plants growing in the Jamaica woods I found that quite 90 per cent, floated (w'ofe the author's Plant Dispersal, p. 181). S = A11 sink. 94 STUDIES IN SEEDS AND FRUITS II. Variable LeguminosjE Ranunculaceae Caryophyllaceae Leguminosae 95 per cent, impe 70 50 80 80 30 50 40 30 35 (Possessing both permeable and impermeable seeds) Abrus precatorius Acacia Farnesiana Albizzia Lebbek Aquilegia (species) Arenaria peploides Bauhinia (species) Caesalpinia Sappan „ sepiaria „ Calliandra Saman „ Canavalia gladiata (red „ seeds) „ obtusifolia „ „ _ (species) „ Canna indica Cannaceae Cassia fistula Leguminosae „ grandis ' „ „ marginata „ Entada polystachya „ Enterolobium cyclocarpum „ Erythrina corallodendron „ „ indica „ „ velutina „ Ipomcea tuberosa Convolvulaceae Poinciana regia Leguminosae Thespesia populnea Malvaceae Vicia sepium Leguminosae 70 50 90 90 90 90 50 90 60 50 65 35 65 80 65 meable III. Permeable Achras Sapota (Sapodilla) iEsculus Hippocastanum (Horse-chestnut) Allium ursinum Andira inermis Anona Cherimolia (Cherimoya) „ muricata (Sour-sop) „ palustris (Monkey-apple) „ reticulata (Custard-apple) „ squamosa (Sweet-sop) Artocarpus incisa (Bread-fruit) Arum maculatum Barringtonia speciosa Sapotaceae Hippocastanene Liliaceae Leguminosae Anonaceae Artocarpeae Aroideae Myrtaceae PERMEABILITY AND CLASSIFICATION 95 Berberis (species) Bignonia (2 species) Blighia sapida Cajanus indicus Canavalia ensiformis Cardiospermum grandiflorum „ Halicacabum Carica Papaya (Papaw) Chrysophylluni Cainito (Star-apple) Citrus decumana (Shaddock) Crinum (species) Datura Stramonium Dolichos Lablab Faba vulgaris (Broad Bean) Fevillea cordifolia Gossypium hirsutum Grias cauli flora Hedera Helix (Ivy) Hibiscus elatus „ esculentus „ Sabdarifa Hura crepitans Iris foetidissima „ Pseudacorus Lonicera Periclymenum (Honeysuckle) Luffa acutangula Mammea americana Momordica Charantia Monstera pertusa Montrichardia arborescens Moringa pterygosperma Moronobea coccinea Opuntia Tuna Phaseolus multiflorus (Scarlet-runner) „ vulgaris (French Bean) Pisum sativum (Pea) Pithecolobium filicifolium Primula veris (Primrose) Pyrus Malus (Apple) Ouercus Robur (Oak) Ravenala madagascariensis Ribes grossularia (Gooseberry) Ricinus communis (Castor-oil) Scilla nutans Berberideae Bignoniaceae Sapindaceae Leguminosa Sapindaceae Papayaceae Sapotaceae Aurantiaceae Amaryllideae Solanaceae Leguminosae Cucurbitaceae Malvaceae Myrtaceae Araliaceae Malvaceae Euphorbiaceae Irideae Capri foliaceae Cucurbitaceae Gutti ferae Cucurbitaceae Aroideas T> Capparideae Guttiferae Cactaceae Leguminosae Primulaceae Rosaceae Cupuliferae Musacea? Ribesiaceae Euphorbiaceae Liliacea? 96 STUDIES IN SEEDS AND FRUITS Stellaria Holostea Caryophyllaceae Swietenia Mahogani (Mahogany) Meliaceas * Tamarindus indica (Tamarind) Leguminosae Tamus communis Dioscoreae Theobroma Cacao (Cocoa) Buttneriae Vicia sativa Leguminosae Results concerning some of the above Impermeable and Variable Seeds from Professor Ewart's T'ables [Proc. Roy. Soc. Vict. 1908). His results for seeds more than 1 5 or 16 years old are not given. Adenanthera pavonina^ seeds 8 years old, swelled after filing. JIbizzia Lebbek, seeds 1 1 years old, scratching needed for germination. Canavalia gladiata^ 10 years old, i out of 6 seeds required filing for germination. Canavalia obtusifolia., 16 years old, required sulphuric acid for germination. Erythrina indica^ of 50 seeds, 8 years old, 6 swelled in water. Guilandina bonducella^ seeds 15 years old, required the acid for swelling. Leucana glauca., seeds 1 5 years old, all swelled in water. Mucuna urem^ lO years old, required filing for swelling. Poinciana regia., 9 years old, outer skin impermeable until filed. Imperme- Those who have studied the dispersal of seeds by the dispersafby ocean-currents have laid stress on the circumstance that many- water, of the seeds capable of transportal over wide tracts of sea belong to leguminous plants ; and I need here only allude to the circumstance that the four West Indian and Central American seeds {Dioclea reflexa^ Mucuna urens, Guilandina bonducella^ Entada scandens) that are most frequently stranded intact on the western shores of Europe belong to this order. When Professor Ewart remarked (p. 1 84) that " macro- biotic " seeds show no special adaptation for dispersal and that " none are wind or water-borne," he apparently had forgotten that there are included in his list the seeds of plants like Canavalia obtusifolia, Erythrina indica., and Guilandina * Tamarind seeds absorb water very slowly at first, requiring often an immersion of a week or more before there is any marked increase in the weight. PERMEABILITY AND CLASSIFICATION 97 bonducellay that have long been known to be dispersed by ocean-currents. Putting aside the question of adaptation to modes of distribution, a view which I hold ought either to be universally applied or to be discarded altogether, the strand of a coral island would be deprived of several of its most conspicuous and typical plants, such as Canavalia obtusifolia^ Coluhr'ma asiatka, Erythrina indica, Guilandina bonducella^ Ipomcea pes-capr 7 » Gain .^5 ^i in weight 15-1 grains i6-8 55 177 55 17-0 55 17-2 55 17-2 per cent Although five or six days are usually sufficient in the case of this and other impermeable seeds for the attainment of the The range of maximum weight, the period may be as short as three or as of weighto^f long as ten days, the time being extended or shortened by kernSs^of ^^^ relative dryness or humidity of the air. The varying Guilandina hygrometric conditions of the air also account for some of the differences between the results of experiments, but only to the extent of 2 or 3 per cent., which represents the ordinary range of hygroscopicity. The results of six experiments on the bared kernels of Guilandina bonducella^ mostly in the West Indies, the average weight of a kernel being 1 6 or 1 7 grains, are given below. Results of Experiments on the Bared Kernels of Guilandina bonducella, showing the increase in weight by the absorp- TION OF Aqueous Vapour after an Exposure of a few Days TO THE Air. Nunmber of kernels. Locality of experiment. Gain of weight in air. Jamaica Grenada England Grenada 10 '2 per cent. 17-6 „ I7"2 ,> 15-0 „ IO*2 ,, i6-2 „ ADDITIONAL EVIDENCE 117 These results indicate a range of from 10 to 17 per cent, in the increase of weight which recently collected seeds of Guilandina bonducella experience on being deprived of their coverings. If we allow for the usual hygroscopic reaction, this would probably represent a true range of 1 2 to 1 5 per cent. One other impermeable leguminous seed may here be specially mentioned in connection with the variations in its increase of weight on being exposed to the air after being deprived of its coverings. Four samples of kernels of Entada The range scandens weighing from 100 to 260 grains increased their scandens* weight during an exposure to the air of from four to ten days by 4*2, 5*7, 6-8, and 12*2 per cent. The last result was obtained during humid weather in Jamaica ; and it is evident from the progress of the experiment which is shown below that if allowance is made for the hygroscopic reaction (that is, by deducting half the variation), the excess weight would not have been much over 10 per cent. I here append the particulars of this experiment in Jamaica on the bared kernel of Entada scandens, which is noteworthy as illustrating the rate of increase and the effect of the ordinary hygroscopic reaction on the materials. Origi nal weight 100 After 7 hours 102-3 55 I day . 106-5 )> 2 days I09-I 55 3 55 • 1 10-8 55 4 55 • I 12-2 55 5 55 • 112-25 55 6 55 • 110-8 )5 7 55 • iio-o 55 10 55 • III-5 55 14 55 • I08-I » 16 •>•> ' ' • III-I grains The average results of my experiments on the bared kernels of these and other leguminous impermeable seeds are General tabulated below, together with those for two species of impermeable Ipomcea ; and it is of importance to note in passing that seeds seeds. STUDIES IN SEEDS AND FRUITS of such a different order (Convolvulaceae) display the same quality when impermeable. Table showing the Usual Increase of Weight through the Absorption of Water from the Air displayed by Impermeable Seeds, either after being bared of their Coverings or after being cut across in their Coats. (All are leguminous excepting the two last.) Gain in weight after exposure to Name of species, with average the air for to 7 days. Locality weight of a single seed. Bared kernels. Cut in halves in their coats. experiment. Adenanthera pavonina (5 grains) 2*9 per cent. ... England, Dioclea reflexa (100 grains) . io'6 ,, 60 „ .... Grenada. England. Entada scandens (400 grains) 5'S ,. lo-o ,, '.'.'. Jamaica. Guilandina bonduc (50 grains) . 8-2 ,, ... England. ( IS'2 .. Grenada and , , bonducella (40 grains) A Jamaica. [ IO'2 ,, England. ,, glabra (65 grains) 6-0 „ ... „ Leucaena glauca (o'8 grains) 5 '0 per cent. Grenada. Mucuna urens (90 grains) . 6 '2 per cent. England. Strongylodon lucidum (40 grains) 5 '2 per cent. ,, Ulex europseus (o'l grain) . 5"o .. ,, Ipomoea dissecta (2-5 grains) 6-0 per cent. Jamaica. „ pes-caprse (3 grains) 4 'o per cent. England. The seeds in the foregoing list vary greatly in size and weight, from those of Leuc^na glauca, which average only 0-8 of a grain, to those of Entada scandens, which average 400 grains. The samples of kernels used were generally 50 to 100 grains, but greater in the case of the large seeds. The hygroscopic reaction is as far as possible excluded. It will be inferred that it is not possible to strike an average increase of weight for the bared kernels of impermeable seeds when exposed to the air. Each kind of seed has its own regime in this respect, which is influenced not only by the relative dryness of the kernel, but also by the amount of oil it contains. This probably explains the small excess weight of the seeds of Adenanthera pavonina. ADDITIONAL EVIDENCE 119 Speaking very generally, however, we may infer that legu- minous impermeable seeds when bared commonly increase their weight from 5 to 10 per cent, by abstracting moisture from the air. A question of interest here presents itself as to the duration of the ultra-dryness of the kernels of impermeable seeds. My materials for furnishing an answer, though insufficient, tend to show that this condition may be main- Duration of tained for several years. A seed of Mucuna urens, gathered dJyness^of by me from the plant in Hawaii eleven years before, ii"penneable increased its weight when bared of its coats in England 6-5 per cent, in ten days. In the same way, the bared kernels of two seeds of Guilandina bonducella obtained by me in Fiji ten years before added 7*2 per cent, to their weight ; and a seed of Strongylodon lucidum, picked up amongst the drift on a Fijian beach eleven years before, and perhaps a year or two old then, increased its weight by 6 per cent, when bared in England of its coats. In the case of the two last-named species, seeds collected at the same time and place and tested for germination at the time of the above experiment germinated healthily and supplied plants for my greenhouse. There is but slight indication here of any marked decrease in the ultra-dryness of impermeable seeds during a period of ten or eleven years. The increase in weight {6' c^ per cent.) of the seed of Mucuna urens is rather above the average (6'0 per cent.) for three seeds, six to eighteen months old, which were also tested in England. On the other hand, the rate of increase for the Fijian seeds of Guilandina bonducella ten years old (7*2 per cent.) is considerably under the average for the tropics (15 per cent.). However, the contrast is not nearly so great as it appears, as the Fijian seeds were experimented on in England, and, as shown in the table below, the rate of increase of the weight of the bared seeds of Guilandina bonducella in a temperate climate would average only about 10 per cent. I20 STUDIES IN SEEDS AND FRUITS There are two other points to be referred to in connec- tion with the behaviour of the bared kernels of impermeable seeds, namely, the respective influences of tropical and temperate climates on the gain in weight in air, and the duration of this excess weight. We would expect the bared kernel of a tropical seed to gain more water from the air in the more humid climate of the West Indies than in the drier climate of the south of England. We should also expect the excess in weight to be permanent yet subject to the ordinary hygroscopic reaction, as long as the seed retains its vitality. Results of Experiments on the Seeds of the same Plant in the Tropics (West Indies), and in the South of England. Place of experiment. Gain of weight in air of bared kernel. Jamaica io'2 per cent. 17-6 Guilandina bonducella , Grenada England I7'2 16 -2 10*2 ,, 4'2 Entada scandens . " I'-i :: Dioclea reflexa Jamaica Grenada England IO-2 ,, 10-6 ,, 6-0 The in- fluence of a temperate climate on the absorp- tive capacity of bared impermeable tropical seeds. The first point is illustrated in the foregoing table. Since the seeds there referred to, as well as those named below, are all tropical, the question, as far as this investigation is con- cerned, relates to the influence of a temperate climate on the capacity of the bared kernels of impermeable tropical seeds of increasing their weight by absorbing water from the air. The data of the table indicate that the absorptive capacity is diminished in temperate climates. The next point is concerned with the permanence of the excess weight acquired by the exposure to air of the ADDITIONAL EVIDENCE 121 bared kernels of impermeable seeds. This has already been noticed in the case of the seeds of Guilandina bondu- cella in Chapter IV, where it is shown that after the first gain of 13 per cent, the weight began to diminish slowly, The degree though even after two years there was still an excess of man?nce of 3 per cent., allowing for the hygroscopic variation. This theexcess loss of the excess weight in time is not at first sight easily acquired by explained. However, since the bared seed in absorbing me'abie water from the air assumes the role of the kernel of a ^^^^^• permeable seed, it is likely that light may be thrown on it when we come to discuss the final fate of permeable seeds in time. On the other hand, a different indication is offered where the seed is punctured or filed, when the gain in weight takes place very slowly. Thus it is shown in the table of results given in Chapter IV for punctured seeds of Guilandina hondu- cella that the punctured seeds occupied some months in, reaching the maximum excess weight of 10 or 11 per cent., and even after two years were still 7 or 8 per cent, heavier than before they were punctured or filed. However, experiments of this kind being always con- ducted under dry conditions are by no means imitations of what occurs in nature, though they indicate latent properties or potentialities of impermeable seeds. In the home of the plant, such a seed, if deprived by some defect or injury of the proper protection of its impervious cover- ings, would either pass on to the germinating stage or would become mouldy and decay. But it is only with those seeds where there is a great increase in weight, such as occurs with the bared kernels of Guilandina bonducella^ that one can test the duration of the excess weight by eliminating the ordinary hygroscopic reaction of 2 or 3 per cent. Impermeable seeds, when deprived of their coats, gather weight during the first week or two independently to some degree of the atmospheric conditions. After this they respond normally to the changes in the hygrometric state of 122 STUDIES IN SEEDS AND FRUITS the air ; and if the excess weight, due to the absorption of aqueous vapour by the ultra-dry kernel, is only 3 or 4 per cent., it is difficult to exclude the disturbing influence of the hygroscopic reaction. Such experiments in the tropics are likely to be terminated by attacks of mould, and even in England it is necessary that they should be carried out in a dry room. The appearance of mould is usually preceded by a marked increase in weight. The bared kernels of an inland Jamaican species of Guilandina gained about 6 per cent. in weight during the first ten days, and, subject to slight variation, preserved this excess for about two months, when very damp weather followed, and the seeds, after having augmented their weight to 10 per cent., were attacked by mould. With regard to the " variable " group of seeds, where both permeable and impermeable seeds occur in the same plant, only a few remarks will be needed before giving the tabulated results of my observations. As concerning the The capacity bared kernel's capacity for absorbing water from the air, kernelso?'^ these seeds exhibit the extreme behaviour of the perme- variable ^JqXq and impermeable seeds, in the first case merely the seeds of . ^ . . . . absorbing ordinary hygroscopic variation or i or i'5 per cent, on the air. '^°'" either side of the mean, in the second case a marked and permanent increase often of 10 per cent, or more. If we had to handle two samples of seeds from the same plant which presented this great contrast in behaviour, we should at once know that one sample consisted only of permeable seeds and the other sample only of impermeable seeds. Almost always, however, the sample would be mixed, and then we should get an intermediate result, for instance, an average increase of weight, after allowing for the hygro- scopic reaction, of 4 or 5 per cent. Some seedsman, more practical than the author, might be able to make a scale of comparison which could be used for proving his seeds ; but it would be requisite to have a standard of comparison for each species. ADDITIONAL EVIDENCE 123 Results of Observations on the Capacity of Variable Seeds of INCREASING THEIR WeIGHT BY ABSORBING WaTER FROM THE AlR, EITHER AFTER BEING COMPLETELY DEPRIVED OF THEIR COATS, OR AFTER BEING CUT ACROSS IN THEIR COATS. (The hygrOSCOpic reaction is excluded.) Note. — The term "variable" is applied when a plant produces both permeable and impermeable seeds. With small seeds it is often more convenient to expose them to the air cut in halves than to bare their kernels. The difference in the results of the two methods is not very great, and will be dealt with in the next chapter. The letters A, B, C, indicate only approximate estimates. A. Sample where most seeds are permeable. B. ,, ,, impermeable. C. ,, they are equally mixed. Gain in weight after exposure to a, 1 air for 5 days or more. Locality of experi- cy: Bared kernels. Cut in halves in their coats. ment. Abrus precatorius Acacia Farnesiana A B I -o per cent. 5'o England. C 2'o percent. Grenada. Albizzia Lebbek . C 2-2 ,, Bauhinia (species) A Hygroscopic only England. C^salpinia Sappan . . -j A B 1 -o per cent. 9*0 ,, Grenada. ,, sepiaria . . j A B Hygroscopic only /■Q per cent. Jamaica. Canavalia gladiata . . i A B Hygroscopic only 5 "o per cent. England. ,, obtusifolia . B 6-0 percent. [[ " Canna indica Hygroscopic only Cassia fistula B ... 3-0 percent. ^l ,, marginata B ... i'9 ,, Entada polystachya . . | A B B I '6 per cent. 9"4 3 '5 per cent. Grenada. Enterolobium cyclocarpum . X B / 3-3 percent. \ England. Erythrina corallodendron , - c C { (2-6) ;; } 2-4 percent. " ' A o"S Grenada. A 2'3 .. jj . B 6-2 ,, indica . C C / 5 '0 per cent. \ 1(4-2) „ / 4*5 England. ,, velutina B lO'O ,, Jamaica. Ipomoea tuberosa . . \ A B Hygroscopic only 4 '6 per cent. England. Poinciana regia . A Hygroscopic only '' Note. — The figures in parenthesis in the "bared kernels " column indicate the result when the coats are included, thus enabling a comparison to be made with the data in the next column. 124 STUDIES IN SEEDS AND FRUITS The effect of baring a permeable seed has been already referred to in Chapter IV in the instance of Canavalia ensiformis. Since the kernel is placed in hygrometric relations with the atmosphere by its porous coats, one would not look for any marked result with seeds that have completed the drying process. Indeed, the immediate effect on a seed that has reached a stable weight is merely to give a rather freer play to its hygroscopicity. There is, as one would expect, no attempt to permanently increase its weight. The contrary is, in fact, the case with a seed that has yet water to yield to the air, since the drying process is accelerated by the removal of its coverings. The contrast between permeable and imper- meable seeds in this respect is well exhibited in those plants producing both types capable of being readily distinguished by the eye, as shown in the results below tabulated. Character of seed. Effect of exposing the bared kernels to the air for 4 or 5 days, stated as a percentage of the original weight. Entada polystachya . Csesalpinia Sappan . Permeable Impermeable Permeable Impermeable Gained i -6 per cent. ; mainly hygro- scopic. Gained 9*4 per cent. Varied only 07 per cent. ; entirely hygroscopic. Gained 9 "o per cent. I made a large number of observations on the effect of baring the kernels of permeable completely air-dry seeds, on the results of which are based the above general conclusions. As examples of permeable seeds which merely continue to behave hygroscopically on the removal of their coverings, though often in an increased degree, the following may be cited : — Achras Sapota (Sapodilla) Anona muricata (Sour-sop) „ palustris „ reticulata (Custard Apple) „ squamosa (Sweet-sop) Canavalia ensiformis ADDITIONAL EVIDENCE 125 Cardiospermum grandiflorum Chrysophyllum Cainito (Star Apple) Citrus decumana (Shaddock) Dolichos Lablab Faba vulgaris (Broad Bean) Hura crepitans (Sand-box Tree) Luffa acutangula (Loofah) Phaseolus multiflorus (Scarlet-runner) Pisum sativum (Pea) Ricinus communis (Castor-oil) Of the thirteen genera here named five are leguminous and the rest belong to a variety of other families. This list is simply intended to illustrate the subject. A number of additional examples might have been given ; whilst others, like the seeds of the Horse-chestnut {Msculus Hippocastanum) and of Acorns (Quercus), will be more fittingly dealt with in discussing the drying process of permeable seeds. In this connection it should be observed that this matter has only been handled here in so far as it brings out the contrast in behaviour between permeable and impermeable seeds when deprived of their coverings. The ultra-dryness of impermeable seeds as compared with Additional permeable seeds which has been disclosed by the various theassoda- experiments above discussed is confirmed by the evidence ^\°"a-dr^ness supplied when the seeds of the different types are exposed to ofimperme- a temperature of 100° C. In other words, it is associated with with a low a low water-percentage. This was established for the seeds of ^ntage^'^' Guilandina bonducella in Chapter IV. Here I will illustrate it by a number of fresh examples and will discuss the subject, therefore, from a more general standpoint. For this purpose all my results for the three types of seeds are given in the table in a later page of this chapter. There is but little significance in this feature of impermeable seeds until it comes to be contrasted with the behaviour of permeable seeds ; and even then the contrast must be made with discretion. For instance, if we were to compare imper- meable leguminous seeds indiscriminately with permeable seeds 126 STUDIES IN SEEDS AND FRUITS of other orders, we should find that there is often no sort of relation between them as regards the capacity of absorbing water from the air in the broken condition and the actual water-contents as indicated by the loss of weight in the oven. For example, an average impermeable seed which contained 9 per cent, of water would be able to increase its weight by about 7 per cent, when broken up and exposed to the air. It would be ultra-dry in the entire condition to that extent. On the other hand, this percentage of water in oily seeds like those of Ricinus or Hura or EUis would be no indication of dryness in the seed, since except for the hygroscopic variation they would remain unchanged in weight on exposure to the air in the broken condition. Here matters are on quite another plane, and for a valid comparison of seeds of different orders we must not look in this direction. It is therefore requisite, if we wish to connect the seed's capacity of in- creasing its weight by abstracting water from the air with its deficient water-contents, that we should restrict the comparison to seeds of the same order. In this case we take leguminous seeds ; but even here disturbing influences may come into play, though they are more easily avoided. So, confining ourselves at present to the Leguminosae, we will at first refer to the indications afforded by impermeable seeds in the table that a seed's capacity of increasing its weight when bared of its coats or in the broken condition is determined by a low water-percentage. We can see at once in the results for impermeable leguminous seeds that the seeds which lose least weight when submitted to a temperature of ioo° C. are those which add most to their weight when exposed unprotected to the air. Thus, we see that the three kinds of seeds with the lowest water-percentage, Dioclea reflexa^ Guilandina honducella^ and G. bonduc^ are those which add most to their weight when exposed to the air. On the average these seeds with a water-percentage of 7*5 per cent, add 9-2 per cent, to their weight when exposed in the broken condition to the air. The other impermeable ADDITIONAL EVIDENCE 127 leguminous seeds {Adenanthera pavonina^ Entada scandens, Mucuna urens) give an average gain in air of 5-0 per cent., with an average water-percentage of 11*7. These results may be accepted tentatively as representing the average behaviour of impermeable seeds when broken up and exposed to the air, viz. : Seeds holding 11-7 per cent, of water gain 5-0 per cent. At best this is only a rough indication, as each seed has a regime of its own in this respect. The real significance of these figures becomes more apparent when we contrast them generally with those for permeable seeds of the same order, taking as our examples the seeds of Canavalia ensiformis^ Faba vulgaris (Broad Bean), Phaseolus multiflorus (Scarlet-runner), and Pisum sativum (Pea), which hold on the average about 1 5 per cent, of water when the drying process is complete, and make no permanent addition to their weight when broken up or cut open or laid bare and exposed to the air. Contrasted with impermeable seeds we get these general results : — Impermeable seeds holding 7*5 per cent, of water add 9*2 per cent. to their weight. Impermeable seeds holding 11-7 per cent, of water add 5-0 per cent. to their weight. Permeable seeds holding 15-0 per cent, of water add o-o, behaving hygroscopically. Numerous disturbing influences come into play in making a rough estimate, such as that given above ; but its general indications are confirmed by the results obtained from experi- ments in which such influences are largely eliminated, namely, Theelimina- by contrasting the seeds of the same plant in those species tm-bing in- where both permeable and impermeable seeds are represented, Auences by namely, in the variable group. But even here, as indicated seeds of the in the table, we must be able to distinguish between samples wh^re both containing very diff^erent proportions of these two kinds of typ^s occur, seeds. There is a practical difficulty in ascertaining a seed's 128 STUDIES IN SEEDS AND FRUITS impermeability in water before testing the amount of its water- contents, and this difficulty is very apt to arise in dealing with variable seeds, notably in the seeds of Poinciana regia^ which behave almost like permeable seeds. It is to seeds like those of C^salpinia Sappan and Entada polystachya, where we can with some confidence distinguish the two types of seeds by their external characters before the experiment, that we must appeal. In their case it is plainly shown in the table that the seeds which imbibe in the broken condition most water from the air are those which lose least water in the oven, or, in other words, that the ultra-dryness of impermeable leguminous seeds is simply a diminution in the water-contents as compared with permeable seeds. Thus we find for Casalp'tnta Sappan that when the seeds held about 14 per cent, of water they did not increase their weight when ex- posed in a broken state to the air. On the other hand, when their water-contents amounted to less than 10 per cent, they increased their weight about 9 per cent, by abstracting water from the air. Similar results were obtained for Entada polystachya. Thus : (Seeds with 14 per cent, of water merely behave hygro- scopically when broken. Seeds with 9-7 per cent, of water add 9 per cent, to their weight when broken. (Seeds with 10 per cent, of water add i'6 per cent, to their weight when broken. Seeds with 6 per cent, of water add 9*4 per cent, to their weight when broken. The data given in the table for permeable seeds of other than leguminous plants are interesting, as they illustrate the fact that many permeable seeds may hold as little water as some of the impermeable leguminous seeds. This is particularly clear when we distinguish between the coats and the kernel, as is done in the table. Here we find that the kernels of permeable seeds like those of Citrus^ Hura, etc., may hold less than 9 per cent, of water. Doubtless the presence of oil goes ADDITIONAL EVIDENCE 129 to explain this low water-percentage ; but at all events this fact shows how necessary it was to avoid comparing seeds of different families when connecting impermeability with diminished water-contents. I come now to the additional evidence in support of the Further principle typified by Guilandina honducella in Chapter IV, that show that in the seed-coverings of impermeable seeds possess the same s'JJ'ds'Sfe^'^'^ quality of ultra-dryness as the kernel, though often in a ^^^gg^J^^^.^ somewhat diminished degree, and the same quality of supply- same quaHty ing the deficiency by absorbing water from the air, the nessas'thT larger water-percentage of the coats being usually associated kernel, with a diminished absorptive capacity of the freshly exposed material. Most of my results are given in the table a few pages later ; but I will confine the discussion as before to legu- minous impermeable seeds. All the kinds of seeds there tested possess this quality of ultra-dryness for the coats as well as the kernel, though the presence of oil in the kernel of Adenanthera pavonina somewhat alters the regime. Most of the results represent the average of three or four or more experiments, the absorptive capacity in air and the water- percentage being determined independently. In spite of possible disturbing effects, due to variation in the seeds and in the atmospheric conditions, the data thus obtained go fairly well together. But in two cases, those of Guilandina honducella and Entada scandens^ this disturbing influence was removed by a simple expedient ; and these experiments have been specially added to the others, since they are not only the most critical but the most decisive. With Guilandina honducella the coats and kernel of each seed were divided between two samples, so that the water-percentage and the absorptive capacity in air were simultaneously determined from similar examples. With Entada scandens the two samples of the seed-coverings and the two samples of the kernel were obtained from one large seed weighing about 500 grains. They gave the following results for the water- 9 I30 STUDIES IN SEEDS AND FRUITS Additional evidence to show that the absorp- tive quality of broken impermeable seeds is not affected by exposure to a tempera- ture of lOO" C. whether in the case of the coats or of the kernel. percentage and for the fresh materials exposed to the air in a broken condition : — (Coats hold 7*6 per cent, of water and gain in weight 12 per cent. Kernels hold 4-2 per cent, of water and gain in weight 16 per cent. (Coats hold 13-8 per cent, of water and gain in weight 8 per cent. Kernels hold 7*5 per cent, of water and gain in weight 12 per cent. These two seeds illustrate what is shown by other im- permeable seeds in the table, namely, that the smaller absorptive capacity of the seed's coats is associated with a larger water-per- centage as compared with the kernel. Dioclea reflexa is irregular, however, in this respect. But the behaviour of variable seeds containing a good proportion of impermeable seeds supports the same conclusion. This is shown in the table by samples of seeds of C^salpinia Sappan^ Entadapolystachya^ and two species oiErythrina. In Chapter IV I have already referred to the circumstance that the capacity possessed by, impermeable leguminous seeds of considerably increasing their weight when exposed in the broken condition to the air is but little ajEFected by first sub- jecting the materials to a temperature of 100" C. for an hour or two. In that chapter I took the seeds of Guilandina honducella as a type. Here I will discuss the additional evidence for impermeable seeds of the same order. This double capacity was disclosed in a large number of experiments on impermeable leguminous seeds. In the table I have compared the two results obtained for seeds in the broken condition. In one column we have the gain in weight by ab- stracting moisture from the air when the materials are not heated. In another column we have the gain after the materials have been exposed to a temperature of 100" C. Many of the experiments on the absorptive capacity of the unheated and heated materials were carried out on different samples and under different climatic conditions, so that disturbing influences were likely to affect ADDITIONAL EVIDENCE 131 them, and a very close approximation between the two absorp- tions could not often be looked for. However, in the main, these independent experiments confirm the principle indicated by the seeds of Guilandina bonducella, that exposure to a temperature of 100° C. does not seriously affect the absorptive capacity. But to eliminate such disturbing conditions I made critical experiments on certain impermeable seeds in which the absorp- tive qualities of the unheated and heated materials were simul- taneously tested in similar samples. The seeds in question were those of Entada scandens^ Erythrina indica^ and Guilandina bondu- cella. In the case of Entada scandens^ one large seed weighing nearly 500 grains supplied all the material for the double experi- ment. In the cases of the two last named, each seed was divided between the two samples, the one for exposure without heat to the air, the other for exposure to the air after being subjected to a temperature of 100° C. The results were as follows : — A. Entada scandens Gain of coats and kernel: unheated, 10-8 per cent.; after 100° C, 9-5 per cent. Gain of coats alone : unheated, 8'2 per cent. ; after 100° C, 8'0 per cent. Gain of kernels alone : unheated, I2"3 per cent. ; after 100° C, 10*4 per cent. B. Guilandhia bonducella Gain of coats and kernel: unheated, 13-7 per cent. ; after 100° C, 14*2 per cent. Gain of coats alone : unheated, i2-o per cent. ; after 100° C, 13*3 per cent. Gain of kernel alone: unheated, 162 per cent.; after 100° C, 15-5 per cent. C. Erythrina ind'ica (impermeable seeds only selected) Gain of coats and kernel : unheated, 4-5 per cent. ; after 100° C, 3-4 per cent. Gain of coats alone : unheated, 2*3 per cent. ; after 100° C, i'6 per cent. Gain of kernel alone : unheated, 5-0 per cent. ; after 100° C, 4-1 per cent. 132 STUDIES IN SEEDS AND FRUITS From these similar samples we learn that as a rule exposure to a temperature of ioo° C. for from il to 2 hours but little afFected the capacities of either the seed-coverings or the kernel for increasing their weight by absorbing moisture from the air in the broken condition. It is scarcely worth while to labour this point. The same indications are supplied in the case of other impermeable seeds mentioned in the table, such as those of Dioclea reflexa, Guilandina honduc^ Mucuna urenSj etc., and by the samples of variable seeds where im- permeable seeds predominated, such as those of C^esalpinia Sappan, Erythrina corallodendron^ etc. Coming to permeable seeds of the leguminous type, we notice in the table that during the period of five days following the exposure to a temperature of ioo° C, they all regained from the air most of the water lost in the oven. That they failed to return to the original weight is doubtless to be attributed to the limited duration of their exposure to the air, since it is clearly shown in the instances of Faha vulgaris^ Phaseolus multifiorus^ and Pisum sativum on p. 142 that if the test had covered a period of a week or two instead of only five days, the seeds would have regained their original weight. But they would not have displayed any excess, except such as is included in the ordinary hygroscopic varia- tion of about 3 per cent., and this is the great point of contrast between permeable and impermeable seeds. As respecting permeable seeds other than those of the Leguminosae, the data for several kinds given in the table tell much the same story. After exposure to the same tempera- ture of 100° C, they in most cases regained much of their lost weight by taking up water from the air, and, no doubt, if the test had been prolonged, they would have regained all. The behaviour of the seed-coverings of the Shaddock [Citrus decumand) is abnormal, but I can throw no light on it here. However, taking all the data for permeable seeds given in the table, it is evident that whether leguminous or otherwise, they as a general rule behave in the same way after being ADDITIONAL EVIDENCE 133 exposed to a temperature of 100° C. in the broken or divided condition. In five days they regain most of the weight lost, and in a week or two they would regain all, maintaining their original weight subject to ordinary hygroscopic variation. When comparing the absorptive capacities of permeable The absorp- and impermeable seeds it is requisite that the drying and cities of shrinking process should be complete. In all these experi- the'^drving" ments only seeds were employed that had accomplished the process has drying process and had attained a stable weight. If we place completed, in the oven a seed that has not yet begun to dry, or has not yet completed that process, we meet with a very different behaviour. A bared fresh Horse-chestnut seed {Msculus) cut up in slices, that had had its weight reduced in the oven from 100 to 52 grains, increased its weight by only 7 grains (52 + 7) during eight days. In the same way the seed of a fresh Acorn {Quercus Robur), after its weight had been reduced in the oven from 100 to 58 grains, added only 4 grains to its weight during an exposure to the air of eight days. A broken seed of Dioclea reflexa which had not completed the drying process lost 20 per cent, of its weight in the oven, and after five days was still 5 per cent, short of its original weight. On the other hand, normal resting seeds of the same plant, which lost 8 "6 per cent, of their weight in the oven, behaved like typical impermeable seeds during an exposure of five days to the air, increasing their original weight by 9 per cent. On the behaviour of seeds exposed to a temperature of 100° C. before they have commenced or before they have completed the drying process, the principle of Berthelot, discussed at length in Chapter VII, throws a flood of light. The water which they subsequently regain from the air is merely the water of hygroscopicity, which they would hold whether living or dead. This amounts on the average to only about 5 per cent, of the weight of the moist fresh seed that has not begun to dry. Most seeds lose quite 50 per cent, of their weight in the normal drying process, so that there would be a large proportion of the water which a fresh 134 STUDIES IN SEEDS AND FRUITS moist seed loses in the oven that it could never regain. Such considerations render necessary a review of the general behaviour of seeds after exposure to the oven test. It will be shown in the following chapters that the whole problem can be stated in quite a different manner if we introduce the principle of Berthelot as a resolving factor. The following table is intended to illustrate the contrast between impermeable and permeable seeds in their capacity of increasing their weight by absorbing water from the air when exposed in the broken condition, either with or without a previous exposure to a temperature of ioo° to iio° C. It will be noticed that in impermeable seeds both the coats and kernel possess this quality of adding considerably to their weight when exposed unheated to the air ; whilst with perme- able seeds both coats and kernel retain their weight, merely displaying the normal hygroscopic variation of 2 or 3 per cent., 98*5 to 101-5. Usually the inability to increase the weight, except in the ordinary course of hygroscopic variation, is indicated by 100. A previous exposure in the oven for I J or 2 hours does not materially affect the behaviour of the seed or its parts. In the case of impermeable seeds much the same excess weight is attained in about five days after the oven test, whilst in permeable seeds the original weight is more or less regained during the same period after the heating ; and in those cases where there is a marked failure to return to the original weight, it can be shown either that the seed had not completed its drying process before the experiment, or that the period of exposure to the air was too short, a week or two being in their case required (see p. 142). The behaviour of variable seeds is of course intermediate in character. ..." Similar samples " used in certain experi- ments were samples where each seed tested had been divided between the two samples, so that truly mixed samples were simultaneously experimented on. The same object was effected in the case of large seeds like those of Entada scandens (weighing some 500 grains) by employing the same seed for both samples. ADDITIONAL EVIDENCE ^3S General remarks. (For the explanation of ' ' similar samples " see the foregoing ex- planation of the table. ) Similar samples. Similar samples. Hygroscopic only, most seeds permeable. Change of weight of coats and kernel to- gether and of coats and kernel separately, after exposure to the air of about 5 days, the original weight in each case being taken as 100. After being first subjected to a tem- perature of 1 00° C. for i^ to 2 hours. S & p _•* p _-*oo jH _io p _W CO w-> ON Vo "i^ : : 'li-i : : : : b O«O«0m« • -0 • - • -o 6 j~v>o p p _a^ M m N i roos b M CO : : Vo : : : : « 0000>H«« - -0 • • • -o |lj '5° ^^ ^ ."^ ."* P .'^ .'^ M CO 2'222'^n?^ ■2'2'^22 c D 1 r^O^^\l^l00rIvOvn•vD : :vo "O ooO|-o««o :o : :© :o ° 00000-"0-0----0 ■i2 ,Q 1j I ."^ .".'*'» .=^ ."^ .l~^ N p N p y~, S c >- f^ i-^vij 00 M CO In vni l/i : : V U"'«3 ,oooi-o«i-oooo--oo t a 00 10 CT> tri a> CO N iri U-, o> ONoo t-^Tt-voTi-: :« : : ; :u-, 1 « CO N 00 M covo . . 1/^tv.u-icot^cr.r^: : N : : : :n il| M vp _rt- ON p p N . . vp .0 . 00 II 3 JkJJ-JJJhJhJJJJhJ i ij Adenanthera pavonina Dioclea reflexa Entada scandens Guilandina bonduc . , , bonducella . ,, glabra . Leucsena glauca Mucuna urens Strongylodon lucidum Ulex europseus Ipomoea dissecta ,, pes-caprse . Abrus precatorius . . ' 136 STUDIES IN SEEDS AND FRUITS O ^"1 S C - n •a 2g„q .^^^ ■ -'S^ ^J ON « O O vo 00 eX3 tv. w-i -rf- n t^oo p MTh: : : : : : : :'riVnr^: CT^ : : : :r^oooooo«ioooov 00 OnOON* on ..«•cy^OO^O^O^Oa^C^ 00 ' ' ' ' a\ ' as •'•*a^a^O^O^a^OO^ OOO^tl-r^ t-v »<-iCTnO vOoOwO 00- • O O O O O ON O O i=> r . t^ . 00 Tt-00 00 t^ N w^^o r^ -^ . . . . ^ • ;^ c^mThw CT^T^roT^-r'^u^T^- yr, : : : : 00 N . 000 . . "-1 ON vO r^Tj-N ^^0 OvO • r-. H w tv. « ON k4 hJ J hJ h-1 t-1 hJ J 1-1 llll,-.ll»l 3 0:1 - e ■^•^ ^c S g.S rt ■K-J .- g rt -2 ,, 8 ^ ^ -Sf3 HO O rt s O — CI « by)_5 j^ J^ S I ^ 'III rt s 2 = S) = III ,11 Jllsl^ lii^l 138 STUDIES IN SEEDS AND FRUITS We have seen that the ultra-dryness of the kernel of an impermeable seed is maintained through the impervious character of its coats. In further illustration of the great resistance which the seed is able to offer on account of its impermeable coverings to the injurious influence of external conditions, I will discuss the results of some experiments illustrating their behaviour under high temperatures. Although none of the seeds germinated after exposure to a temperature of 100° C, their behaviour under the test was very instructive ; and it would seem that in no better way can the contrast between permeable and impermeable seeds be shown than in their modes of responding to different stages of heat. It is of course natural that a seed-covering which is neither As ex- hygroscopic nor pervious should have this influence. Yet Entada ^ some very curious effects are produced when impermeable scandensand ggg^g such as those of Entada scandens and of Guilandina Guilandina ' bonducella. bonducella, are exposed in an oven to a temperature of 100° to 110° C. They are well brought out in the accompanying table, which contains the results of two simultaneous experi- ments on these seeds ; and in order to emphasise the peculiarity in the behaviour of the impermeable seed when heated with its coats intact, I have added the results for the same seed when subjected to a similar high temperature in the broken condition. It is shown in the columns of this table that in their coats these seeds behaved in a very similar fashion after an exposure for two hours to a temperature of 100° to 110° C. They lost respectively 2*7 and 1-9 per cent, of their weight, the subsequent efforts of both to supply the loss by absorbing water from the air having a very slight result. Both of them then doggedly resumed in their altered condition their previous im- permeability, making no hygroscopic response to the variations in atmospheric humidity and gaining back no weight on being immersed in water. If we contrast this with their behaviour when deprived of the protection of their coats (as indicated by the average results of several experiments on other seeds of the same species), we find that in the oven test they lost ADDITIONAL EVIDENCE 139 Results of Simultaneous Exposure for two Hours to a Temper- ature OF 100° to 1 10° C. of entire Seeds of Entada scandens and GUILANDINA BONDUCELLA, ILLUSTRATING THE PoWERS OF RESISTING Heat in their Shells or Coverings possessed by Impermeable Seeds, as Compared with the Behaviour of the Seeds of the same Species exposed to the same Temperature, but in a Broken Condition, and therefore no longer protected by their Shells. (The results are stated in percentages, the materials tested being 2 entire seeds of Entada scandens weighing in all 861-3 grains, and 3 entire seeds of Guilandina bonducella weighing io5-8 grains. The data given for the seeds in their broken condition represent the average of a number of experiments made in other connections during the course of this investigation. ) Changes in weight during and after the oven test. Condi- tion of seeds. Original weigh't. After 2 hours' expos- ure to 100° to 110° c. Weight after the oven test. After 2 to 7 days' im- mersion in water. ^ ■o ri C S 97'3 CI tJ- Entire in their 100 97 '6 97-6 97-6 97 '6 97-6 Entada coats. scandens In a broken condition. 100 88-6 99 'o 104-6 104-5 103*0 Entire in their 100 98-1 98-4 98-4 98-4 98-4 98-4 98-4 Guilandina coats. bonducella ^ In a broken 100 92*0 105 "o Ill's III 4 io6-o condition Entada scandens protected by its coats lost 2 7 per cent, and regained o"3. ,, unprotected ,, ,, 11-4 ,, ,, i6-o, Guilandina bonducella protected ,, ,, 1*9 ,, ,, 0-3. M unprotected ,, ,, 8*o ,, „ i9'S« 11-4 and 8'o per cent, respectively. This was more than regained after an exposure of some days to the air, there being a final excess over the original weight of 4*6 per cent, in the case of Entada scandens and of 11*5 per cent, in the case of Guilandina bonducella. Unfortunately, all the seeds failed to germinate, though I40 STUDIES IN SEEDS AND FRUITS examination proved that their kernels were in appearance sound. A more careful test of their germinative capacity might have produced different results. However, the interest of the experiment lies in its suggestiveness as to the mode in which an impermeable seed might be able to resist a temperature of 1 00° C. Provided that its water-contents are reduced to a minimum, it could withstand even a greater amount of heat and yet germinate. Becquerel, basing his opinion on the experiments of Dixon, considered 120° C. as the limit for desiccated seeds {Annales des Sciences Naturelles, v., 1907). With regard to my own results it may be added that the loss of weight in the oven is not so surprising as the subsequent resumption of impermeability. Behaviour somewhat similar, though under different conditions, is noticed below in the case of the impermeable seeds of another plant. Another illustration of the method by which impervious coats may enable seeds to resist high temperature was afforded by selected impermeable seeds of Canavalia obtusifolia under the strain of a great variety of thermal conditions, both with coats intact and with coats punctured, as shown in the diagram below. Exposed to alternating dry and damp conditions at ordinary temperatures and to extremes of dry and moist heat in the oven, the seed with coats intact varied only about 1-2 per cent, of its weight during a period of seven weeks, whilst the range of the weight of the seed with punctured coats under the same tests and during the same period was 7*5 per cent. But if we disregard the first loss of weight as concerned with influences preceding the experiment, then the variation under these highly contrasted conditions of the seed with coats intact amounted only to 0-7 per cent., which is probably not much greater than the variation that a quartz-pebble would exhibit under the same circumstances. Since the impermeable seeds of Canavalia obtusifolia normally increase their weight by 5 or 6 per cent, when exposed to the air bared of their coats, some proportion of the great variation displayed by the punctured seeds must be ascribed to their original ADDITIONAL EVIDENCE 141 Diagram illustrating the Different Behaviours under a great Variety of Conditions of Intact and Punctured Seeds of Canavalia obtusifolia of the Impermeable Type. (The experiment extended continuously from August 14 to October carried out in England.) and was Grains. Warm, dry air in cupboard, temp. 70°-75° F. Cool,d.imp air in room, temp. bo'-bs' F. Ten hours' dry heat temp. 85°- loy F. Cool, damp air in room, temp. 60-65" V- Three hours' moist heat in oven, temp. .50»F. Cool, damp room, temp. 60' F. One hour in temp. 2I2°F. Cool, damp room, temp. 60° F. Grains. Aug. 14 Aug. 19 Sept. 6 Sept. 14 Sept. 14 Sept. I 6 Sept. 1 8 Sept. 18 Sept. 28 Sept. 28 Oct I .09 109 108 108 107 i. •••■■•••. P nctured 107 106 106 105 Intact / 105 104 ^, ^■.'^' ^^ -— vL. \ .-■•. : 104 .03 '% /■•• .....•■ '. -r- — ''••' '* 'Intact '03 102 ; / V 102 101 101 100 •. / 100 99 99 98 98 In one case the seed -coats were punctured in several places. In the other case the coats were left intact. Nine or ten seeds were employed in each experiment. The first fall in weight was due to the seeds having been previously kept under less dry conditions, the small loss in the case of the seeds with coats intact being probably connected with tissue adherent to the scar. The loss in weight of the punctured seeds on September 18 after being exposed for three hours to moist air in the oven at a temperature of 150° F. is remarkable. The punctured seeds here displayed a caving-in or collapse of their coverings around each puncture. A vessel of water was placed in the oven during the test, and after cooling down the inside of the oven was covered with moisture. ultra-dryness. None of the seeds germinated at the close of the experiment, a negative fact which probably depends as 42 STUDIES IN SEEDS AND FRUITS much on the imperfection of the method as on the failure of the seeds, since, as above remarked, experiment has established the ability of certain seeds to withstand for some hours a temperature of ioo° to 120° C. However, the original purpose of this experiment has been served in demonstrating the protection an impervious covering affords against extreme thermal and hygrometric conditions. Very different is the behaviour of the permeable seed under the strain of a high temperature, a difference which its hygroscopicity would lead one to expect. Permeable seeds give up their moisture in the oven almost as readily when protected by their coverings as in the exposed condition. Since we have already seen in the chapter on type seeds that with such seeds the coats merely restrain but do not prevent the hygroscopic reaction of the kernel, the results given below are such as we should have looked for. Table showing the Influence of their Coverings on xfiE Behaviour OF Permeable Seeds when exposed for two Hours to a Temper- ature OF 100° TO 110° C. (Two samples of each kind of seed were employed, one with the seeds entire, the other with the seeds cut across so as to be deprived of the protection of their coverings. The seeds were eight or nine months old, the samples of the peas weighing loo grains and of the others 200 grains. The results are given in percentages. ) Condition. Original weight. Weight after the oven test. Loss in the oven. Time occupied in regaining original weight. Faba vulgaris J (Broad Bean) j P i s u m sativum J (Peas) \ Phaseolus multi- florus (Scarlet- runner) Entire Cut across Entire Cut across Entire Cut across 100 100 100 100 100 100 88-8 85-9 89-2 84-9 lO'O 13-9 II"2 141 IO-8 15-1 6 weeks 2 ., 2 .. 1 week 6 weeks 2 ,, By contrasting the results above tabulated with those given a few pages back for Entada scandens and Guilandina bonducella, where the seeds were exposed to the same test, we can frame a numerical estimate of the difference in the degree of protec- tion against high temperatures which the coverings offer in the ADDITIONAL EVIDENCE 143 case of permeable and impermeable seeds. Thus, assuming that the seed gave up approximately all its water when deprived of the protection of its coats, then the permeable seeds entire in their coverings may be regarded as having lost as much as from 70 to 80 per cent, of their water-contents, whilst the im- permeable seeds in the same entire condition and exposed to precisely the same test lost barely 25 per cent. Further contrasts in the behaviour of permeable and im- permeable seeds when exposed in their coats to a temperature of 100° to 110° C. for two hours are exhibited when we compare the results of their efforts to regain the lost water from the air. With impermeable seeds we have seen that very little effort is made to gain back their original weight, and that after the oven test they resume their impervious character, doggedly refusing to make any response to the hygrometric changes of the air and adding nothing to their weight when immersed in water. On the other hand, after the oven test permeable seeds slowly regained the water lost, but so very slowly that six weeks in the cases of Faba vulgaris and Phaseolus multiflorus and two weeks in the case of Pisum sativum were occupied in returning to their original weight. The influence of the coats is especially brought out in the case of the two types of seeds, if we regard their behaviour when the seed is broken up and is thus deprived of the protection of its coats. After the oven test the imper- meable seed gains back in a few days all the water lost and a considerable percentage (5 to 10 per cent.) more ; whilst the permeable seed returns to its original weight in a week or two, and is subject then to only the normal hygroscopic variations. The differences in the nature of the protection afforded by Summing-up the coats of permeable and impermeable seeds when exposed for ences in the" two hours to a temperature of 100° to 110° C. may be thus behaviour of briefly stated. In the impermeable seed the coats only allow and imper- one-fourth of the water-contents to escape, and by subsequently when ex-^ ^ resuming their imperviousness practically frustrate the seed's coats^to wJh effort to gain back the loss. In the permeable seed the in- tempera- hibitive influence of the coats is so slight that three-fourths of 144 STUDIES IN SEEDS AND FRUITS the water-contents are lost in the oven ; whilst the seed's attempt to regain from the air the water lost is retarded, but not prevented, the original weight being in time attained. The retardation in the process of re-absorption is, however, very marked, since in the Broad Bean and Scarlet-runner the period is extended from two to six weeks and in the Pea from one to two weeks. There are several points raised in the foregoing remarks which will be elucidated in the chapter on Hygroscopicity, notably, that concerned with the return of permeable seeds to their original weight after being heated. Only completely air- dry seeds would thus behave, since seeds that are still drying would fall considerably short of the weight they possessed before being placed in the oven. SUMMARY (i) This chapter contains the bulk of the data on which are based the distinctions in behaviour between permeable and impermeable seeds, which are described and illustrated by typical examples in Chapter IV. (2) The capacity of increasing their weight considerably by absorb- ing moisture from the air, when impermeable seeds are deprived of the protection of their coats, is confirmed by the results of ex- periments on the seeds of other plants. Other results also go to confirm the conclusion drawn for type seeds in Chapter IV, that the gain in weight is far more rapid when the kernel is completely bared than when the coats are merely punctured. The average results of all experiments on this absorptive capacity of impermeable seeds are tabulated ; and it is generally concluded that leguminous seeds of this type when deprived of the protection of their coverings as a rule increase their weight between 5 and 10 per cent, in a few days (p. 115). (3) The indications, though limited, go to show that impermeable seeds retain their ultra-dryness for a number of years (p. 119). (4) The data show that the capacity of absorbing water from the air when an impermeable seed is bared of its coats is greater in the tropics than in temperate climates, and that the gain in weight is maintained longer when it is acquired slowly, as in filed seeds, than it is when acquired rapidly, as in bared kernels (p. 120). ADDITIONAL EVIDENCE 145 (5) The results for variable seeds (permeable and impermeable seeds in the same plant) of baring the kernel, or exposing it by cutting it in halves in its coats, are also tabulated ; and it is shown respecting their relations with the moisture of the air (a) that the permeable seeds behave hygroscopically, like ordinary seeds of the type ; (/;) that imper- meable seeds behave also like seeds of their type and add considerably to their weight ; (c) that mixed samples of the two types of seeds display intermediate qualities (p. 122). (6) Additional data are given concerning the behaviour of bared permeable seeds that have completed their drying in air ; and it is shown, as in Chapter IV, that the main result is to give a rather freer play to the hygroscopic reaction, the average weight remaining about the same (p. 124). (7) Further evidence is then supplied of the association of the ultra- dryness of impermeable seeds with a low water-percentage, and in the first place the necessity of restricting the inquiry to seeds of the same order is pointed out. (8) Thus with leguminous seeds we obtain the following general results for seeds broken up or cut open and exposed to the air : — Impermeable seeds holding 7-5 per cent, of water add 9*2 per cent, to their weight. Impermeable seeds holding 1 1-7 per cent, of water add 5-0 per cent, to their weight. Permeable seeds holding 15-0 per cent of water add o-o, behaving hygroscopically. (9) The indications of this rough estimate are confirmed by the results of experiments on seeds where both the permeable and imper- meable types of seeds are produced by the same plant. (10) More evidence is adduced to show that in impermeable seeds the seed-coats possess the same quality of ultra-dryness as the kernel. (11) Additional data are given in support of the conclusion that the capacity possessed by impermeable seeds of considerably adding to their weight when exposed to the air in a broken state is not afiected by first exposing the materials to a temperature of 100° C, whether in the case of the coats or of the kernel. (12) Permeable seeds in this respect present a great contrast to impermeable seeds, since in a week or two they gain back from the air only the water lost in the oven and assume a stable weight subject merely to ordinary hygroscopic variation. (13) In the case of both permeable and impermeable seeds it is necessary, when comparing their absorptive capacities after heating, to employ only seeds that have completed the drying and shrinking process, since seeds of either type, when incompletely dried, fail to 10 146 STUDIES IN SEEDS AND FRUITS return to their original weight. On this point the principle of Berthelot, discussed in Chapter VII, throws a flood of light. (14) The contrast between impermeable and permeable seeds is further illustrated by their different modes of responding to high temperatures under the protection of their coats. Whilst impermeable seeds, when exposed for two hours to a temperature of 100° to 110° C, lose only about 25 per cent, of their water-contents, permeable seeds that have completed their drying in air lose under these conditions in the oven as much as 75 per cent. (p. 138). (15) But the contrast is extended when we compare the results of their efforts to regain from the air the water lost in the oven. The impermeable seed makes a very slight effort in this direction ; and whilst its coats quickly resume their impervious character, the seed doggedly refuses to make any response to the variations in the atmospheric humidity and adds nothing to its weight when placed in water. On the other hand, after the oven test the permeable seed slowly regains from the air the water lost, but so gradually that weeks are taken up in the process, the original weight being ultimately attained subject to the ordinary hygroscopic reaction. The return of the air-dried permeable seed to its original weight after the heat test is a point of importance, since seeds that have not completed their drying in air fail to reach their original weight, a critical distinction discussed in detail in the chapter on Hygroscopicity (p. 139). (16) The great resistance which a seed protected by impermeable coats is able to offer to extremes of moist and dry heat ranging up to 100° C, and to alternating dry and damp conditions, is shown in the behaviour of impermeable seeds of Canavalia obtusifolia. Kept under the strain of a great variety of extreme conditions for seven weeks, seeds with coats intact did not vary i per cent, in weight, whilst those with punctured coats varied as much as 7-5 per cent. (p. 140). CHAPTER VII HYGROSCOPICITY Hygroscopicity in a seed may be defined as the variation of Definition of its water-contents in response to the changes in the hygrometric scScity. state of the atmosphere, such a variation being due to its property of readily imbibing moisture from the air and as readily parting with it. This interesting quality, which is characteristic of all vegetable substances and of much besides, has in recent times been studied by numerous investigators, including amongst others Jodin (1897), Berthelot (1903), Leo Errera (1906), Becquerel (1907), and Demoussy (1907), the years in- dicating the date of publication of the author's paper consulted. The most comprehensive treatment of the subject is to be found in the memoir of Leo Errera entitled " Sur I'Hygrosco- picite comme cause de Faction physiologique a distance " {Recueil de TInstitut Botanique Leo Errera^ Universite de Bruxelles^ tome vi., 1906). Assisted by the previous researches of Classifica- Warburg and Ihmori, he was able to direct attention to certain dXrent^ principles involved and to show that hygroscopicity in its Errera'^ ^^° widest sense has a far more extended significance than is generally imagined, as is suflSciently brought out in his classification of the various forms of this property, both physical and chemical, which I have reproduced in Note 9 of the Appendix. The matter as here dealt with appeals mainly to the physicist and the chemist ; but it is essential, before studying hygroscopicity as affecting plants, to bear in mind the broad treatment of the subject which gives so much importance 147 148 STUDIES IN SEEDS AND FRUITS to this paper. Hygroscopicity is there exhibited in the most comprehensive sense, as displayed (d) in the condensation of the water-vapour of the air on the cold surface of a glass ; (b) in the capillarity of hair, wool, cotton, wood shavings, etc. ; {c) in the imbibition of water from the air by gelatine ; (d) in the deliquescence of common salt ; (/) in the absorption of water from the air by concentrated sulphuric acid ; and (/) in the behaviour of quicklime. Becquerel, in applying this classification to seeds, suggests two kinds of hygroscopicity : (i) physical, where condensation is affected by the cold, smooth sides of the seed or by the walls of very fine capillary pores ; (2) chemical, when induced by the afl&nity of certain substances for water {Annales des Sciences Naturelles Botanique^ tome v., 1 907). Coming to the display of this quality by vegetable materials in general, I will, before handling my own observations, take my cue from the researches of Jodin on peas, and will then look to the principle laid down by Berthelot for guidance in the search after the significance of hygroscopicity in plant- tissues. But it is necessary to preface my remarks by point- ing out that the hygroscopic reaction understood by these investigators, and always intended in these pages, is the response of the permeable seed (by absorbing or yielding up water-vapour) to the varying hygrometric condition of the air, a never ceasing " give and take " process by which the equilibrium between the seed and the air is maintained. Jodin approached the subject from the biological and Berthelot from the physical side, and both arrived at the same conclusion, that we are concerned with a quality that is inde- pendent of life. Jodin, in his paper published in the Annales Agronomiques for October 1897, tells us that living and dead peas (those recently grown and those that had long lost their germinative capacity) exhibited much the same hygrometric variation in the course of a year's exposure to ordinary air- conditions. Stated as a percentage of the average weight of the air-dry resting seed, his results give a variation for the live peas of 8 to 23 per cent., and for the dead peas of 11 to 21 HYGROSCOPICITY 149 per cent. It was the work of Jodin that led Becquerel in the paper before quoted to make the critical distinction in a seed's water-contents between the water of hygroscopicity and the water concerned in the latent life of the embryo. Whilst several years have passed since Jodin directed his attention to the ordinary hygrometric variations experienced by peas, Berthelot in more recent times has opened up the whole subject of the hydratation of vegetable matter, and in so doing has thrown an important light on the nature of hygroscopicity in plants (" Recherches sur la desiccation des plantes et des tissues vegetaux ; conditions d'equilibre et de reversibilite," Annales de Chimie et de Physique^ April 1903). He shows that the peculiar property possessed by air-dried vegetable matter of regaining from the air the water it has been made to lose by heat and other artificial means is a function of the hygrometric condition of the atmosphere. It is not easy for me to state Berthelot's principle tersely, and Berthelot's accordingly I have above followed Maquenne in his reference feverSbUity. to this subject in Comptes rendus, October 1905. Nor is it easy to grasp its full significance at first, since, as is natural in such abstruse inquiries, much will seem pointless that does not cross the boundary of one's own researches. To me perhaps it is not so hard, since the principle has cast a flood of light upon the results of my studies of permeable and impermeable seeds. A plant, says Berthelot, does not dry entirely in air like porcelain or metals (see Note 21 of Appendix). It retains after being thus dried a certain amount of water, which varies in response to the changes in the hygrometricity of the atmo- sphere. When this water has been driven ofi^ by exposure to a its applica- temperature of 110° C, it is gained back little by little from tissuls uIT"*' the air up to a limit practically the same as that reached when general, the plant was dried in air. In a word, the water which the air-dried material loses in the oven is regained in the air. This is Berthelot's principle of reversibility, and it is characterised by him as essentially a physico-chemical process independent of life. It applies equally to the plant that has I50 STUDIES IN SEEDS AND FRUITS been dried in air, to the plant that has died spontaneously, and to the plant that has been subjected to almost absolute desicca- tion by heat and other means, all ultimately reaching the same condition of equilibrium with regard to the atmosphere. We will illustrate Berthelot's principle by combining in one statement the results of his different experiments. Portions of a living grass, a species of Festuca^ weighing, we will suppose, lOO grammes, are allowed to dry in the air of an ordinary room for some days, until they acquire a stable weight affected only by the usual small hygroscopic fluctuations. Their weight is thus reduced by loss of water to about 66-^ grammes. They are then exposed to a temperature of iio° C. for some hours, with the result that the weight is further reduced to 6i grammes. After being laid aside for some days, the material, by the absorption of moisture from the air, returns to the original air-dried weight of about 66' c, grammes, and there remains, varying slightly with the changing humidity of the atmosphere. It is in the ^^ grammes which the air-dried material lost in the oven and regained when subsequently exposed to the air that the secret of the hygroscopicity of plants lies. Table illustrating Berthelot's "Principle of Reversibility." (The Results of Different Experiments are here combined, the Materials employed being Portions of a Species of Festuca.) Weight in grammes under different conditions. Fresh weight. Dried in air of an ordi- nary room. Died spon- taneously. After exposure to temperature of iio°C. After a subsequent exposure of some days in the air of an ordinary room. lOO lOO lOO 66-5 66-s 61 6i 6i 66-5 66-5 66-5 Note. — The data representing the effect of exposing fresh and air-dried material to a temperature of iio° C. belong to the same experiment, the rest of the data being supplied from the indications of other experiments. HYGROSCOPICITY The feature in this table which will prove the greatest significance to us, in respect to the of impermeable seeds when exposed to the air, is which shows that the water regained from the air, 151 to be of behaviour that after fresh plant-tissues have been exposed to a temperature of 100° to 110° C, is the water of the dead plant and of the plant dried in the air, and is therefore independent of vitality. In the above table I have pieced together the indications of different experiments in order to emphasise certain points in the behaviour of plants when exposed to natural and artificial desiccation. After reading Berthelot's paper I experimented on some fresh leaves of the Hazel {Corylus Avellana)^ with the results below given. His principle is well illustrated there, and we can see at a glance that the water which the fresh leaves gained back from the air after being exposed to a temperature of 105° C. is the water which they would have contained as ordinary air-dried leaves. For about three years I have been The testi- mony of Hazel leaves. Experiments by the Author on fresh Hazel (Corylus Avellana) Leaves in illustration of Berthelot's " Principle of Reversibility." Results of the drying and heating tests. The water of hygroscopicity. Fresh weight. After dry- ing in air of room. After ex- posure to temperature of 105° C. in oven. Three or four days after the oven test. Lost in oven after drying in air. Gain in air after the oven test. Stated as a per- centage of the dry weight. A B C D 100 100 100 100 31-3 32-8 28-0 29-4 31 -o 29-6 31 '4 33-0 34*5 32-9 3 '3 3*4 3 "4 3-6 3'5 3"3 I2*I 12-2 ii'3 ii-i Hundred-grain samples of the fresh material were used. In the case of experiments A and B, the sample was first dried in air for about five days, when it reached a stable weight. It was then subjected to a temperature of 105° C. for ih hours, and afterwards left exposed to the air of the room for three or four days, when it ceased to gain weight. In the case of experiments C and D the fresh materials were placed in the oven without previous drying in air. 15- STUDIES IN SEEDS AND FRUITS putting this question to myself as to the significance of the gain in weight of plant-tissues, and more particularly of seeds after being exposed to desiccation in the oven, never dreaming that such a simple experiment would supply the answer. To have been contented with attributing it to hygroscopicity would have explained little. As a disconnected fact it appeared without interest and without meaning. What was required was the estabhshment of a relation between this property and some other attribute of plant-tissues ; and this 1 ultimately found in the principle of Berthelot. From this standpoint the water-contents of plants could be divided into two parts, the water of hygroscopicity and the water of vitality. The first, being independent of life, is equally characteristic of the plant living and the plant dead. It is the residuum left in the air-dried material ; and it is the water that the same material loses in the oven and regains in the air. The second is the water that distinguishes the plant as a living organism. Its quantity is regulated only by the needs of that organism. Unlike the water of hygro- scopicity, it does not directly respond in its variations to the hygrometricity of the air. On the other hand, hygro- scopicity being a non-vital process represents the response of the non-vital part of a plant's water-contents to the varying humidity of the atmosphere. We can thus understand how the residuum of water in the air-dried plant is the water that represents a function of the hygrometric state of the air. This mode of differentiating the water-contents of plant- tissues is of practical importance. There are many ways of stating the proportions, and I have spent much time in trying to harmonise them with the results of my observations. Finally, it became evident that of the numerous methods of describing the " hydratation " of plants there were none so simple and none so true as that implied in the principle of Berthelot. There is the water that the organised tissues contain, whether living or dead, the water of hygroscopicity ; HYGROSCOPICITY 153 and there is the water which they hold only as living structures, the water of vitality. The chemist, when producing by synthesis organic vegetable matter, would allow the atmosphere to supply the water of hygroscopicity, whilst he himself in his creative role would have to supply the water of vitality. This in a sense is very much what a baker does when he adds water to his flour in making bread. As shown below, bread behaves like fresh plant-tissues when dried in air and when desiccated by heat. The water it regains from the air after heating is the water originally existing in the flour as supplied by the miller, and the water it does not gain back is what the baker put into it. Experiments by the Author on ioo-grain Samples of Bread in Illustration of Berthelot's " Principle of Reversibility." Treatment of samples. Results of the drying and heating tests. The water of hygroscopicity. Original weight. After drying in air of room. After exposure to temperature of 105° C. in oven for i| hours. Three or four days after the oven test. Lost in oven after dry- ing in air. Gain in air after oven test. Stated as a per- centage of the dry weight. Dried in air and then placed in oven Placed at once in oven 100 100 7o'3 6i-2 6o*5 70-4 70*0 9"! 9-2 9*5 15-0 '57 Divided into small squares, the bread occupied about six days in reaching a stable weight in air before being placed in the oven, where it was kept for ij hours. Three to four days were passed in acquiring a stable weight after the oven test. When we come to apply the test of experiment to this The same principle as it affects seeds, we get the same indications. The mistrated in simple experiment of drying a fresh seed first under ordinary air-conditions, then in the oven at 100° to 110° C, and after- wards allowing it to remain exposed to the air for a few days until it assumes a stable weight, supplies results that make the the be- haviour of a seed. 154 STUDIES IN SEEDS AND FRUITS statement of a seed's water-contents as easy as it was difficult. All my results on the varying water-percentages of seeds, and on the capacity of regaining from the air the water lost in the oven, arrange themselves in an intelligible system in the light of the principle below typified in the behaviour of the full- grown, moist pre-resting seeds of Phaseolus multiflorus. The hydratation of these seeds in this condition of active vitality may be thus stated : — Water of vitality . Water of hygroscopicity Solids . . . . 64-8 5-2 30-0 lOO-O The Principle of Berthelot illustrated by the Behaviour of THE Full-grown Unripe or Pre-resting Seeds of Phaseolus multiflorus. (Three seeds weighing 140 grains were experi- mented on by the author ; the results are given as percentages.) Results of the drying and heating tests. The water of hygroscopicity. Unripe or pre-rest- ing seed from the green pod. After drying in air of room. After exposure to a temperature of 105° to no° C. in oven for 2 hours. Six or seven days after the oven test. Lost in the oven after dry- ing in air. Gain in air after the oven test. Stated as a per- centage of the dry weight. 100 35"3 30*0 35 "o 5*3 5-0 17-3 Important as the principle of Berthelot is in the differen- tiation of the water-contents of plant-tissues generally, its application is still more interesting in its results in the case of the hydratation of seeds. It not only enables us to recognise in clear outlines the nature of the contrast between permeable and impermeable seeds ; but this implication supplies quite a novel way of viewing the problem of the latent life of seeds. If the implication is valid, its influence on our views should be revolutionary. HYGROSCOPICITY 155 In the first place, as regards the contrast between (a) That air- permeable and impermeable seeds, it is evident that an seed?^both air-dry permeable resting seed, which has assumed a stable permeable . , 1 . , 3 . , . . . and imper- weight, subject only to ordinary hygroscopic variations, meable, con- contains only the water of hygroscopicity, and that the wate°"of water of vitality disappeared in the drying process. It g^^T'it also becomes apparent that the impermeable seed con- tains only the water of hygroscopicity, but in a diminished amount, so that when deprived of the protection of its impermeable coverings it at once begins to supply the deficit by abstracting moisture from the air until a stable weight is reached. The implication of course is that resting seeds completely air-dried, whether permeable or impermeable, possess only the water which is independent of vitality. If Berthelot's principle is true and the implication is valid, there is in the typical resting seed no water that is associated with any vital function. (I am not here speaking of water locked up in chemical combination in the seed's tissues, since that may be a property of both living and dead matter.) Should the seed exposed to a temperature of 100° C. in the oven yield up (6) that in more water than it subsequently regains from the air, the fngleedr^" inference is that it had not completed its drying; in air and there is no Ml -1 f 1 f • 1- --r-i • • , water associ- still contained some 01 the water or vitality. 1 his residuum ated with of the water of vitality left in the deficiently air-dried function, seed has nothing to do with the life of a resting seed, but merely represents the remains of the water of the large, soft pre-resting seed of the moist green fruit, a seed that would have proceeded with its growth and with its development into a young plant without any pause, if the resting period had not been imposed on it through external influences. The resting seed needs no water to prolong its life, the presence of water being more likely to curtail its existence than to endow it with longevity. Indeed, there would seem to be more than fancy in the speculation of M. Demoussy that a perfectly dry seed kept protected from 156 STUDIES IN SEEDS AND FRUITS the air has the potentialities of immortahty (Comptes rendus^ December 1907). Yet it cannot be doubted that the view expressed so clearly by Becquerel {Ann. des Sciences Nat. Botan.^ v., 1907), that it is absolutely necessary to distinguish in a seed between the hygrometric water that can vary and the water enclosed in the cellules of the embryo and albumen which is invariable, at present holds the field. We must distinguish, he says, between the water of hygrometricity and the water that plays a part in the phenomena of the latent life of the seed. But if Berthelot's principle is correct, the only water that a seed retains after it has completed its drying in air is the water of hygrometricity. According to the implications of this principle, the true resting seed, as already observed, needs no water for the support of its latent life ; and the latent life itself becomes almost a figure of speech. Several years ago Schroder ascertained that grains of three cereals (species of Hordeum and Triticum) retained their germinative capacity, notwithstanding that after undergoing a process of artificial desiccation for nearly three months their water-contents had been reduced respectively to 0*5, I'O, and 2-0 per cent. {Untersuch. Botan. Inst, zu Tubingen^ 1886). It would be quite as legitimate to infer from this experiment that resting seeds can dispense with water altogether as to assume that Schroder in his experiment reached the minimum com- patible with the preservation of the germinative powers. There is an obsession in the human mind respecting water and active life that makes it difficult to assimilate the notion that a resting seed could possibly do without it. The standpoint adopted in this chapter is that the occurrence of water in a properly air-dried seed is accidental as far as it is concerned with the retention of the germinative capacity. It could have no concern for the student of the latent life of seeds, since its quantity would be the same whether the seed be living or dead. HYGROSCOPICITY 157 Later investigations on the desiccation of seeds have been numerous ; but many of them are summed up by Becquerel Becquerel on in his paper on the latent life of seeds (^Ann. Sci. JVat.y 1907). seeS^^^ There are certain seeds, he points out, which are able to resist the most powerful desiccating agencies at our command ; and very significant is his conclusion, after a review of the liquid- air results, that it is the seed where the water and " gaz " have been reduced to the narrowest possible limits by the most active desiccators of the laboratory that best withstands these tests. There seems no necessity to assign a function to the extremely minute amount of water that mi^ht survive the desiccating process. On the contrary, it might be urged that its water is the Water is the greatest foe to a seed's longevity. What, we may ask, is the fo'^l^eed's^ real biological significance of the hygroscopicity of seeds, as ^°"ge^i^y- far as their longevity is concerned ? It is their hygroscopicity that limits the life of permeable seeds ; or, in other words, the constant reaction between the seed and its atmospheric The hygro- environment places a term to its existence. Not the least ticmiimfts^" interesting of the conclusions drawn by Jodin from his ^^^dg^^^hilst observations on peas lies in this direction, and we may apply its absence ui J • 1 T-u '• J u • favours their It to permeable seeds m general. The contmued hygroscopic longevity, reaction, he points out, would in the course of time bring about molecular changes in the seed, terminating in its loss of germinative power and death. Thus we can perceive by implication how the impermeable seed, by not responding to the changes in its atmospheric surroundings, is secured against one great risk to its longevity. Here, again, we perceive that the long life of the seed presents itself as an affair of the coats rather than one concerned with the dormancy of the protoplasm of the embryo. It is the free play of the hygroscopic reaction that curtails the life of a permeable seed. It is the absence of this reaction that gives long life to the impermeable seed. After this long digression on the significance of hygro- scopicity in seeds, I come to my own studies in this connection. 158 STUDIES IN SEEDS AND FRUITS My observations on the hygroscopic behaviour of seeds are naturally concerned only with permeable seeds, since the impermeable seed in the usual sense of the word is non-hygro- scopic, and when employed in experiments merely serves to give contrast to the results. Hygroscopic seeds weighed daily during a fortnight of changeable weather usually vary ij or 2 per cent, of their average weight, an amount, however, which is only about half of what may be regarded as the ordinary extreme of the hygroscopic range, which, as ascertained by a method to be subsequently described, is usually 3 or 4 per cent. This reaction figures as a possible disturbing cause in all experiments on seeds where the balance is employed. Seeds as a rule continue to lose weight by drying during a period varying from a few weeks to two or three months after being gathered from the plant. If we extend the experiment over a year or more, employing only seeds that have completed the drying process and have acquired a stable weight, we find a response to the varying humidity of the air not only in the minor changes during short intervals, as between day and night, and in the greater changes from week to week, but also between the different seasons. It may here be remarked that the seasonal changes in weight were well exemplified in my experiments on the seed of Msculus Hippocastanum (Horse-chestnut). Seeds that had been kept for three years were usually i or i J per cent, heavier in the winter than in the summer. Reference will now be made to the changes of weight which many seeds undergo in transference between regions where different hygrometric regimes prevail, as between tropical and temperate countries. I made some observa- tions in this direction in the case of seeds taken from England to Jamaica in November 1907, seeds which had been gathered fresh in Jamaica in the spring of the same year. Permeable, impermeable, and variable seeds were here represented. HYGROSCOPICITY 159 Table illustrating the Changes in Weight experienced by Tropical Seeds when transferred from England to Jamaica, AND from Jamaica back to England. (Seeds obtained fresh in Jamaica in spring of 1907.) Character. Weight in grains. Change stated as a percentage of the total weight. Eng- land, Oct. 1907. Jamaica, Jan. 1908. 169-60 102-80 52-15 105-70 102-50 306-60 809-65 104*80 Eng- land, April 1908. Rise. Fall. Range. Canavalia ensiformis (7 seeds ; Leguminosse) Achras Sapota (10 seeds ; Sapotacese ; fruit baccate) Chrysophyllum Cainito (4 seeds ; Sapotacete ; fruit baccate) Abrus precatorius (69 seeds ; LeguminosK) Canavalia obtusifolia (8 seeds ; Leguminosa;) Guilandina bonducella (7 seeds ; Legu- minosse) Entada scandens (3 seeds ; Leguminosa') Adenanthera pavonina (22 seeds ; Legu- minosse) Permeable Permeable Permeable Variable Variable Imperme- able Imperme- able Imperme- able 167-85 .0.55 51-70 105-40 102-40 306-60 809-60 104*80 163-40 99-80 50-30 105-00 io2*55 306*60 809*45 104*85 1-04 1-23 0-87 0*28 0*15 O'OO coo 0-05 3-66 2-91 3'55 0-66 0-00 0-02 37 2*9 3-6 o*7 0-2 0-00 o-oo 0-00 The seeds had been five months in England when weighed in October 1907 ; two months in Jamaica in January 1908, when weighed in that island ; and one month in England when weighed in April 1908, All the seeds retained their germinative powers after the experiment, with the exception of those of Achras and ChrysophylluTn, which were sound, but did not germinate. It may be concluded from the data in this table that the following were the results of the transportation of these seeds from the temperate zone to the tropics and back. The range of the variation in weight of the permeable seeds was 3 to 4 per cent., and that of the variable seeds considerably under i per cent. ; whilst the range of the impermeable seeds may be regarded as " nil," or, if present, as probably instrumental and therefore negligible. In the case of both the variable kinds i6o STUDIES IN SEEDS AND FRUITS of seeds the proportion of permeable seeds was probably very- small, that is to say, under lo per cent.; and doubtless the change of weight was experienced by a very few seeds. The rise in weight of the three permeable kinds of seeds in Jamaica and the fall after returning to England are characteristic ; but the difference between them is probably due to the varying hygrometric conditions of the air when the two sets of obser- vations in England were made. On the whole, however, it may be considered that the changes in weight experienced by these permeable seeds between warm and cool latitudes are within the limits of the hygroscopic range defined a few pages back. In all experiments for determining the limits of hygroscopic variation it is requisite that the seeds should have completed the drying process and that they should be, as far as their water-contents are concerned, in a state of equilibrium with the air. The method at first employed in investigating the hygroscopic behaviour of seeds was, as already indicated, to weigh them daily for ten or fourteen days when the weather was changeable. The variation was then stated as a percentage of the total weight of the seed, and this was termed the " hygro- scopic range." In such experiments four or five kinds of seeds, of both the permeable and impermeable types, were experi- mented on at the same time. The results of one of these experiments in Jamaica are given below in the form of a diagram ; and in order to obtain a more graphic effect, they have all been computed for looo grains of each kind of seed, whilst the prevailing weather conditions as regards rain have been roughly indicated by black and white squares. The following seeds were employed : — 112 seeds of Anona palustris, weighing 74 „ Anona muricata (Sour Sop), „ 85 „ Citrus decumana (Shaddock), „ 212 „ Adenanthera pavonina „ 3 „ Entada scandens „ The first three are permeable seeds, the ranges for the two species of Anona being 1-3 and \"i per cent., whilst that for 440-8 grai 470-8 •)•> 3347 „ lOOO-O ji 1087-3 5? HYGROSCOPICITY i6i the Shaddock was i'6 per cent. But these do not represent the maximum hygroscopic ranges. The Shaddock seeds in another experiment in England gave a range of 2 per cent. ; and probably the swing of the range, including ordinary extremes, would amount to nearly 3 per cent, for all the permeable seeds here experimented on. The last two are typically impermeable, and the small variation exhibited is probably instrumental. Diagram contrasting the Behaviour of Permeable and Imperme- able Seeds as respects their Variation in Weight during Ten Days of Changeable Weather in Jamaica, the Rain BEING indicated BY BlACK. (For further explanation see the preceding meable and the last two are impermeable.) remarks. The three first-named are per- Range in grains. 1 >.„,. 1 Fcbr. 1 ^^H 1 1 ^3 24 25 26 27 28 29 30 3' I Anona pjlustris Anona muricata . Citrus decumana . Entada scandcns . Adcnanlhera pavonina . 997-3- 1010-4= ij-l 997-4- 1009-3= 1 1-9 1000-0-1016-1 = 16-1 1000-0—1000-1= o-i 1000-0- 1000"2= 0-2 ■■••••... .••■■ ■••••.. --- -r.T- ..... -... -. In course of time, however, I discovered that although the method described and illustrated in the previous pages The method exhibited the ordinary hygroscopic response of the permeable adopted of seed to the usual weather changes within a limited period, f^e^h™?©"^ it did not give me the whole range of the variation, such as scopic range. 1 62 STUDIES IN SEEDS AND FRUITS one would look for if the experiment was continued for a year. A chance observation led me to believe that the whole range was quite twice as much as that indicated in the fore- going experiments ; and I soon found, on transferring the seeds from a cool, damp room without artificial heat to a dry- cupboard, where the temperature (owing to the vicinity of hot- water pipes) was from io° to 15° F, warmer, that the range of the hygroscopicity was much increased. In such circumstances the seeds of Anona^ Shaddock, Canavalia ensiformis^ etc., which, as first tested, exhibited a variation in weight of I to i|- per cent., now showed an increased range of 2 or 3 per cent. Accordingly, I subjected the seeds to this new proof, and the results are those given in the following pages. The seeds were transferred from the cool room to the warm cupboard for two days and back again to the room, the mean of the two results being taken. Typical impermeable seeds used as checks in many experiments displayed no change in weight, except of a trifling nature. Results of the Author's Observations on the Hygroscopicity of Seeds. I. Impermeable Seeds. (Hygroscopicity " nil") A typical impermeable seed exhibits no change in weight, except such as can be attributed to instrumental error, or to a little loose tissue adhering to the scar or to the raphe. For example, a seed of Guilandina bonducella 40 grains in weight, and a seed of Entada scandens weighing 400 grains, would show the same variation during a month of O'l grain, which is so small that it may be safely attributed to other causes than to normal hygroscopicity. The seeds actually tested by me include the following : — Adenanthera pavonina, Dioclea reflexa^ Entada scandens^ Guilandina bonducella^ Leuc^na glauca^ Mucuna urens, etc. But of course all the seeds mentioned in the Impermeable Group in Chapter V would, when typical, display no hygroscopic re- H YGROSCOPICITY 1 6^ action, excluding seeds like those of Ipomcea pes-caprc€^ where the covering of hairs would give hygroscopicity, though the seed bared of hairs is non - hygroscopic. The influence of hairs is dealt with a few pages further on. Since impermeable seeds are not infrequent in nature, as is shown in Chapter III, it follows that a large number of seeds fail to make any response to the weather changes, and are therefore non-hygroscopic. Here would belong the seeds of many of the Australian Acacias and a number of the i8o " macrobiotic " or long-lived seeds included in Professor Ewart's list {Proceedings^ Royal Society of Victoria^ 1908), though a large proportion also would belong to the Variable Group, where plants possess both permeable and impermeable seeds. As illustrating the behaviour of impermeable seeds in the balance, there are appended the actual results obtained in the case of those of four species during periods of from ten to sixty days. It will be noticed, as before remarked, that the same slight variation of from O'l to 0*2 grain is displayed by large and small samples, being evidently in great part instrumental. We are here only concerned with the absence of the ordinary hygroscopic reaction in the course of a few weeks. The extent to which this behaviour is persistent will be discussed in Chapter X. [Table 164 STUDIES IN SEEDS AND FRUITS Some Results illustrating the Absence of a true Hygroscopic Reaction in Impermeable Seeds, the Slight Variation being Instrumental. (Daily observations made in the short experiments only.) Number Length of Variation of weight Locality of of seeds. experiment. in grains. experiment. Guilandina bonducella . 30 18 days 990-43- 990-60 or 0*17 England. 60 „ 35-03- 35-10 or 0-07 England. Quartz pebble 60 ,, 38-40- 38-47 or 0-07 England. Dioclea reflexa II 10 ,, 1042-00-1042-2 or 0-20 Grenada. Entada scandens . 3 10 ,, 1087-4 ~^°^7'S or o'lo Jamaica. Leucsena glauca . 332 10 ,, 250-0 - 249-9 o"^ °''° Grenada. II. Permeable Seeds. (Hygroscopicity 2 to 5 per cent.) (The hygroscopic range of a seed is the amount of variation stated as a percentage of the average weight. Thus, if a seed varied between 98 and 102 grains, the range would be 4 per cent. The methods of obtaining these results are described a page or two back.) ^sculus Hippocastanum (Horse-chestnut) Allium ursinum .... Anona muricata (Sour Sop) . ,, palustris .... ,, reticulata (Custard Apple) . ,, squamosa (Sweet Sop) Bignonia (species of) . Canavalia ensiformis . Chrysophyllum Cainito (Star Apple) Citrus decumana (Shaddock) Datura Stramonium Dolichos Lablab .... Faba vulgaris (Broad Bean) . Hura crepitans (Sandbox Tree) . Iris foetidissima .... Luffa acutangula (Loofah) . Phaseolus multiflorus (Scarlet-runner) Pisum sativum (Peas) wrinkled ,, ,, smooth Primula veris (Primrose) Ravenala madagascariensis (Travellers' Ricinus communis (Castor Oil) Sapota Achras (Sapodilla) . Scilla nutans (Bluebell) Swietenia Mahogani (Mahogany) . Tamus communis Palm Hygroscopic range. 4 '5 per cent. 4"o ,, 3 3 3"S 2-4 3"o 4-0 2-2 4-0 2-2 5*0 6-0 4"5 2-0 I '5 1*9 4-0 2-0 4-0 Average weight of a seed in grains [30-0 grains. 4*5 HYGROSCOPICITY 165 III. Variable Seeds. (Hygroscopicity variable.) (Variable seeds are those where some are permeable and others impermeable, the pro- portions being very inconstant, so that the hygroscopic range for different seed-samples would vary much, and could only be characterised as intermediate between that for im- permeable seeds (nil) and that for permeable seeds (2 to 4 per cent.). Abrus precatorius Acacia Farnesiana Albizzia Lebbek ...... Aquilegia (species of) . Bauhinia (species of) Csesalpinia Sappan (A), mostly permeable . ,, (B), mostly impermeable Csesalpinia sepiaria . . . . . Canavalia obtusifolia ..... Canna indica ...... Cassia fistula ,, marginata ...... Entada polystachya (A), mostly permeable . ,, (B) ,, impermeable Enterolobium cyclocarpum .... Erythrina corallodendron .... ,, indica ...... Ipomoea tuberosa ...... Poinciana regia ...... Tamarindus indicus Thespesia populnea ..... Hygroscopic range (variable in different samples). o'5 per cent. I'S .. 2'0 ,, 07 „ o-S .. 3"o ., o'6 ,, 2'0 ,, o'3 .. 0-3 „ 0-6 ,, 07 ,, 2"0 ,, 0'2 ,, o'6 o'S .. I "o ,, Average weight of a seed m grains. I '5 grains. 2*0 2-3 0-33 4-0 lo-o 10 -o 4*5 I2'0 27 4-0 lO'O 6-6 5"o 17-0 3 "2 13-0 25-0 lo-o 20 -o 3-0 conclusions drawn from From the results of the observations on the hygroscopicity of a number of seeds which are given above it may be inferred — (i) That permeable seeds under ordinary weather conditions General vary to the extent of 2 or 3 per cent, of their weight in the course of two weeks, though the range may the author's . ' o ,. . observations be reduced to i per cent, m equable conditions, and increased to 4 or 5 per cent, when the weather is characterised by extreme changes ; (2) That impermeable seeds have practically no hygroscopic reaction, such small changes as do occur, as of i in 1000 or of I in 10,000 of the weight, being either instrumental or connected with loose tissue adhering to the scar or to the raphe ; 1 66 STUDIES IN SEEDS AND FRUITS (3) That with variable seeds, where there is a mixture of permeable and impermeable seeds, the range is inter- mediate in amount, frequently about i per cent., but varying of course with the proportion of permeable seeds. With those " variable " seeds, as with Entada polystachya and Casalpinia Sappan^ where it is possible to distinguish by inspection between the two types of seeds, the difference in the hygroscopic reaction is well marked. Thus, with the first-named plant, a sample of permeable seeds displayed a range of 2 per cent., whilst a sample of seeds almost all impermeable gave a range of O'l per cent. So also with C Great contrasts. " • The same pod 7 days West Indies 4'2 ,, Wet and dry. Pisum sativum (Pea) Two pods, total 120 grains See note England 47 ,, See note. JVoU. — In all the experiments except that on Pea-pods, the pods were kept in one room. The experiments, occupying two months, lasted from June to August. The data for the Pea-pods were obtained by transferring them from a cool and moist room (temperature 50° to 55° F.) to a warm and dry cupboard (temperature 65° to 70° F.) and weighing them after four days. B. — Comparison of the Hygroscopic Ranges of the Air-dried Seeds and Fruit-case of Pisum sativum (Pod) and of Iris fcetidissima (Capsule) under the same Conditions. Iris fcetidissima Pisum sativum , H. R. of seeds 4*0 per cent. ,, 4-5 .. H. R. of fruit-case 3 "6 per cent. >> „ 47 ,. It will be seen from these tables that the variation in weight in response to the changes in atmospheric humidity was usually HYGROSCOPICITY 175 3 or 4 and never over 5 per cent,, whatever the nature or size of the pod. These pods, all of them dry, intact, and contain- ing their seeds, range in length from 3 or 4 inches, as in the case of the Pea, to a couj^le of feet, as with the pods of the species of Cassia and Poinciana ; and there is a great contrast between the relatively fragile appearance of the first-named and the tough, woody aspect of the two last. However, not- withstanding the great differences in size, weight, and texture of these pods, their changes in weight due to the hygroscopic reaction are not far apart. The long pods of Cassia fistula are very sensitive to the hygrometric condition, and vary in weight whilst being handled for the balance. One of them lost I per cent, of its weight whilst a wet morning was giving place to a fine evening. This amounts to a great deal in the balance, since the pods when dry range between 1500 and 2000 grains in weight. They are durable, easily weighed, and might prove useful as hygrometers. It would seem from the foregoing data that permeable leguminous seeds possess much the same degree of hygro- scopicity as their pods. Thus under the same conditions peas gave a range of 4*5 per cent., and the pods with seeds removed 4-7 per cent. (The same principle is also indicated in the case Pods are of Iris fcetidissima^ where the ranges for the seeds and the capsular vflK;ther°t?ie valves were 4*0 and 7*6 per cent.) However, there is no seeds are ^ . , , II- permeable or such relation between the impermeable seed and its pod. All impermeable. the impermeable seeds with which I am acquainted, such as those of Guilandina bonducella, Mucuna urens, Diodea reflexa^ Entada scandens^ etc., are enclosed in pods that in the dry state readily take up and absorb moisture ; whilst in the case of the hygroscopic pods of the species of Cassia, Poinciana, Entada, and Albizzia, dealt with in the table, many of the seeds are impermeable, and the hygroscopic range for any sample of seeds chosen at random is consequently low, usually not over I or 2 per cent., but varying according to the proportion of permeable seeds. As illustrating the hygroscopic behaviour of pods emptied of their impermeable seeds, I may mention 176 STUDIES IN SEEDS AND FRUITS that the dry open pods of Guilandina bonducella and C^salpinia Sappan^ as they lay on the table in my room In Grenada, used to vary in their weight as much as i or 2 per cent, from day to day, especially between the e\iening and the following morning. In concluding this chapter we may observe that the question we put to ourselves a few pages ago as to the extreme limit of a seed's hygroscopicity is very far from being answered, and perhaps it may prove to be not altogether pertinent to the subject we are discussing. With the data at our disposal it seems unnecessary to follow up this special point any further at present ; and indeed it will appear from the subsequent chapters that a number of queries claim a reply first, SUMMARY (i) Hygroscopicity in a seed is defined as the variation of its water- contents in response to the changes in the hygrometric state of the atmosphere (p. 147). (2) After referring to the important memoir on the subject of hygroscopicity in general by Leo Errera and to the display of this quality by vegetable materials in particular, special attention is directed to the researches of Jodin and Berthelot. Whilst the first-named investigator approached the subject from the biological and the second from the physical side, both arrived at the same conclusion : that we are here concerned with a quality that is independent of vitality (p. 148). (3) Jodin, experimenting on living and dead peas, found that they exhibited much the same hygroscopic variation in the course of a year's exposure to ordinary air-conditions (p. 148). (4) Berthelot, experimenting on different vegetable materials (leaves and stems of grasses, etc.), shows that the peculiar property possessed by air-dried vegetable substances and some other materials of regaining from the air the water which they yield up when exposed to a temperature of 100° to 110° C. is a function of the hygrometric state of the atmosphere and is essentially a physico-chemical process independent of life. We will take as illustrating this principle 100 grammes of fresh plant-substance which is reduced to 50 grammes by air-drying in an -ordinary room. Exposing it then to a temperature of 100° to 110° C, its weight is further reduced to 40 grammes ; but on being allowed to HYGROSCOPICITY 177 remain in the room for a few days, it replaces the water lost in the oven by abstracting it from the air and returns to its original air-dried weight of 50 grammes, subject only to the ordinary hygroscopic variation. Precisely the same air-dried weight is ultimately reached if the fresh material is put at once in the oven. In the same way, the leaf that dries and dies naturally on the plant, if placed in the air after the heat test, regains the water lost in the oven. This is Berthelot's principle of reversibility (p. 149). (5) The author, after confirming this principle by appealing to his experiments on Hazel leaves, shows that it presents us with the simplest mode of differentiating the water-contents of plant-tissues, namely, into (a) the water of hygroscopicity, which they hold whether living or dead, and (b) the water of vitality, which they lose when they die (p. 151). (6) He points out that the chemist, when producing by synthesis organic vegetable substances, would allow the atmosphere to supply the water of hygroscopicity, whilst he himself in his creative role would have to furnish the water of vitality. To emphasise this point he gives the result of an experiment on bread, and shows that just as moist fresh plants after being exposed to a temperature of 100° C. can only recover from the air the water lost by air-dried plants in the oven, so bread after the same heat test only gains back what was originally in the miller's flour, but cannot recover the water the baker put into it (P- 153)- (7) After giving a further illustration ot the principle of Berthelot in the results of an experiment on the seeds of Phaseolus multifloriis^ the author proceeds to deal with the implications of this principle (p. 154). (8) The first is that air-dried resting seeds, both permeable and impermeable, contain only the water of hygroscopicity ; and the second, which follows from it. Is that there is in such seeds no water associated with any vital function (p. 155). (9) Since some seeds can retain their germinative powers after being desiccated to an extreme degree, the presumption arises that in their resting state they can dispense with water altogether ; and it is urged that we are not called upon to assign a function to the minute amount of water that may remain after extreme desiccation. A perfectly dry seed protected from the air alone possesses the potentialities of immortality (p. 156). (10) Instead of favouring the longevity of a seed, water is regarded here as the source of its greatest danger. Appeal is then made to the biological significance of hygroscopicity in seeds, and it is shown that whilst the constant hygroscopic reaction limits the life of permeable seeds, its absence in impermeable seeds ensures their longevity. The seed that has the best chance of living for 12 178 STUDIES IN SEEDS AND FRUITS ever is one where a perfectly dry embryo is locked up in a hard impermeable shell or covering (p. 157). (11) Coming to the details of his ow^n observations on hygro- scopicity, the author, after observing that in the nature of things considerations of hygroscopicity are concerned only with permeable seeds, proceeds to discuss the effects of change of climate on the weight of seeds. In this connection he shows that, as a result of transporta- tion from the temperate zone to the tropics and back, permeable seeds varied from 3 to 4 per cent, of their weight, variable seeds (both types in the same plant) less than i per cent., and impermeable seeds practically not at all. The changes of weight, he points out, are included within the ordinary hygroscopic range (p. 158). (12) The methods of determining the range of hygroscopicity in seeds in terms of the variation of their weight are then described, and after giving the results of his observations on more than sixty kinds of seeds, of all sizes and characters, he forms the following general conclusions : — («) That permeable seeds vary usually to the extent of from 2 to 3 per cent, of their weight, though the range may be as little as i per cent, in equable atmospheric conditions, and as great as 4 or 5 per cent, when the changes in the relative humidity of the air are extreme ; [b) That impermeable seeds have practically no hygroscopic reaction, their weight remaining unchanged in spite of great variations m the hygrometric state of the air ; [c) That with variable seeds, where there is a mixture of permeable and impermeable seeds, the range for any ordinary sample is usually about i per cent., differing according to the proportion of permeable seeds (p. 165). (13) Special stress is laid on the value of the hygroscopic reaction stated in terms of weight as a test for proving seeds (p. 166). (14) Additional data are given, as supplementing those already given in Chapter IV, on the influence of the coats on the hygro- scopicity of permeable seeds, and supporting the previous conclusion that the coats tend to restrain the hygroscopic range (p. 167). (15) Results of observations are then given on the effects of age on the hygroscopic behaviour of seeds, and they are shown to be slight (p. 168). (16) Contrary to one's expectation, it is found that a hairy covering, whilst it gives a small hygroscopic range of less than i per cent, to impermeable seeds, but slightly if at all increases the hygroscopicity of permeable seeds (p. 169). (17) Since it follows from the principle of Berthelot that the water of hygroscopicity could have little to do with germination, appeal HYGROSCOPICITY 1 79 is made to the results of experiments made by Nobbe, Jodin, and the author in this connection 5 and it is established that the normal hygroscopic reaction cannot provoke germination, the minimum amount of water required for germination being far in excess of that which a seed would take up unaided from the air (p. 171). (18) The nature of the extreme limit of a seed's hygroscopicity is then discussed ; and the author, after laying stress on the especial risks of mould and condensation to which such experiments are liable, follows the indications of his own experiments in accepting a range in weight of 5 or 6 per cent, of the average weight of the seed as the greatest amplitude of normal hygroscopic variation, the usual range, however, being only 2 or 3 per cent. Hoffmann's results point in the same direction, and precisely the same limits to the hygroscopic range of weight are established by the author's observations on the weight of dry pods (p. 172). (19) The results of these observations on air-dried legumes are then tabulated, and it is shown that leguminous pods with permeable seeds display much the same degree of hygroscopicity as their seeds, the pod being hygroscopic whether the seeds are permeable or impermeable (p. i 74). (20) Although the ordinary maximum of the normal hygroscopic range has been above stated, the author regards the possible extremes of a seed's hygroscopicity as outside the present inquiry. CHAPTER VIII A LAST WORD ON THE HYGROSCOPICITY OF SEEDS The seed's capacity of regaining from the air the water driven off by exposure to a temperature of ioo° C, has been discussed pretty much in the order in which the investigation has been pursued. We have plodded on without always knowing where the road was leading to, diverging first to one side and then to the other, feeling our way often in the darkness, until at length we have stumbled on a clue that throws much light on the whole inquiry. The principle of Berthelot only came under our notice when the investigation was far advanced, and of course it would be possible to recast much of what has been already written in the light now displayed to us. But it has been found easier to present the work in the stages in which it shaped its course and to introduce this principle much as the romance-writer produces his climax towards the end of the story. In my last word on the subject of permeable and imperme- able seeds and the contrasts they present, I therefore view the whole matter from a different standpoint. Up to now the im- permeability of seeds has been the most conspicuous feature in the discussion ; and most of the results obtained shaped them- selves in one way or another in some relation to this quality. Here I propose in a few pages to make the water of hygro- scopicity the centre-point of the discussion. The hydratation of all living vegetable matter may be thus simply stated. There is the water that is lost when the materials are allowed to dry 1 80 THE HYGROSCOPICITY OF SEEDS i8i under ordinary air-conditions, the water of vitality ; and there is the water that this air-dried material loses in the oven when exposed to a temperature of ioo° C. and subsequently regains from the air, the water of hygroscopicity. Air-dried vegetable material, whether living or dead, contains only the water of hygroscopicity, which in these circumstances is one and the same with the water-percentage. There is no room for any more free water in a plant's economy than that which is included in the water of vitality and the water of hygro- scopicity. It is the water of hygroscopicity that we are here concerned with, and we have just seen that in air-dried vegetable substances this is the water that such materials lose in the oven. Let us then determine how this principle applies to seeds. But before we can apply this principle to seeds it is The cor- requisite to determine the correlative of the seed in its pre- [he^g*^ d°^ th resting and resting stages with other portions of the plant, other Manifestly the correlative of the living leaf and the living stem the plant, is the large soft seed of the ripe fruit before the drying and shrinking process begins. Manifestly also the correlative of the dried-up leaf and stem is the same seed after it has com- pleted its shrinking process and has entered upon the resting stage. As far as their water-contents are concerned, there is no difference between the resting seed and the dried-up leaf and stem, all of them holding only the water of hygroscopicity, which is driven off in the oven and subsequently regained from the air. As far as vitality is concerned there is also no differ- ence. We term the leaf dead and say that the vitality of the seed is suspended. What is the difference ? There is room for none. It is thus clear that the seed's hydratation acquires a new significance when we allow the discussion to centre around the water of hygroscopicity and apply the principle of Berthelot. The logical This principle raises the whole question of the necessity prlnciple*of of water for the resting seed, and the logical issue is the Berthelotas J . 1 r • -1 1 T- f- • r • respects the denial or its necessity altogether, l^rom this point or view seed. 1 82 STUDIES TN SEEDS AND FRUITS there is no use for any free water in the economy of a seed that has gone through the normal shrinking and drying process. With these remarks I would now draw attention to the contents of the following table, which is intended to illustrate the hydratation of vegetable substances from this standpoint. Comparison One notices at once in the columns on the right side of ?ontenS^or" the next table how constant in amount is the water of leaves, fruits, hygroscopicity (the water of the air-dried tissue) in the living leaf, fruit, and seed. Stated as a proportion of the living substance, its amount varies usually between 3 and 6 per cent, in different cases ; but since its quantity is regulated by the degree of atmospheric humidity, it is evident that if all experiments were conducted under precisely the same conditions, the range of the variation would be much less, probably only 4 to 5 per cent, for these different parts of a plant. It is, therefore, not in the water of the air-dried tissue (the water of hygroscopicity), but in the water lost by the living tissue when drying under ordinary air-conditions (the water of vitality) that the several parts of the living plant differ from each other. This varies considerably, the range of the results given in the table being 23 to 80 per cent. But notwithstanding this variation, the most important sugges- tion that this table offers to us is that the leaf, the fruit, and the seed are, as respecting their water-contents, all comparable in the living condition, the living seed being the soft uncon- tracted seed, such as we see in the ripe capsule and legume before drying begins. So, again, these parts of a plant are all comparable in the air-dried state, the air-dried seed being the typical resting seed. We cannot consider the seed as a thing apart and place it in a category by itself. The resting seed from this point of view is just as dead as the dried leaf. Both are equally inert vegetable substances, and the only free water that they both contain is what they yield up in the oven and gain back from the air. Occasionally in the leaf, and THE HYGROSCOPICITY OF SEEDS 183 ■B I, To the weight of the air-dried material. c a. : f^ _Tj- _f^ p t^ p 00 p p p P p P P >P P V^ "in C^oo V 00 'vn V 'vncxs "o O 00 r^ ?^ = - : = — — :: ::::: — : 1 r^ p >o -0 ^y^.-*- p r P "? .''"^ .0 P r^ r Vni m \i-,VinV V r^i ^Od r» V^nVi 1 Water lost by the air-dried material in the oven (water of hygro- scopicity). m . U-. >p .f^yi.'l- p _« p >p _-*>p p p r^j- . 00 .^ .■* ."^ ."^ "^ .0 .0 .0 P p p p p p 00 : Vo V k c^ " "0 b b b k i-v b b b 'm VO t^ v£) tv. r'lOO 00 00 vo vo vnvo vo u-, «n U-. 3^.|^ a r^avr«i pt^Ow oomo vo ■+'0 p p _rnvn"-i > r^ « t-. p _mp 0> _D CTvp _Tl--0 Th p p _t^C3\ 00 00 V< b 00 - 'r-. Ini V VOd b vD ">n In b p . r» vp vp yi _r^ .N .0 .0 .0 p p p p p p V» : vb In 'm vja 00 'ct> b b b Vn "c b b b k Iftilj 000 0000 000 0000000 000 0000 000 0000000 1 s J! 38-0 15-0 85-0 Green Windsor variety. Cassia fistula 4-0 15*0 85-0 Canavalia ensiformis . 25-0 i6-o 84-0 Seeds white. ,, gladiata 45 'o 20 -o 8o-o Seeds brownish-red. ,, (species) i8-o 21 -O 79 "o Vicia sativa .... o'3 2I-0 79-0 Bauhinia (species) 4-0 23-5 76-5 Cassia grandis 9-0 25-0 75*0 Vigna luteola 07 25-0 75-0 Mucuna urens 90*0 26-3 73 7 Canavalia obtusifolia I2-0 27-0 73-0 Ctesalpinia Sappan IOO 27-5 72'S Tamarindus indica 20'0 28 -o 72 Entada polystachya 6-6 28-5 7i"5 Seeds permeable. Cassia marginata . lO'O 29*0 71 "o Abrus precatorius . 1-5 30 '0 70*0 Erythrina indica . 12-5 30*6 69-4 ,, velutina 7"S 3i"4 68-6 Sophora tomentosa 2-4 32-0 68-0 Entada polystachya 5"o 32-8 67-2 Seeds impermeable. Erythrina corallodendron 3'2 33*1 66-9 Dioclea reflexa 100 'o 40*0 60 "o Entada scandens . 400 "o 39 'o 61 -o Calliandra Saman 4-0 40-6 59 '4 Leucaena glauca . 0-8 46*0 54 '0 Guilandina bonduc 50*0 49 'o 51-0 (glabra?) . 60 'o 490 Si-o An inland Jamaican species with oblong yellow seeds. Poinciana regia . IO"0 49 '5 50*5 Albizzia Lebbek . 2*2 50*0 50 "o Enterolobium cyclocarpum . 17-0 52-0 48-0 Adenanthera pavonina . 47 53 '0 47-0 Guilandina bonducella . 40*0 58-0 42-0 Acacia Farnesiana 2-0 61 -o 39 'o Csesalpinia sepiaria 4-0 6, -5 Ih 190 STUDIES IN SEEDS AND FRUITS Table showing the Relative Weights of the Coats and Kernels OF Resting Seeds belonging to other Orders than the Leguminos^, the Entire Seed being taken as 100. Relative vireights Average of coats and weight kernel taking Order. of a the entire seed Remarks. single as 100. seed in grains. Coats. Kernel. Tamus communis Dioscorege 0-3 5-0 95-0 Arum maculatum Aracese 07 12-5 87'S Ipomcea tuberosa Convolvulacese 25-0 20 '0 8o-o Iris Pseudacorus Iridese 07 20 'o 80 -o Canna indica . Cannacege 2'S 21 -O 79-0 Mammea americana . Guttiferae 550-0 23-5 76-5 Swietenia Mahogani (Ma- Meliacese 3 7 25-0 75 'o Without wing hogany) 18 and 82. Citrus decumana (Shad- Aurantiacese 4-0 26 'o 74 'o dock) ^sculus Hippocastanum Hippocastaneie 130*0 27-0 73-0 (Horse-chestnut) Anona reticulata Anonacece 4-0 27-0 73-0 Citrus Aurantium (Orange) Aurantiacece 2-5 28-0 72-0 Ricinus communis (Castor Oil) Bignonia (near jequinocti- alis) Hura crepitans . Euphorbiacese 3-0 28-0 72-0 Bignoniaceas 5-0 28'0 72 '0 Euphorbiacese 20 "o 30*0 70*0 Anona palustris Anonaceae 4-0 31-0 69*0 Moringa pterygosperma . Moringese 5"o 33 'o 67-0 Without wings 30 and 70. Anona muricata Anonaceae 6-0 34 -0 66-0 ,, squamosa ,, S'o 35 "o 65 "0 Momordica Charantia Cucurbitaceae 30 37'5 62-5 Ipomoea dissecta Convolvulaceae 27 37*5 62-5 Jatropha Curcas Euphorbiaceae I2-0 38-0 62*0 Iris foetidissima Irideas 0-8 40*0 60 'o Bignonia (species) . Bignoniacece 7'5 40'o 60 "o Anona Cherimolia . Anonaceae 8-0 42-0 58-0 Ravenala madagascariensis Musacese 6-0 42-5 S7'5 Chrysophyllum Cainito Sapotaceae II "o 43 "o 57-0 (Star Apple) Cardiospermum Halica- Sapindaceae I "4 43 'o 57 "o cabum Ipomcea tuba . Convolvulaceae 5"o 44 "0 56-0 Without hairs 417 and 58-3. pes-caprse . >> 3'o 47-0 53'o Without hairs 45-5 and 54-5. Hibiscus Sabdarifa . Malvaceae o'4 47 'o 53-0 Thespesia populnea . " 3"o 487 5i'3 Without hairs 48 and 52. Cardiospermum grandi- Sapindaceae 2-5 50-4 49-6 florum THE SHRINKING AND SWELLING SEED i Table showing the Relative Weights, etc. — continued. Relative weights Average of coats and weight kernel, taking Order. of a single the ent ire seed as : 00. seed in grains. Coats. Kernel. Hibiscus esculentus . Malvaceae 07 51-0 49 "0 Colubrina asiatica . Rhamnese 0-6 517 48-^ Achras Sapota (Sapodilla) Sapotacea; 9'5 535 46-5 Luffa acutangula Cucurbitacese i-o 55 "0 45 'o Hibiscus elatus Malvaceae 0-6 56 'o 44 "o Without hairs 5 3 "4 and 46'6. Lucuma mammosa . Sapotacese 220 "o 56-0 44-0 Gossypium barbadense Malvaceae I '4 57-0 43 "o Without hairs (Cotton) 40 and 60. Chrysophyllum (species) . Sapotacese 6"o 59 '0 41-0 Calotropis procera . Asclepiadese 0"2 61 'o 39 'o Without hairs 50 and 50. Sapindus Saponaria . Sapindaceae 13-0 64'o 36-0 Gossypium hirsutum . Malvaceae 1-6 687 3'"3 Without hairs 44-6 and 55-4. If, then, we group together the results for all these seeds of 82 species, we get the following arrangement : — Grouping of the Seed-coat Ratios for the Resting Seeds of 82 Species of Plants, the Weight of the Entire Seed being taken as 100. Below 20 per cent. 20 to 29 per cent. 30 to 39 per cent. 40 to 49 per cent. 50 to 59 per cent. 60 to 70 per cent. Total. Leguminous . Non-leguminous . 9 2 12 II 8 8 5 10 4 9 2 3 40 43 II 23 16 IS 13 5 83 Note. — The total here is not 82 but 83, as Entada polystachya with its two types of seeds has been entered twice. Since just half of the species possess seed-coat ratios The general indications between 20 and 40 per cent, (both inclusive), it is probable that supplied by the average weight of the seed-coats would be about 30 per of^tlS^°eed cent,, rather below for leguminous seeds and rather above for coat ratios, 192 STUDIES IN SEEDS AND FRUITS seeds of other orders. The ratio may have a generic value in some cases, as in Erythrina, Guilandina^ Phaseolus, and Hibiscus ; and in others it may vary considerably within the limits of a genus, as in Anona and Ipomcea^ so that as possible disturbing influences the one may be set against the other. Whilst ranging widely in some orders, as in Leguminosae, it may be comparatively uniform in others, as in Malvaceae and Sapotaceae. Size has little or nothing to do with the ratio, excluding very small seeds less than one-tenth of a grain, which are not here discussed. Looking down the lists, we find large and small seeds frequently associated on account of the similarity in their ratios. Thus the seeds of Canna indica and of Mammea americana have similar ratios, although about 220 seeds of the first will be required to make the weight of a single seed of the second. So also, if we compare the imper- meable seeds of the two species of Entada^ we find that their ratios are not far apart, notwithstanding that at least 80 seeds of E. polystachya are required to weigh down a single seed of E. scandens. Then, again, with the two sapotaceous plants, Lucuma mammosa and Achras Sapota^ we notice that the propor- tional weights of the seed-coats are nearly the same, although the difference between the weights of a single seed are as 220 to 9-5. Size, of course, goes with weight in all these cases. The con- In resting seeds the seed-coat ratio is sufficiently constant seed-coat ^ ^^ form a character for the species, though, as we have seen th*^r^'t^^"f ^^°^^> ^^ "^^y ^^^7 considerably within the limits of a genus. a species. Its constancy for a species is illustrated in the results for a few plants given below ; but my data do not often lend themselves for such a comparison, since in determining this relation it was my usual practice to employ a number of seeds at the same time. Guilandina bonducella^ number of seeds tested 22, range of seed-coat ratios 53 to 64. Guilandina bonduc^ number of seeds tested 8, range of seed-coat ratios 44 to 52. Entada scandens^ number of seeds tested 5, range of seed-coat ratios 377 to 40-5. THE SHRINKING AND SWELLING SEED 193 Anona mur'icata^ number of seeds tested 12, range of seed-coat ratios 31 to 38. Carina ind'tca^ number of seeds tested 10, range of seed-coat ratios 18 to 25. With regard to the influence of variation in size as inter- preted by weight on the seed-coat ratio in the same species, I am not able to give many data, as my mode of work rarely admitted such a comparison. The following results for 22 The influence seeds of Guilandina bonducella indicate that there is no great difference. The range of the variation in weight was from 33 to 45 grains, the proportion of the seed-coats being slightly greater in seeds above than below 40 grains. Thus : — Above 40 grains in weight, average seed-coat ratio 58*5. Below „ „ „ 57-5. However, carefully guarded observations on a large number of seeds of the same age and from the same plant are necessary for the elucidation of this point. As respecting the influence of appendages on the propor- The influence tional weight of the seed-coverings in resting seeds, we will deal ages on"the first with hairs and then with wings. In the table subjoined ratio'^°** there are given the data for eight kinds of hairy seeds belonging (A) Hairs. to three families, Asclepiadeae, Malvaceae, and Convolvulaceae, of which the two first are especially notable for the hairiness of the seeds in some genera. The results are arranged in order accord- ing to the proportional weight of the hairs, commencing with the seeds where the relative weight is smallest. It is evident that ordinary pubescence as illustrated in the case of Ipomcea pes-capr^ adds but little to the weight of a seed. Nor does it make much difference if a pubescent or puberulous seed, as in the case of Ipomcea tuha^ is bordered by longer hairs at the angles ; but when these hairs are abundant and woolly, as in the seed of Ipomcea peltata^ the hairy covering may make up 9 per cent, of the total weight. The proportions for these convolvulaceous seeds are probably typical of a good many seeds of other families. The great development of hairs that we find in some Asclepiads and in some malvaceous genera is not common in the plant-world. 13 T94 STUDIES IN SEEDS AND FRUITS Table illustrating the Effect of Hairs on the proportional Weight of the Seed-coats in Resting Seeds. Thespesia populnea Ipomoea pes-caprse Ipomcea tuba Hibiscus elatus Ipomoea * peltata Calotropis procera Gossypium t barbadense Gossypium t hirsutum (?) Aver- Order. age weight of a seed in grains. Malvaceae 3-0 Convolvulacese 3-0 do. 5"o Malvaceae 0-6 Convolvulacese I'S Asclepiadeae 0"2 Malvaceae x-4 do. 1-6 Character and extent of the covering of hairs. Villous at the base and angles Pubescent over most of sur- face Puberulous, but villous at the angles Villous down over all sur- face Pubescent, but with abund- ant woolly hair at the angles Terminal coma with hairs i- 1 1 inches long and spread out like a pappus, the whole easily carried by winds Seed invested with abund- ant woolly hair (cotton) do. Proportion of parts, taking the weight of the entire seed as loo, 28-5 43*5 47-8 44 "4 40 "o 50*0 28-5 25-2 So'6 52-8 56*0 44 'o 43 "o Effect of the hairs on the relative weight of the coats. Percentage of the seed-weight. With hairs. 49 4 47-2 44 'o 56*0 57-0 Without hairs. 48-6 457 417 53*3 50*0 44 '6 * The proportional weight of the hairs was alone ascertained. t In Gossypium {hirsutum ?) the cotton adheres firmly to the black seed, becoming dirty grey and matted next to the seed. Of the relative weight of the hairy covering (43 '5 P^"^ cent.), white cotton forms 37*1 and the grey matted portion 6 "4. In G. barbadense the cotton easily separates from the black seed, there being no adherent covering of matted hairs. THE SHRINKING AND SWELLING SEED 195 We there have genera, like those of Calotropis and Gossypium, where the hairs may form a quarter or even nearly half the weight of the seed ; and, as in one species of Gossypium named in the table, the hairs may be heavier than the kernel. In illustration of the effect of large wing-like appendages (B) Wings, on the relative weight of the seed-coats, I will take the seeds of three familiar tropical plants, Swietenia Mahogani, Moringa pterygosperma^ and Tecoma stans^ the data for which are given in the following table. Table illustrating the Effect of Wings on the proportional Weight of the Seed-coats in Resting Seeds. Character of the wing or wings. Weight in grains. Seed-coat ratio, taking the entire seed as loo. Seed and wing. Coats and wing. Entire seed. Wing or wings. Wing percen- tage. Coats without wing or wings. Coats with wing or wings. Exclu- ding wing or wings. Inclu- ding wing or wings Swietenia Mahogani Moringa pterygosperma Tecoma stans A terminal oblong wing as in Pinus Seed round with three vertical wings Thin, flat, ob- long seed with margin- al winggreat- ly prolonged at the two extremities 3'S o-is 0-35 0"20 0'02I 4 .. 14 ., o'S5 '•45 0*90 1-65 17-5 30-2 257 33 Pinus Terminal ob- long wing, but formed from the scale, and not truly comparable with the above o-is 0*030 20% ... Note, — In Pinus the seed is only partially enclosed in the base of the wing, and is exposed on one side. 196 STUDIES IN SEEDS AND FRUITS Withered leaves and dried seeds are in the same category. Excluding the Pine seeds, which are not strictly comparable with ordinary winged seeds, we observe that in the three types of seeds here exemplified the weights of these appendages vary in amount between 4 and 14 per cent, of that of the entire resting seed. The wings are here greatly developed, so we may infer that in ordinary "margined" seeds, where the alar appendage is reduced to a narrow border, there is very little addition to the seed's weight. Now wings are functionally useless as we observe them in the resting seed, or we may put it in another way, and say that though actively functioning in the soft living seed within the living fruit, they have no biological significance in the dry seed of the withered fruit. Being dried up and dead they could only serve an accidental function in a resting seed which is practically in the same condition. A more natural comparison of these types of winged seeds would therefore be obtained by contrasting them in the living moist condition within the fruit when the wings are actively functioning organs. In the resting seed the wings are dead and dried up and could only serve acci- dental ends. As far as the capacity for transportal by wind is concerned the withered leaf and the dry winged seed are in the same category. That the seed possesses the power of reproducing the plant is an accident as regards its fitness for wind- transportal. The puff of wind will carry along both the seeds that are germinable and the seeds that have lost this power ; and we cannot distinguish between them as respects either appearance, weight, or size. Yet the dispersal of seeds by winds is real enough. The seeds of Tecoma stans^ which are about an inch long, weigh just about the same as a piece of newspaper cut to the same size. The wind when strong could carry them great distances, and the like may be said for the seeds of the Pine. The much heavier winged seeds of Moringa^ as experiment shows, are but little aided in this way by their appendages. In an ordinary breeze a Mahogany seed as it falls out of the dried dehiscing THE SHRINKING AND SWELLING SEED 197 fruit would be transported, as I find, only a few paces ; but during a strong gust I have known it to be carried a hundred feet. In this connection the remarks of Dr Goebel on the Goehei's " parachute-apparatus " of fruits are well worth quoting, since concerning they would apply in a sense also to seeds. In his Organo- f''^^^^- graphy of Plants (English edition, ii. 570), he writes as follows : — " . . . Many arrangements which have hitherto been considered merely as a parachute-apparatus on the ripe fruit are in my view to be considered as a trans- piration-apparatus for the ripening fruit, and these sub- sequently can be used for distribution, but are not necessarily for this. . . ." However, the function of the wings of the moist seed in the living fruit would probably be haustorial. In other words, these appendages would greatly The author's ,, . r 11- ^ T ^u viewascon- increase the seed s capacity tor absorbmg water, in tne cerning instance of the moist white seeds of the Mahogany tree the f^^^tg^ by^' increase of the area of the receiving or absorbing surface due Mahogany to the wing is very large, the alar surface-area being more than double that of the seed, as is indicated in the following measurements : — Surface-area of the seed without the wing, 450 square millimetres. „ wing alone, 1050 „ „ winged seed, 1500 „ In the case of the seed of the Pine it was long ago suggested by Goeppert, as quoted by Nobbe in his Handbuch der Samenkunde (p. 49), that the wing exercised the function of a funicle or umbilical cord. With the Mahogany seed it is probable that in the closed living fruit the wing also serves for storage of water. As shown in the tabulated results of my observations given below, the wing and coverings of the soft white seed in the full-grown fruit hold nearly twice as much water as the kernel, losing about 89 per cent, of their weight in the drying and shrinking process, as against 49 per cent, lost by 98 STUDIES IN SEEDS AND FRUITS the kernel. This excess of water in the wing is probably associated with the watery condition of the fleshy placental column in the living fruit. It will be seen from the discussion of the regime of the drying Mahogany capsule in Note 22 of the Appendix that the placental column loses about 70 per cent, of its weight as the fruit dehisces and dries, the loss of the capsular walls being about 60 per cent. Thus we find a regular gradation in the water-contents of a living Mahogany fruit, viz. about 60 per cent, in the fruit walls, about 70 per cent, in the placental column, and about 90 per cent, in the wings of the seeds. We are here referring to the indications supplied by the loss of weight during the natural drying process. The Shrinking Regime of the Pre-resting Mahogany Seed, that IS TO SAY, OF the Soft, Uncontracted, Moist Seed of the Full-grown Fruit. (The data required for this purpose are the weights of the seeds and the proportion of parts in each condition. ) pre-resting and resting Condition of the wing. Absolute and relative weights, the first in grains. Loss of weight during the natural drying of the fruit. Pre-resting. Resting. Pre-resting. Resting. Wing and seed-coats Kernel . Entire seed White, heavy, thick, soft, and flabby Brown, light, thin, and crisp 8-4 (60) 5-6 (40) 14-0(100) 0-95 (25) 2-85 (75) 3-80(100) 887%. 73%. The pro- portional weights of parts in the three condi- tions of the seed supply data for determining the regime of the shrinking and swelling seed. We now pass on to the determination of the proportional weight of the seed's coverings in the two other conditions of the seed, the pre-resting state, when the seed attains its full size in the moist ripe fruit, and the state immediately preceding germination, when the resting seed has absorbed much water, and the embryo is on the eve of continuing its growth that was brought to a standstill during the shrink- ing process. These data being obtained, we shall possess THE SHRINKING AND SWELLING SEED 199 the seed-coat ratios for the three conditions of the seed, the pre-resting, the resting, and the swollen state pre- paratory for germination ; and in contrasting them we shall be constructing the regime of the shrinking and swelling seed, of the seed as it enters upon the rest- ing period, and of the seed as it subsequently swells for germination. It has already been established in Chapter II. that the The return seed takes up when swelling for germination the water that ingseedto it lost in the shrinking process, the weight lost during ^erghf-Jfthg the shrinking being approximately regained during the P[|^gYs*^°^t a swelling. But it will be brought out in this chapter that in simple the attainment of this result the parts of the seed take P'^^^^^s. different shares, the coverings of the swelling seed never regaining all the water originally lost, whilst the kernel takes up more water than it held in the pre-resting state. However, the loss of the one tends to counterbalance the gain of the other in the following fashion. The deficiency of the coats is generally larger than the excess of the kernel ; but since the coats as a rule are only one-half or one-third of the weight of the kernel, the ultimate result is usually not much affected. Nevertheless, this is sufficient to show that the return of the resting seed when swelling for germination to its original weight as a pre-resting seed is not a simple process, and that such a result of experiments can only be regarded as approximate in value. In the following table are given the data for construct- Theeiements ing the regime of the shrinking and swelling seed in a jngthe'^""" considerable number of cases. Knowing the weight of |nd"swening the resting seed, the shrinking and swelling ratio, and the regime, proportional weight of the coverings (the seed-coat ratio) in the three conditions of the seed, the determination of the shrinking and swelling regime for any seed named in the table can be readily effected, as shown in the example added. 200 STUDIES IN SEEDS AND FRUITS Table showing the Proportional Weight of the Seed's Coverings (Seed-coat Ratio) in the Three Conditions of the Seed, THE Large, Moist^ Pre-resting Seed of the Ripe Fruit, the Resting Seed, and the Seed Swollen for Germination. (From the data given in this table the regime of the shrinking and swelling seed can be readily constructed, as explained in the example given, P. = permeable ; I. = imper- meable ; V. = variable. ) The The seed-coat ratio in the three con- '6 Order. Average weight ol a resting seed in shrinking and swell- ing ratio, the weight of the rest- ditions of the seed, the weight of the entire seed being taken as 100. '0 1 grains. ing seed Cj being , ^ ^ Ss.o taken as r. ^ S3 ■S % ^ c V. 30*0 28-0 V. Abrus precatorius . Leguminosa; I '5 2-25 Acacia Farnesiana . )) 2"0 2 'GO 59-6 60 -8 53-6 V. Adenanthera pavonina . 47 2-42 53'o 50-1 I. ^sculus Hippocastanum Hippocastanese IjO'O 2 -20 35'o 27-0 p. Albizzia Lebbek . LeguminosEe 2 '2 2-27 50-0 3^ '5 V. Anona muricata Anonaceee 6-0 I '43 ... 33-0 26-4 p. Arum maculatum . Aracese 07 1-63 33 'o 12-5 p. Bauhinia (species) . Leguminosse 4-0 2'lO 23-0 23"5 24-0 V. Bignonia (near aequi- Bignoniacete 5"o 2-30 38-0 28-0 p. noctialis) Caesalpinia Sappan Leguminosse lO'O 2-20 32'S 27*5 26-4 V. , , sepiaria ,, 4-0 2-22 6i'o 61 -4 53'o V. Cajanus indicus ^ 3*0 2"IO 28-0 13-0 16 -o p. Calliandra Saman . ]'^ 4"o 2-50 40-6 35 "o V. Canavalia ensiformis \\ 25-0 2-17 i6'o 21-5 p. ,, gladiata ,, 45-0 2'II 19-8 23-3 V. ,, obtusifolia . ,, I2*0 2-50 45 '0 27-0 34-6 V. ,, (species of) . jj i8-o 1-94 21'0 "•3 V. Cassia fistula . ,, 4-0 2-50 26 -3 15-0 17-2 V. ,, marginata . ,, lO'O 2-10 29-0 28-3 V. ,, grandis 9-0 2'II 25-0 26-0 V. Dioclea reflexa — (A) Impermeable ,, 90 "o 2-07 48-5 40 "0 33'S I. (B) Permeable seed . jj 1 00-0 1-86 48-5 38-0 36-1 p. Entada polystachya— (A) Impermeable ,, 5'o 2-50 27-2 32-8 25-1 I. (B) Permeable seed . ,, 6-6 2-25 27-2 28-5 24*0 p. Entada scandens . 40o'o 2-50 46*0 39'o 31 'O I. Enterolobium cyclo- ,, i7"o 2-30 ... 517 46*0 V. carpum Erythrina corallodendron ,, 3 "2 2-i6 33-1 30-4 V. ,, indica jj 12-6 2-49 30'6 28-1 V. ,, velutina . ,, 7*5 2-40 3 1 '4 34-0 V. Faba vulgaris (Broad jj 30-0 2-00 33 "0 15-0 13-0 p. Bean) Guilandina bonduc " 50*0 2-47 487 48-2 I. THE SHRINKING AND SWELLING SEED 201 Table showing the Proportional Weight, etc. — continued. The The seed-coat ratio in the three con- TJ Order. Average weight of a resting shrinking and swell- ing ratio, the weight ditions of the seed, the weight of the entire seed being taken as loo. ^0 seed in of the rest- 1 grains. ing seed being £-5 w ii S.2 5 taken as i. "13 ■0 ^^ ^l 1 ^ o.S Guilandina bonducella . Leguminosse 40-0 3-00 6o-8 57-8 53 "o ^ glabrae?) . ,, 60 -0 2-52 49 '0 44 "4 \. Hura crepitans Euphorbiacese 20'O 2-10 57 -0 30*0 44 "o P. Ipomoea pes-caprse Convolvulacese 3-0 2-50 S5'o 47 -o 40 'o L ,, tuba ,, 5-0 2-63 46*0 44-0 32-0 L ,, tuberosa . 25'0 2-50 21-0 19-0 V. Iris foetidissima Irideoe 0-8 310 65*0 40*0 6o"o P. ,, Pseudacorus . ^j 07 2'00 40-0 20 'O 25-0 P. Leucaena glauca Leguminosse 0-8 2-6o 45-2 46-2 38"5 L Luffa acutangula . Cucurbitaceae i-o 177 SS'o 62 "o P. Mucuna urens Leguminosse 90"o 2'00 47'S 26-3 20 '4 L Phaseolus multiflorus i8-o 2'00 25-0 12-5 9-0 P. Pisum sativum— (A) Smooth seeds ,, 6-0 1-90 25-0 8-5 9-0 P. (B) Wrinkled seeds* . ,, 7-0 2-40 30-0 10 "o 9-4 P. Poinciana regia ,, 10 2-30 45'S 49'5 41 '0 V. Ricinus communis . EuphorbiaceK 3'o I '33 280 23-0 P. Swietenia Mahogani MeliaceK 37 370 6o-o 25-0 P. Tamarindus indica Leguminosse 20 'o 2-I5 28-0 33 'o V. Thespesia populnea Malvacese 3'o 1-82 37-0 487 45 '0 V. * The collection of water under the coats seriously affects the regime of the wrinkled seed. I will take the seed of Entada scandens to illustrate the illustration employment of the data in the above table for determining SJedlt'ain^^ the shrinking and swelling regime of a seed. All that is the table, required are the shrinking and swelling ratios of the entire seed and the proportional weight of parts in the three conditions of the seed, the pre-resting, the resting, and the swollen state. The ratios are given in two ways, one where the weight of the resting seed is taken as i, the other where that of the pre-resting seed is taken as 100, the first method being required for purposes of comparison when, as often happens, the data for the swelling process are alone available. 202 STUDIES IN SEEDS AND FRUITS Entada scandens. (Weight of resting seed 400 grains. Shrinking and swelling ratio 2*5, taking the weight of the entire seed as i. Pr. =pre-resting seed ; R. = resting seed ; and Sw. =the seed swollen for germination. ) Proportional weights stated Weights Shrinking and swelling ratios. as a percentage of the weight of the entire seed. grains. Weight of rest- ing seed as i. Weight of pre- resting seed as 100. Coats . Kernel Entire Pr. 46 54 100 R. 39 61 100 Sw. 69 100 Pr. 460 540 1,000 R. 156 244 400 Sw. 310 690 1,000 Pr. 2-95 2-21 2-50 R. I I I Sw. 2-50 Pr. I op 100 100 R. 34 45 40 Sw. 67 128 100 We will commence with the comparison of the regimes of two types of leguminous seeds, that of Guilandina honducella and that of Faha vulgaris (Broad Bean), the first with im- permeable, the last with permeable coverings. Employing the data given in the table, we get the following results : — Comparison of the Shrinking and Swelling Regimes of an Impermeable and a Permeable Leguminous Seed, taking the Weight of the Pre-resting Seed as 100. Pre-resting seed. Resting seed. Seed swollen for germination. Guilandina honducella (im- J permeable) . . . 1 Coats Kernel Entire 100 100 100 32 36 33 87 120 100 Faba vulgaris (permeable) - Coats Kernel Entire 100 100 100 23 63 50 39 130 100 Note that the pre-resting seed is the soft seed of the full-grown moist fruit. Illustrations We here see that whilst in both seeds the coats of the oftheshni^^- Swelling seed fail to regain the water lost in the shrinking ingand process and the kernel ultimately holds more water than swelling .,11., • 1 • • 1- seed. It held in the pre-resting state, there is an important dis- THE SHRINKING AND SWELLING SEED 203 tinction in their behaviour, the deficiency in the gain of the coats of the permeable seed being much the largest. But the difference is only one of degree, since both types of seeds illustrate the same general principle. There is much that is extremely suggestive in these figures ; and if we compare them with those tabulated below which represent the average results for impermeable, variable, and permeable leguminous seeds, we perceive that the regime illustrated by the two type-seeds just dealt with applies generally to seeds of the family. It is as true of the individual species as it is of the aggregate. Of the seventeen species of leguminous seeds, for which complete data are given in the preceding general table, all but one, that of Bauhinia^ conform to the principle that during the swelling of the resting seed for germination the coats fail to regain all the water lost in the shrinking process, whilst the kernel not only regains all the lost water but absorbs Table giving the Average Shrinking and Swelling Ratios for THE Coats, Kernel, and Entire Seed of Impermeable, Vari- able, AND Permeable Leguminous Seeds. The impermeable seeds used are those of Dioclea reflexa, Entada scandens, Entada polystachya (impermeable type), Gui/andina bondticella, Lencana glanca. The variable seeds are those of Acacia Farnesiatta, Bauhinia, Ccesalpinia sepiaria, CcEsalpittia Sappan, Canavalia obtiisifolia. Cassia fistula, Poinciana regia. The permeable belong to Cajanus indicus, Pisum sativum, Faba vulgaris, Phaseolus 7nultiflorus. Shrinking and swelling ratios for the pre-resting (Pr.), resting (R.), and swollen (Sw.) seed stated in two ways, the pre-resting seed being taken as loo in the first way, and the resting seed as i in the second way. Number of species. Coats. Kernel. Entire Seed. Impermeable = 1 Pr. R. 100 39 (2-6 I Sw. 80 2-1) Pr. R. 100 42 (2-4 I Sw. iiS 2-8) Pr. R. Sw. 100 40 100 (2-S I 2-S) Variable . 7 100 40 (2-S I % 100 47 (2-1 I 112 2-4) 100 44 100 (2-3 I 2-3) Permeable ^ i TOO 22 (4-S I 42 I -9) 100 60 (17 I 122 2-0) 100 50 100 (2'0 I 2'o) 204 STUDIES IN SEEDS AND FRUITS considerably more. It is likely that further investigation may bring the behaviour of the Bauhinia seeds into line with that of other seeds. It would appear from the table that as a general rule, although the coats of the per- meable seed shrink more than those of the impermeable seed, both double their weight in the swelling process. On the other hand, the kernels of permeable seeds, though shrinking less, only double their weight when swelling for germination, whilst the kernels of impermeable seeds shrink more and increase their weight threefold in the swelling process. It becomes at once apparent from all the tabulated results that the resting state leaves its impress on the germinating seed, especially with reference to the behaviour of the coats. The impress But the impress is also evident in the contrasts in appearance state^on1;be^ presented by the coverings in the pre-resting seed and in the forgennina? ^^^^ Swollen for germination. In the pre-resting state the tion. coats are usually thick, moist, fleshy, easily marked by the finger-nail or cut with a knife, and elastic or yielding. In the swollen seed on the eve of germination the seed's coverings are thinner, relatively dry, tough, and unyielding. With a seed like that of Entada scandens these characteristics are well displayed, and we may take it as a good example of the mechanism of germination in seeds of this type. In the pre-resting or so-called unripe state we have a white, moist, flabby seed with coats 3 to 4 millimetres thick. In the swollen, germinating stage we have a dark brown seed with tougher and drier coverings between 2 and 2^ millimetres thick. But a more important indication of the impress of the resting state appears when we compare the changes in weight with the changes in size. The seed on the point of germi- nation is usually rather smaller than the so-called unripe or pre-resting seed. This is brought out in the measure- ments for four leguminous seeds in their several stages below given. THE SHRINKING AND SWELLING SEED 205 Measurements in Millimetres of Leguminous Seeds in the three conditions Pre-resting Resting. Swollen for germination. Entada scandens Guilandina bonducella Csesalpinia sepiaria Canavalia obtusifolia . 66-5 25-0 14-0 21'0 47*5 i8-o 10*0 iS'5 64 24 13 20 Taking the case of the disc-shaped seed of Entada scandens^ we find that the embryo or kernel of the seed swollen for germination is no longer enclosed, as in the pre-resting seed, in soft, yielding coverings, in which it lies easily, filling the space without tension. It is now invested by tough, unyield- ing, drier coverings ; and, whilst the kernel has increased its weight by about 28 per cent, through water-absorption, the coverings are drawn tighter round it than in the pre-resting state. As shown in the regime before illustrated, the kernel that weighed 540 grains in the pre-resting state now weighs 690 grains, and the seed, as we have just seen, is rather smaller. The strain produced by the swelling embryo within its tightened coats must be great, and the coverings yield opposite the hilum, the process being purely physical and not necessarily followed by germination, itself a purely biological process in which the active growth of the hypocotyl is involved. The tension within the seed is well illustrated at times when the rupture of the coats is from some cause delayed and one or both of the thick cotyledons within break right across. The selection of the hilum or scar for the seat of the first rupture of the coats is on physical grounds not easy to explain, since the coverings are here thicker than elsewhere ; but this is a point that is discussed in Note 23 of the Appendix. The great tension existing within the seed on the eve of germination is also well exemplified in the regimes of the three other leguminous seeds, for which measurements have above been given, viz. Casalpinia sepiaria^ Canavalia obtusifolia^ The condi- tions of strain within the swelling seed. 2o6 STUDIES IN SEEDS AND FRUITS and Guilandina bonducella. The swollen seed ready for ger- mination is rather smaller than the pre-resting seed of the ripe moist pod. Its coats are tougher and drier, holding in one instance 23, and in two instances 13 per cent, less water than in the pre-resting condition. Yet within these tightened relatively unyielding envelopes lies in each case a kernel that is about 20 per cent, heavier through the absorption of water, and correspondingly larger than in the pre-resting state, a condition of strain necessarily resulting. As a rule the conditions of strain within the swelling seed are at first rather more pronounced in permeable than in imper- meable leguminous seeds. This is well shown in the average results for these two types of seeds already given on a previous page. But in the case of the permeable seed the relatively thin coverings soon give way, whilst with the impermeable seed the tougher and thicker coats resist the strain of the swelling kernel for many days, and the strain, though less at first, becomes very great when the swelling of the kernel is far advanced. It occasionally happens with impermeable seeds that the conditions of strain are intensified from the beginning, and, as in the case of the seed of Mucuna urens, a somewhat abnormal regime is displayed. Whilst its coats behave like those of a permeable seed, its kernel plays the role of a seed of its type, but in an exaggerated fashion. Coats : pre-resting 100, resting 28, swollen for germination 43. The regime of Kernel : pre-resting 100, resting 70, swollen for Mucuna urens \ germination 152. Entire : pre-resting 100, resting 50, swollen for germination 100. Whilst studying this seed in the West Indies, I noticed that the complete soaking of the coats was not requisite for germination, the radicle often protruding through coverings partly dry. This occurs occasionally with seeds like those of Dioclea reflexa and Entada scandens. In some other leguminous THE SHRINKING AND SWELLING SEED 207 seeds, usually of the variable type, the tension produced during the swelling of the resting seed is but slight, as with those of Entada polystachya and Poinciana regia. In other variable seeds again, as with those of Cassia fistula, the strain may be great ; whilst in rare instances, as with the seeds of Bauhinia, there may be none at all. The case of Cassia fistula is interesting, as it indicates that albuminous seeds may behave like exalbuminous seeds in this respect. ' Coats : pre-resting 100, resting 23> swollen for germination 65. ia fistula ■ Kernel : pre-resting 100, resting 46, swollen for germination 112. Entire : pre-resting 100, resting 40, swollen for germination 100. ' Coats : pre-resting 100, resting 49, swollen for germination 105, mhinia Kernel : pre-resting germination 99. 100, resting 47, swollen for Entire : pre-resting 100, resting 48, swollen for germination 100. Be This is all that can be said here for the combined shrinking and swelling regime of leguminous seeds, and the data in the general table before given must be allowed to tell their own story in individual cases. The reader can work out for himself the regime of any particular seed. Should he wish to contrast the impermeable and permeable types in the same species, he will find in the cases of Dioclea and Entada that the permeable seed in its behaviour, strictly speaking, comes between the two. But the seeds of other orders may behave like leguminous seeds when shrinking before entering upon the rest-period and when swelling previous to germination. What we may term the convolvulaceous regime, as exemplified in the The con- behaviour of the two species of Ipomxa illustrated below, does rlghjj|'.*^^°"^ not seem to differ materially from the leguminous regime, although the seeds themselves differ much as respects the albumen, embryo, and other characters. In its conspicuous 2o8 STUDIES IN SEEDS AND FRUITS features it comes near the average regime of a leguminous impermeable seed as previously given. Additional data relating to the swelling ratios of the parts of seeds. The Convolvulaceous Regime. Pre-resting seed. Resting seed. Seed swollen for germination. Ipomoea pes-capras— Coats Kernel .... Entire I GO lOO lOO 34 47 40 73 133 100 Ipomoea tuba — Coats Kernel .... Entire lOO 100 lOO 37 40 38 70 126 100 There are seeds where the shrinking and swelling regime presents special difficulties, as in the case of those of Hura crepitans^ the well-known Sand-box tree, belonging to the Euphorbiaceae. In this instance when the pre-resting seed is fully matured, the seed-coverings have already lost consider- ably in weight, whilst the kernel has just attained its maximum development. We encounter here the same difficulty that we meet with when following the development of certain fruits, such as those of Quercus Robur, Barringtonia speciosa^ and Cocos nucifera (Coco-nut), in Chapter XIV. The behaviour of the seed of Hura crepitans is fully dealt with in Note 24 of the Appendix and requires no further mention here, except the remark that, like other seeds with oily kernels, such as those of Ricinus and Anona^ there is much less water lost in the shrinking process and regained in the swelling process than is required by a typical leguminous seed. Evidently oily seeds have a regime of their own. We will now utilise the results obtained for the seeds of a considerable number of plants where the swelling phenomena were alone observed, and we shall thus obtain a large accession to our data relating to the swelling ratio of seeds preparing for germination with especial reference to the respective parts THE SHRINKING AND SWELLING SEED 209 taken by the coats and the kernel in the process. We have, in fact, the " swelling " data for the coats and the kernel in forty-four species, of which all but ten are leguminous, the residue belonging to a variety of orders, such as Convolvulaceae, Euphorbiaceae, Iridaceas, etc. Many influences come into play in determining the swelling The dis- ratios of the coats and of the kernel of the resting seed pre- influeiKes. paring for germination, influences that are more numerous, however, with the seed's coverings than with its kernel. In the case of the kernel there is the relative dryness of that of the impermeable seed as compared with that of the permeable seed, and there is the peculiar regime of the oily seed, as in Rkinus, where the oil seems to take a vicarious part, much less water than usual being required for germination. In the case of the coats there is also the relative dryness of the seed- coverings in typical impermeable seeds ; but this influence is at times masked in the swelling process of seeds like those of Mucuna and Dioclea, where the kernel may begin to germinate whilst its coats are still partly dry. Then we have the diverse influence of the various textures of the coverings themselves, influences that are far more diverse than any that could be off^ered by the kernel, however much it may vary in consistence, as indicated by such terms as farinaceous, fleshy, horny, oily, etc. If we begin by comparing the swelling ratios of coats and The swelling kernel for all the plants named in the following table, we find kernel is that in just about two-thirds the kernel has the largest ratio. "^r"^gj.^than But if we except the fact that all the four species of Canavalia that of the and both the species of Iris are included in the smaller group where the coats have the largest ratio, there is little that is determinate in such an arrangement, species of the same genera being sometimes separated, as in the cases of Cassia and Erythrina^ whilst permeable and variable seeds are more or less divided between the two groups, the larger group being made up of the three types of seeds, impermeable, permeable, and variable. It is not therefore from such an arrangement that we should expect to be able to frame any definite infer- 14 coats. 2IO STUDIES IN SEEDS AND FRUITS ences in this direction, and accordingly I have not burdened my pages with it, though the reader can construct it for him- self from the table subjoined, where all the data are grouped in another fashion. In order to wrest their story from all these data, I will make use of another arrangement as given below. Here the seeds are arranged in two columns. In the first column the seeds are placed in order according to the amount of the swelling ratio of the coats, those with the greatest ratio being placed first. In the second they are placed according to the value of the swelling ratio of the kernel, those with the largest ratio coming first. There is nothing definite to be made out of the arrangement of the seeds in the first column, since permeable, variable, and impermeable seeds, whether or not we restrict ourselves to the Leguminosae, are fairly well distributed. On the other hand, if we turn to the second column, which contains the data for eleven impermeable, twenty-one variable, and thirteen permeable seeds, we find that all but one of the im- permeable seeds occur in the upper half of the column, where the swelling ratio for the kernel is the largest, and all but one of the permeable seeds in the lower half, where the ratio is smallest, the variable seeds being about equally distributed. We perceive the same contrast between the indications of the two columns when we compare the places occupied by seeds when both permeable and impermeable seeds occur in the same species. Thus with Entada polystachya and Dioclea reflexa^ the permeable and impermeable seeds come close together in the column where the seeds are arranged accord- ing to the amount of the swelling ratio of the coats, but lie far apart when the arrangement chosen, as in the second column, is that of the amount of the swelling ratio of the kernel. Evidently, therefore, although the question of permeability or of impermeability is largely shaping itself in its influence on the swelling ratio of seeds, it is an influence that chiefly aff'ects the kernel. The coats also respond, though in a less degree, to this influence, but their behaviour is often masked by THE SHRINKING AND SWELLING SEED 21 Table giving the Swelling Ratios of the Coats and Kernels OF Seeds, the Weight in the Resting State being taken as i, (The ratios are arranged in order, the largest being placed first. Permeable seeds are marked P., variable seeds V., and impermeable seeds I. Leguminous seeds are denoted by L.) Swelling ratio of the coats. Swelling ratio of the kernel. Iris fcetidissima P. 4-65 Guilandina bonducella . L. I. 3'34 Canavalia obtusifolia l! V. 3-20 Ipomcea tuba . I. 3'i4 Hura crepitans P. 3-08 Albizzia Lebbek . l'. V. 306 Canavalia ensiformis L. P. 2-87 LeucKna glauca L. I. 2-98 Cassia fistula . L. V. 2-87 Ipomcea pes-caprce . I. 2-83 Guilandina bonducella L. I. 27s Entada scandens l! I. 2'83 Cajanus indicus L. p. 2-59 ,, polystachya . L. I. 279 Erythrina velutina L. V. 2-59 Guilandina (species) L. I. 274 Tamarindus indica . L. V. 2'53 Calliandra Saman . L. V. 273 Iris Pseudacorus p. 2-50 Caasalpinia sepiaria . L. V. 271 Canavalia gladiata . l'. V. 2-48 Poinciana regia L. V. 2-69 Guilandina bonduc . L. I. 2-45 Erythrina indica L. V. 2-59 Adenanthera pavonina . L. I 2-29 Adenanthera pavonina . L. I. ^■57 Erythrina indica L. V. 2-28 Enterolobium cyclocarpum L. V. ^■57 Guilandina (species) L. I. 2-28 Ipomcea tuberosa . V. 256 Ipomcea tuberosa V. 2-27 Guilandina bonduc . L I. 2-49 Cassia grandis . l! V. 2"20 Cassia fistula . L. V. 2-44 Calliandra Saman . L. V. 2-i6 Entada polystachya. L. p. 2-39 Leuccena glauca L. I. 2-i6 Acacia Farnesiana . L. V. 2-38 Bauhinia (species) . L. V. 2-15 Erythrina velutina . L. V. 2'3I Ipomcea pes-caprse . I. 2-13 Abrus precatorius . L. V. 2-31 Pisum sativum (smooth) . L. p. 2'12 Dioclea reflexa L. I. 2-29 Abrus precatorius . L. V. 2*11 Erythrina corallodendron L. V. 2-25 Cassalpinia Sappan . L. V. 2'II Canavalia obtusifolia L. V. 2-24 Enterolobium cyclocarpum L. V. 2-05 C.-esalpinia Sappan . L. V. 2-23 Canavalia (species) . L. V. 2-05 Mucuna urens . L. I. 2-i6 Cassia marginata L. V. 2-03 Cassia marginata L. V. 2-13 Luffa acutangula p. 2-00 ,, grandis L. V. 2 -08 Entada scandens U I. I 99 Bauhinia (species) . L. V. 2-o8 Erythrina corallodendron L. V. 1-98 Phaseolus multiflorus L. p. 2 -08 Ipomcea tuba . I. 1-91 Iris fcetidissima p. 2-07 Poinciana regia L. V. 1-91 Faba vulgaris . U p. 2-05 Csesalpinia sepiaria . L. V. 1-91 Cajanus indicus L. p. 2-03 Entada polystachya . L. I. 1-91 Canavalia ensiformis L. p. 2-03 ,, ,, L. p. 1-89 ,, gladiata . L. V. 2-02 Dioclea reflexa L. p. ^'77 Tamarindus indica . L. V. 2-00 Acacia Farnesiana . L. V. 175 Pisum sativum (smooth) . L. p. 1-99 Dioclea reflexa L. I. 173 Thespesia populnea . V. I '95 Faba vulgaris . L. p. 173 Dioclea reflexa L. p. 1-92 Thespesia populnea . V. 1-68 Canavalia (species) . L. V. 1-91 Mucuna urens . L. I. I '55 Iris Pseudacorus p. 1-88 Albizzia Lebbek . L. V. I '47 Hura crepitans p. 1-68 Phaseolus multiflorus L. p. 1-44 Anona muricata p. I '57 Anona muricata ... p. 1-15 Luffa acutangula p. 1-49 Ricinus communis . p. I 'lO Ricinus communis . p. I '43 disturbing causes, and, speaking generally, they tend to play a neutral part in the matter. This relative independence of the 212 STUDIES IN SEEDS AND FRUITS coverings is well shown, not only in the general lack of correspondence of seeds in this respect, but in extreme cases like that of Hura crepitans^ where the seeds, on account of the large swelling ratio of the coats, stand nearly at the head of the first column, and on account of the small swelling ratio of the kernel are placed nearly at the bottom in the second column. The run of the data in the above table would therefore lead us to expect that whilst the coats of permeable, variable, and impermeable seeds would on the average possess swelling ratios not far apart from each other, the kernels of these three seed-types would differ markedly in this feature, the imper- meable seed displaying the largest, the permeable seed the smallest, and the variable seed a ratio intermediate in amount. This expectation is fulfilled in the following tabulated results of the table, whether for all the seeds or for the leguminous seeds only, though it is on the indications of the leguminous seed that we must mainly rely, since the disturbing influences of different ordinal characters are then eliminated. Tabulated Summary of the preceding Table showing the Average Swelling Ratios of the Coats and Kernels of Permeable, Variable, and Impermeable Seeds when preparing for Germination. Character of the seeds. Number of species tested. Swelling ratios, taking the resting seed as i. Coats. Kernel. Leguminous only Permeable Variable Impermeable 7 9 2 -06 2"20 2-12 2-07 2-35 2-69 Leguminous (35 species) and ( seeds of other orders (10 4 species) (^ Permeable Variable Impermeable 13 21 II 2-22 2-i8 2MO 1-89 2-34 274 Their con- The interesting indications afforded in the tabulated the water- summary just given become more important when we connect percentage, them with the water-contents of the resting seed as signified THE SHRINKING AND SWELLING SEED 213 by the loss of weight of the materials when exposed in the oven to a temperature of about 100° C. We should expect to find with all three types of seeds, where the swelling ratios for the coats are not far apart, that the water-percentages for the seed-coverings would not differ much in amount. We would also expect in the case of the kernels that where the swelling ratio is greatest, as with impermeable seeds, the water-percentage would be lowest ; that where the ratio is smallest, as with permeable seeds, the water-percentage would be largest ; and that where it is intermediate in value, as with variable seeds, the water-percentage would be also intermediate in amount. These indeed are the actual results that are represented in the tabulated summary to be now given as respecting seeds for which all the requisite data have been obtained. In order to avoid disturbing influences, the summary is restricted entirely to leguminous seeds, not to all leguminous seeds, but to exalbuminous seeds of that order. Tabulated Summary of Results showing the Average Relation BETWEEN the SWELLING RaTIOS OF THE COATS AND KERNELS OF Permeable, Variable, and Impermeable Seeds and the Water- percentages. (The swelling ratios are given in the general table in this chapter, and the water- percentages in the table illustrating the absorptive capacities of seeds in Chapter VI.) Character. Number. Coats. Kernel. SweUing ratio. Water- percentage. Swelling ratio. Water- percentage. Permeable Variable Impermeable . 4 4 6 2-04 2"I2 2-13 I3-4 x^-3 II'O 2-04 2-34 2-6i 14-9 11-4 8-3 The Permeable seeds are Canavalia ensiformis, Pisum sativum, Faba vulgaris, and Phaseolus niultijiorus. The Variable seeds are Abrtis precatorius, Ccesalpinia Sappan, Erythrina corallodendron, and E. indica. The Impermeable seeds are Adenanthera pavonina, Dioclea rejlexa, Entada scandens, Guilandina bonducella, G. bonduc, and Mucuna urens. 214 STUDIES IN SEEDS AND FRUITS We see here that whilst with the coats the differences in the swelUng ratios and water-percentages of all three types of seeds are small, the differences that do exist are in accord with the principles laid down in Chapters IV. and VI. Thus, the coats of permeable seeds have a smaller swelling ratio and a larger water-percentage than the coats of seeds that are more or less impermeable. But the difference is small, and it is to the kernels, where the contrast between the three types of seeds is pronounced, that we chiefly look for evidence in this direction. The determination of the amount of the swelling ratio by the quantity of water held by the seed is fully established by the behaviour of the kernels, the kernel of an impermeable seed holding on the average not much more than half the water held by the kernel of an average permeable seed, and possessing a much larger swelling ratio. In the first we have a water-percentage of 8*3, associated with a swelling ratio of 2*6 1. In the second the water-percentage amounts to I4'9, and the swelling ratio to 2*04. The effect of oil on the absorption of water by the kernels absorption of of the seeds preparing for germination is another point to be referred to. The effect is shown in the small swelling ratio of such kernels, a fact indicating a relatively small absorption of water. Such seeds have a regime of their own when swelling for germination. Ordinary permeable resting seeds, like those of Canava/ia, Faba^ Pisum, and Phaseolus^ dealt with in the tabulated summary given above, display an average swelling ratio for the kernel of 2*04, taking the weight of the kernel of the resting seed as i. But permeable seeds with more or less oily kernels exhibit a much smaller swelling ratio. Thus, the kernels of the seeds of Hura crepitans, Anona muricata, and Luffa acutangula possess sweUing ratios of i-68, 1*57, and 1-49 respectively ; whilst with Ricinus communis the swelling ratio for the entire seed is only i'33, as against 2*04 for typical entire permeable seeds belonging to the four leguminous genera above named. It will thus be seen that whilst an ordinary permeable leguminous seed doubles its weight by THE SHRINKING AND SWELLING SEED 215 absorbing water when preparing for germination, a Ricinus seed adds only one-third to its weight. It is probable that the small swelling ratios of many of the seeds mentioned in the table of the results obtained by Hoffmann and Nobbe in Chapter II. {Brassica, Raphanus, Cannabis^ Camelina^ Helianthus^ and Pinus) result from the oil in the kernels. That oil takes the place of water in the kernel is shown when we compare the water-percentages of the kernels of permeable resting seeds. In Hiira crepitans the water-contents of the kernel amount to 8-7 per cent. ; in Ricinus they form 6 per cent. ; whilst with the kernels of the four leguminous plants dealt with in the tabulated summary, the water-percentage ranges between 14 and 16. In the cases of certain Palm seeds oil plays a prominent part in determining the regime. The water-percentage is low when the spontaneous drying is complete. Thus in ELeis guineensis it is 9 per cent., and in the Coco-nut about 7 per cent., this small percentage of water being connected with the large amount of oil present. It will be appropriate here to make a few remarks on the The decrease decrease in the relative weight of the seed-coats as the seed weight of matures in the ripening fruit. Just as the growth of the poatTas'the pericarp is always in advance of the seed-growth, so the growth seed ripens, of the seed's coverings is always in advance of that of the kernel or seed proper. Thus with Guilandina bonducella the weight of the coats of the full-sized soft seed in the mature moist pod is about 61 per cent, of the total weight of the seed. A little earlier in the history of the seed's growth it is about 66 per cent., and earlier still, when the seed is not much over half size and the embryo incompletely formed, it is about 70 per cent. The behaviour of the seeds of the Horse-chestnut i^Msculus Hippocastanuni) as they ripen in the capsule well illustrate this progressive change. When the seed is only one-fourth grown the coats form 46 per cent, of the weight of the entire seed, when half size 42 per cent., and when full size 35 per cent. In this connection again let us take the data afforded by Peas {Pisum sativum) from the same set of plants. 2l6 STUDIES IN SEEDS AND FRUITS In young seeds weighing about 8 grains (when the coats make up one-fourth of the seed's diameter) the weight of the coverings is about 37 per cent, of that of the entire seed. In the mature soft seeds, weighing 12 grains, the proportional weight of the coats is reduced to 27 per cent. In all cases, one may add, there is a further decrease in the relative weight of the coverings in the resting seed. Taking the approximate data for the different stages, we obtain the following results : — The Relative Weight of the Coats of the Pre-resting Seed in DIFFERENT STAGES AND OF THE RESTING SeED, STATED AS A Percentage of the Weight of the Entire Seed. Pre-resting seed. Resting seed. i size. 1 size. 1 size. Full size. Guilandina bonducella yEsculus Hippocastanum . Pisum sativum . 46 70 42 66 ... 37 61 35 27 58 27 9 Not the least interesting feature of this discussion relating to the proportion of parts in the three conditions of the seed is that concerning the role taken by the embryo in albuminous seeds. I deal here only with types of those dicotyledonous seeds where the embryo has attained the maximum size permitted by the limits of the seed. The general subject of the size and other features of embryos is handled in Chapter XVIII. Here we deal only with those embryos, with large, flat, more or less foliaceous cotyledons, that occupy the length and breadth of the seed and lie usually between two slab-Hke masses of albumen. A few of the leguminous seeds used in this inquiry, viz. those of species of Bauhinia, Cassia^ and Poinciana, have these characters ; and for most of them I possess the requisite data for the embryo's weight-relation In the pre-resting, resting, and swollen conditions of the seed. In addition I have also materials for two euphorbiaceous plants possessing albuminous seeds with similar embryos, viz. Ricinus THE SHRINKING AND SWELLING SEED 217 communis and Hura crepitans. The seeds of the two plants just named are permeable, whilst those of the leguminous seeds are variable in this respect. If the kernels of the leguminous seeds experimented upon were, as is probable in some cases, ultra- dry, it would only be to a small degree (2 or 3 per cent.), and not sufficient to materially affect their behaviour. For practical purposes we may assume that the kernels of all these seeds attain a more or less stable weight in the resting state. A singular feature of these dicotyledonous embryos, and we may say the same of other similar embryos, such as those of the sapotaceous genera, Achras and Chrysophyllum^ is that they do not display in the resting seed the dried-up, shrivelled appearance so characteristic of monocotyledonous embryos in the naturally dried seeds of Palms, Cannas, etc. This is probably due in part to the difference in form and situation of the embryos in these two types of seeds. In the dicotyle- donous seed the embryo lies flat between two slabs of albumen and contracts with the kernel, leaving no unfilled space. With ordinary seeds of Palms the embryo lies in an elongated cavity, which it completely fills in the moist seed, but in the dry seed it has shrunk away from the walls of the cavity. At all events the water lost by the embryos of the moist seeds of both types when drying spontaneously does not differ much in amount, those of Palm seeds (Cocos, Areca, etc.) losing about 6G per cent, of their weight, and those of the leguminous embryos {Bauhinia, Cassia, Poinciana) 50 to 62 per cent. We have already dealt with the behaviour of the kernel Differentiat- in the shrinking and swelling of seeds. Here, then, we are JhfbSiavlour going to differentiate in the case of albuminous seeds between of the the two components of the kernel, the embryo and the the albumen, albumen. Allowing for the crudeness of the method, the indications of the table below are clear enough. The embryo and the albumen evidently go fairly well together in the three conditions of the seed, and consequently in the shrinking and swelling stages, each of them taking its proportionate share in the processes. The deviations in opposite directions suffi- 2l8 STUDIES IN SEEDS AND FRUITS ciently establish this point, and it would be unwise to place much stress on individual differences. Thus, to take the swelling ratios, although in Cassia and Rkinus this ratio for the embryo is rather greater than for the albumen, in Poinciana it is considerably less, and in Bauhinia it is about the same. Table showing the Behaviour of the Embryo in the three Con- ditions OF Albuminous Seeds, the Pre-resting, the Resting, and the Swollen State preceding Germination. (All are leguminous except the last.) Poinciana regia . Bauhinia (species) Cassia fistula ,, marginata , , grandis . Ricinus communis The relative weight of the embryo, taking the kernel Pre- resting. 47'9 42-8 27-1 Resting. 53*5 45-8 22*4 i8-3 147 5-6 Swollen. 45-8 I 46-1 I 24-2 j 19-1 j The shrinking and swelling ratios of the embryo and albumen, the first for the embryo, the second in parentheses for the albumen. The pre-resting state as 100. Pr. R. Sw. 100 45 103 (100 36 113) 100 50-5 106 (100 44-9 93) 100 38 100 (100 49 1.7) The resting state as i. Pr. 2"22 (278 1-98 (2-23 2-63 (2'03 Sw. 2"30 3-13) 2-10 2-07) 2-63 2-38) 2-23 2-IO) 2-30 2-05) 1-67 ••41) In order to obtain the data summarised in the foregoing tabular statement a considerable amount of work was done. For the study of the part played by the embryo in the shrinking and swelling regime of an albuminous seed of the type before de- scribed it was necessary to make use of a formula containing the elements for the requisite determinations, these elements being : (a) The weight of the resting seed ; (J?) The shrinking and swelling ratios of the entire seed ; {c) The relative weights of coats, albumen, and embryo in the three conditions of the seed, that of the un- contracted pre-resting seed, the contracted resting seed, and the swollen seed ready for germination. THE SHRINKING AND SWELLING SEED 219 The elements for the seeds under discussion are given in the following table. From its columns we can obtain the materials for constructing the complete regime of the shrink- ing and swelling seed in the case of three of the species of seeds there dealt with, the seeds of Hura crepitans being not here included, as they present special difficulties which are treated in Note 24 of the Appendix. In three other cases we have only the elements in part, namely, those for the determina- tion of the swelling regime. As an example of how the formula can be worked, I have appended to the table the complete regime, as thus indicated numerically, for the seeds of Cassia fistula. Whilst preparing the tabulated summary given above, I have had before me the regimes of all the seeds dealt with. Guided by the example given, the reader, by using the elements in the following table, can determine the regime for any of the other seeds. Albuminous Seeds. — Table showing the Proportional Weights of THE Coats, Albumen, and Embryo in the three Conditions OF the Seed. (From these data the complete regime can be constructed for the shrinking and swelling seed. ) Aver- Shrink- ing and swelling ratio, taking the weight of the Relative weights, taking the entire seed as loo. 1 "^\, weight of a resting seed Pre-resting seed. Resting seed. Seed swollen for germination. grains. resting seed 1 1 i c 1 1 as I. S .Q H fa < w < 23*5 27-0 41-0 <: 32-0 27-0 Poinciana regia . 10 2*3 4S"S 28-4 26-1 49 '5 Bauhinia (species] 4 2-1 2VO 44-0 33"o 23'S 41-5 350 24-0 41 -o 3S"o Cassia fistula 4 2*5 26-, S3 7 20-0 1 IS'O 66 -o i9'o 17-2 62-8 20'0 ,, marginata 10 2*1 29*0 58 . i3"o 28-3 58-0 137 ,, grandis . 9 2'II ... 25-0 64*0 II -o 26'0 62*0 I2-0 Ricinus com munis 3 i'33 28-0 68-0 4-0 23-0 72-0 S-o Hura crepitans . 20 2-1 I S7-0 38-2 4-8 30-0 61-5 8-5 44-0 48-5 7-S The seed of Hura crepitans presents special difficulties, its regime being discussed in Note 24 of the Appendix. 220 STUDIES IN SEEDS AND FRUITS Cassia fistula. — The Shrinking and Swelling Regime of a Seed of Cassia fistula in Illustration of the Mode of employing the Data in the above Table as described on the preceding Page. Weights in grains. Shrinking and swelling ratios, taking the pre-resting state as 100. Pre-resting. Resting. Swollen for germination. Coats . Albumen Embryo Entire . 2-63 5-37 2-00 o*6o 2-64 076 172 6-28 2 '00 P. R. S. 100 23 66 100 49 117 100 38 100 lo-oo 4-00 IO"0O 100 40 100 SUMMARY (i) In order to illustrate the shrinking and swelling regime of a seed, a comparison is made of the relative weights of the coats and kernel in the three conditions : the pre-resting, the resting, and the swollen state preceding germination (p. 187). (2) In connection with the requisite data the proportional weights of the coats and kernel of the resting seed are first dealt with, results being given for thirty-nine species of leguminous seeds and for forty- three species belonging to twenty other orders, including Anonaceae, Convolvulaceae, Euphorbiaceae, Malvaceae, Sapindaceae, Sapotaceae, etc. For convenience only one value is generally employed in the discussion, that of the relative weight of the coverings, which is termed the " seed- coat ratio" (p. 188). (3) Respecting the great range in the proportional weight of the coats of the resting seed, it is shown that although the range is from 5 to 69 per cent, of the seed's weight, nearly the whole of it is presented by the leguminous seeds, and that the maximum of the range is but slightly extended in the case of seeds with abundant hairs (p. 189). (4) By grouping the results for the eighty-two species, it is found that the average weight of the coats of the resting seed is about 30 per cent, of the total weight, rather below for leguminous seeds and rather above for seeds of other orders. Excluding very small seeds, " size " has little or nothing to do with the ratio (p. 191). (5) As regards the constancy of the seed-coat ratio within the THE SHRINKING AND SWELLING SEED 221 limits of a species, it is shown that the relative weight is sufficiently stable, although it may vary considerably within a genus (p. 192). (6) With reference to the influence of appendages, such as hairs and wings, on the proportional weight of the coverings of a resting seed, that of hairs is first dealt with. Although ordinary pubescence adds but little to the weight of a seed, a copious covering of long hairs may increase it by 10 per cent. ; whilst in the case of the great hair- development occasionally found with seeds, as in certain Asclepiads and in some malvaceous genera {Gossypium\ the weight of the hairs may amount to one-fourth or even to nearly a half of the entire weight (P- 193)- (7) Coming to the influence of wings on the weight of the resting seed, it is shown in the case of Swietenia, Moringa^ and Tecoma that the wing or wings may make up between 4 and 14 per cent, of the total weight. With ordinary " margined " seeds there would be but little addition to the seed's weight (p. 195). (8) Discussing the function of wings in the resting seed, the author points out that they serve only an accidental purpose in aiding dispersal, and that they have no biological significance except in the actively functioning soft seed of the living fruit. Withered leaves and dry resting seeds stand in this respect in the same category. The possession of wings, he shows, does not always materially assist dispersal. In the case of Pine seeds and the light seeds of Tecoma stans^ wind may carry them a long distance ; but Mahogany seeds are usually only carried a short way ; whilst the seeds of Moringa pterygosperma are not much assisted in this respect (p. 196). (9) With regard to wings in resting seeds, the author adopts the standpoint taken by Dr Goebel respecting the parachute-apparatus of dry fruits, namely, that although at times serviceable for their dispersal, they were not originally developed for this end, but performed quite a different function in the moist pre-resting seed. As regards winged Mahogany seeds, the author contrasts the heavy, moist, flabby seeds of the living, closed capsule with the dry, crisp seeds — only one-fourth of their moist weight — of the dehiscing dead or dying fruit ; and he regards the function of the wing in a seed in the first-named condition as mainly concerned with the absorption and storage of water, the absorbing surface being increased threefold and the quantity of water for the seed's wants greatly augmented (p. 197). (10) The proportional weight of the coverings in the two other conditions of the seed is then considered, and it is pointed out that, knowing the relative weight of parts (coats and kernel) in all the three conditions of the seed (pre-resting, resting, and swollen for germination), we possess the requisite data for determining the regime of the shrinking and swelling seed (p. 198). 222 STUDIES IN SEEDS AND FRUITS (I I) Although, as established in Chapter II., the seed when swelling for germination regains approximately the water lost in the shrinking process, or, in other words, returns to about its original weight in the pre-resting state, it is shown that this result is in a sense accidental, since the coats do not win back all the water they gave up in the shrinking stage, whilst the kernel takes up as a rule considerably more, the two results going to counterbalance each other in determining the ultimate swelling weight (p. 199). (12) The materials for constructing the shrinking and swelling regime for a considerable number of seeds are then tabulated and explained. Commencing with a typical impermeable and a typical permeable leguminous seed, it is shown that in both cases the coverings of the swelling seed fail to regain all the water lost in the shrinking process, the deficiency being greatest with the permeable seed. On the other hand, the kernel in both cases ultimately holds more water than in the pre-resting state. The same principle, it is pointed out, applies generally to leguminous seeds ; but by the employment of the additional materials we are enabled to see a little more into the average details of the processes. As a general rule, although the coats of the permeable seed shrink more than those of the impermeable seed, both double their weight in the swelling process ; whilst the kernel of the impermeable seed shrinks more and increases its weight threefold as compared with the kernel of the permeable seed, which shrinks less and only doubles its weight when swelling for germination (p. 200). (13) It becomes apparent from all the tabulated results that the resting state leaves its impress on the seed swelling for germination, especially with regard to the coats. This is also evident in the contrast in appearance between the drier, tougher, and relatively unyielding coverings of the seed swollen for germination and the moister, softer, and more yielding coats of the pre-resting seed. But it becomes particularly noteworthy when we find that the seed on the point of germinating is a little smaller than the so-called unripe or pre-resting seed (p. 204). (14) The result is that conditions of strain arise within the swelling seed. Whilst with the pre-resting seed the kernel lies easily within its coverings, with the swollen seed ready for germination the kernel holds considerably more water than in the pre-resting state and lies constramed within its tightened, unyielding coats. Examples of the strain within the seed thus produced are given, and it is shown that within its tightened envelopes lies a kernel on the average more than 20 per cent, heavier through water-absorption than in the pre-resting state. The result is the rupture of the coverings, the process being purely physical and not necessarily succeeded by germination (p. 205). (15) That the seeds of other orders may possess a regime similar to THE SHRINKING AND SWELLING SEED 223 that of leguminous seeds is brought out for purposes of illustration in the case of the seeds of the two species of Ipomoea^ which represent the convolvulaceous regime (p. 207). (16) The special difficulty presented by the seeds of Hura crepitans is next referred to. These seeds follow the fashion of some indehiscent fruits, like those of the Oak [Quercus] and of the Coco-nut, the seed-coverings losing considerably in weight before the kernel attains its mature size (p. 208). (17) The author then utilises a large amount of data obtained for seeds where the swelling phenomena of the coats and kernel were alone observed. The results for forty-four species are thus employed, of which all but ten are leguminous ; and it is remarked that although in about two-thirds the kernel has a larger swelling ratio than the coats, there is no very determinate result in this direction, genera behaving sometimes consistently and at other times being divided in this respect Cp. 209). (18) In order to wrest their story, another arrangement of these "swelling" data is adopted, from which it appears that whilst with the coats the swelling ratios, as respects impermeable, variable, and permeable seeds, are not far apart, with the kernels the ratios for all three types of seeds differ markedly from each other, the impermeable seed showing the largest, the permeable seed the smallest, and the variable seed a ratio intermediate in amount (p. 210). (19) It is then elicited that this behaviour of the coats and the kernel in these three seed-types is to be connected with the water- percentage of the resting seed, but less in the case of the coats, where the differences are small, than with the kernels, where the differences are large, the kernel of an impermeable seed holding on the average not much more than half the water held by the kernel of a permeable seed and possessing a much greater swelling ratio. With the kernels of variable seeds, where the swelling ratio is intermediate in extent, the water-percentage is also intermediate in amount (p. 212). (20) Two other points are then referred to, the first being that in seeds with oily kernels, where the unusually low water-percentage indicates that the oil supplies the deficiency, the swelling ratio, as in the case of the seeds of Ricinus, is also unusually small. The second is the decrease in the relative weight of the coats as the seed ripens ; and in this connection it is remarked that just as with fruits the growth of the pericarp is always in advance of the seed-growth, so with seeds the growth of the coats is always in advance of the kernel (p. 214). (21) Having dealt with the shrinking and swelling processes of the kernel, we now differentiate in the case of dicotyledonous albuminous seeds between the two components of the kernel, the embryo and the albumen. The role of the embryo in this connection is now studied. For this purpose only seeds with embryos possessing large, flat cotyledons. 224 STUDIES IN SEEDS AND FRUITS and occupying the length and breadth of the seed, are employed, seeds such as those of Poinciana, Bauhinia^ Cassia.^ Ricinus^ and Hura. It is shown that the embryo and the albumen go fairly well together in the three conditions of the seed, and consequently in the shrinking and sweUing stages, each of them taking its appropriate share in the processes (p. 217). (22) Such a statement implies that the author, with the aid of the balance, has been able to construct a complete regime for the shrinking and sweUing stages of a dicotyledonous albuminous seed, not merely for the seed in its entirety, but independently for all its parts : the coats, the albumen, and the embryo. The materials for such determinations are tabulated, and the seed of Cassia fistula is taken to illustrate the employment of the data (p. 218). CHAPTER X THE FATE OF SEEDS AS INDICATED BY THE BALANCE We realise from the results of the recent researches of Becquerel Long years and others that long years are needed for the satisfactory study for the study of the latent life of seeds, a period, to employ the words of J^gof seeds this French investigator, far longer than the time during which the seeds preserve their germinative powers. This is as true for the indications of the balance as it is for those of any other method of physical or chemical research. It is necessary that the life of the inquirer should extend beyond that of the seed he is studying ; and too often, as Jodin aptly observes, the savant who commences such an experiment will never know the results. This is the spirit in which we should approach such a difficult subject as the life of the seed, one of the deepest and most mysterious problems that can occupy the wits of man. A study, though unfinished, is not altogether incomplete if we leave it so that others can take it up. There- fore, although the results given in this chapter are derived from experiments two to four and a half years in length, they represent only the beginning of a long series of experiments which, it may be, someone else may continue after my term of life has ended. Yet we have here a record of the start and sufficient indications to enable us to look a little ahead in the matter. After seeds have entered upon the resting stage and have quite completed the process of drying in air, four possibilities present themselves. As years go on they may either lose or 225 15 226 STUDIES IN SEEDS AND FRUITS gain in weight very slowly, or they may remain absolutely un- changed, or they may acquire a stable weight subject only to hygroscopic variation about a mean. A period of a few weeks or of two or three months is usually sufficient for the completion of the process of drying in free air. For most permeable and impermeable seeds this is quite enough, as is shown in Note 12 of the Appendix, though longer periods may be occasionally required, as in the case of Mammea americana^ specially referred to in Note 6, where the drying was continued for about a year. Speaking generally, my data show that impermeable seeds can retain their weight unchanged for at least four years, excluding such small variations, amounting in the case of Entada scandens to only about 2W0 °^ ^^^ ^°^^^ weight, which are mainly instrumental in their nature. On the other hand, during the same period permeable seeds, assuming what we may call "the hygrometric state," display a variation in weight of 2 or 3 per cent, around a fairly constant mean in response to the varying humidity of the atmosphere. Variable seeds again behave in an intermediate but change- able fashion, the result of their varying proportions of permeable and impermeable seeds, sometimes approaching one type, sometimes the other, but usually holding a half- way position and exhibiting a small hygroscopic range of 1 per cent, or less. Coming first to the impermeable seed, we are justified, I think, in believing that as long as it preserves the same weight and is non- hygroscopic it retains its vitality. It enjoys complete immunity from the dangers of the hygroscopic reaction, which, if continued through years, must, as M. Jodin observes, ultimately induce molecular changes and lead to the loss of germinative capacity in the case of the permeable seed. The impermeable seed, on the contrary, as long as it is true to its original weight, is exposed to no such risk. However, Professor Ewart, in his paper on " The Longevity of Seeds," contends that " even when perfectly inert a macro- FATE OF SEEDS INDICATED BY BALANCE 227 biotic seed (that is to say, a long-living seed with more or less impermeable coats) is subject to slow and gradual molecular changes and rearrangements, such as take place in glass or wood in the progress of centuries, and that these chano-es cannot take place in the contents of the seed without destroy- ing the molecular arrangements and groupings requisite for the restoration of life." Admitting for argument's sake the force of this contention, we should be compelled to credit such seeds with the capacity of retaining their germinative powers for many centuries. But this seems to me to be hardly a correct comparison. Glass and wood are at all events exposed to atmospheric influences, and the last named in particular would be continu- ously subjected to the hygroscopic reaction. The feature of the non-hygroscopic impermeable seed is that as long as its coverings are intact it remains hermetically sealed up and beyond the influence of atmospheric conditions. If we allow that such a seed may be expected to retain its germinative powers as long as it retains its weight, then the question left to determine is concerned with the duration of its capacity of preserving its weight unchanged ; and in illustration of this point there are appended the results of some experiments on impermeable seeds extending over a period of from two to four years. It has already been established in Chapters IV. and VI. that impermeable seeds weighed from day to day give no indication of any response to the varying hygrometric states of the air. The data in the table below given make it plain that these seeds preserve their non-hygroscopic behaviour from year to year. Time alone will show how long they will remain in this irresponsive condition as far as any reaction with their surroundings is concerned. Although, as pointed out in a previous chapter, the experiment truest to nature would consist in burying the seed deep in the soil in a dry climate, still, these tests in free air go to show that impermeable seeds may remain for many years in this inert condition. 228 STUDIES IN SEEDS ,AND FRUITS The Weights of Impermeable Seeds during Periods of FROM TWO TO FOUR YeARS. Number of seeds. 1 Length of ex- periment in years. Changes in weight stated in grains. Proportion of the change stated as a fraction of the total weight of the seed. Entada scandens . A Guilandina bonducella Adenanthera pavonina Ipomoea pes - caprae (with hairs) .... 22 20 4i 44 4i 3h i 3 301* 8-302*0 266* 7-266*9 241* 1-241*3 406* 5-406*6 306* 6-306*8 104* 8-104*9 41*15- 4i"2 TsVcT ITSS j^oie. — The observations were made at irregular intervals every few months. The first three seeds of Entada scandens were carried to Jamaica and kept there some months during the first year, which accounts for their rather larger variation of 0*2 grain ; but in the last three years in England the weighings only varied 0*1 grain. A pebble of quartz would probably have behaved in the same way ; and evidently the cause of the variation is largely instrumental. This is also true of Guilandina bondncella. The samples of the seeds of Adenanthera pavonina and Ipomcea pes-capm were small in weight ; but the results give the same general indications. In connection with the last- named it is shown on page 169 that its pubescent hairs have a very slight disturbing effect. However, although this is the rule, cases of failure are not infrequent, and their aberrant behaviour is at once detected by the balance. Some slight defect in the coats, due probably to some imperfection in the shrinking process, gives time its opportunity ; and the seed within, brought into relation with its surroundings, responds to the changing atmospheric conditions, first slowly and then more rapidly, until it assumes the role and the limited life-duration of a permeable hygro- scopic seed. The manner in which this change of state is carried out has been described in Chapters IV. and VI . in connection with the results of puncturing or filing the seed- coverings. The seed gradually gains weight in the course of months, taking up from the air the water which it previously lacked ; and, as is shown in the case of filed seeds of Guilandina bonducella^ it may retain its germi native capacity for two years after it has virtually lost the protection of its coats, an event, FATE OF SEEDS INDICATED BY BALANCE 229 however, only possible when kept in a dry room. A punctured seed in nature would soon fall a victim to the attacks of mould and insects, unless it germinated quickly. The failures amongst the impermeable seeds are, as one The failures might expect, very instructive. We can trace the slow LwrseeT" progress of the loss of impermeability in seeds fathered bv areinstruc- ir 11T1 r ^^^^ "^ P''^- ourselves trom the plant. In the course or an experiment senting-per- covering years, a single seed in a sample gradually begins ^quaiif/by to fail, a change unerringly indicated in the balance by the '^^'^"•*- increase of weight. Two seeds of Entada scandens^ the shrinking process of which I had watched after collecting them in the immature, uncontracted state, preserved their weight unchanged for a few weeks, and then began slowly to gain weight, until at the end of a year each had added 3 per cent, to its weight. After this they behaved hygroscopically, like ordinary permeable seeds. The explanation was supplied in the development of fine cracks in the cuticle. In another case three seeds, also of Entada scandens^ which together weighed 1080 grains, gained i grain during the first year. By subsequently weighing them separately the culprit was discovered, two of the seeds remaining quite unchanged in weight. The cause of failure in one of the seeds lay in an imperfection of the cuticle. The same thing occurred during an experiment on six seeds of Guilandina bonducella weighing in all about 200 grains. A gain in weight of i per cent. during the first year led me to weigh them separately, and it was thus discovered that this increase was due entirely to one seed, which on close inspection showed defects in the outer coat. The above experiences of faulty impermeable seeds show that nature fails at times in endowing a seed with imperme- ability, but the suggestive implication is that in such failures permeability is presented to us as a quality by default. Although the indications supplied by my numerous Prof. Ewart experiments seem clear and unmistakable with reference to the °he imperme- ultimate fate of the impermeable seed, the views held by ^'''^ ^^^^- 230 STUDIES IN SEEDS AND FRUITS Professor Ewart respecting the matter are opposed to the position adopted in this work. It has been established by my observations that the impermeable seed, more especially the leguminous seedj holds much less water in the entire state than when broken up and exposed to the air. The process of the change is exhibited when the seed develops some defect in its coverings or when we imitate such a defect by puncturing the coats. The seed then slowly gains in weight by abstracting moisture from the air, until in time it assumes the stable hygrometric condition of an ordinary permeable seed. Now, Professor Ewart takes a very different position. In the case of Acacia seeds, which are notable for their imperme- ability, and often contain less than 5 to 8 per cent, of water, he writes that " such seeds when preserved in a dry atmosphere seem to steadily lose water, until ultimately as dry as if kept in a desiccator " {Proceedings of the Royal Society of Victoria, vol. xxi. p. 199, 1908). This is based on the fact that old seeds lost less weight in the oven than fresh seeds. Fresh air-dried Acacia seeds, he says, contain 5 to 14 per cent, of moisture ; seeds ten to twenty years old contain from i to 3 per cent. ; whilst fifty-year-old seeds hold less than i per cent., losing only o*7 per cent, of their weight after the prolonged exposure of a day to a temperature of 110° C. Fine capillary glass tubes, he writes, show a greater loss of weight than this, and hence he infers that " old, dry, cuticularised macrobiotic seeds become drier than corresponding inorganic material." Professor Ewart thus observes that in course of time Acacia seeds become as dry as if kept in a desiccator, and implies that they behave like inorganic material. This raises the question as to the behaviour of these two types of substances after desiccation. Seeds that have been exposed to such a low temperature that the free water has been completely expelled fall to powder when struck with a hammer and quickly absorb the hygrometric water from the air (P. Becquerel, in Annales des Sciences Naturelles, Botanique, tome v., 1907). This is but an extreme form of what occurs in the kernel of an FATE OF SEEDS INDICATED BY BALANCE 231 impermeable seed when its coats are broken. Being normally in a state of partial desiccation, it at once begins to supply the deficiency by absorbing moisture from the air. From inorganic substances like glass, we should look for no such behaviour. They have little or no water to lose under desiccating conditions, and in consequence have little or no water to regain. This matter is discussed in Chapters VII. and VIII. It will be there seen that Berthelot lays stress on the contrast in behaviour between inorganic substances (like porcelain and metals) and plant-tissues, the first drying com- pletely in air, whilst the second do not. There is nothing to lead us to expect that the kernels of seeds of any type would ever lose their capacity for behaving hygroscopically and become non-hygroscopic, inert substances like porcelain and metals. If this were the case, Berthelot's principle of reversibility would lose much of its significance. It must, however, be admitted that the semi- stony consistence of the kernels of some old permeable seeds would at first sight seem to justify such a belief. At one time I thought that the seeds of Msculus Hippocastanum (Horse- chestnut) and of the Acorn {Quercus Robur) would in time assume some of the characters and behaviour of inorganic material, such as chlorite and opal, as regards the diminished water-contents and the small hygroscopic reaction ; but this proved to be incorrect. As regards hygroscopicity, reference has already been made to the effect of time on the behaviour of the seeds of the Horse-chestnut in their coats, the results of the experiments being tabulated in Chapter VII. We saw there that, whether six months or thirty months old, the hygroscopic reaction was much about the same, namely, 2*o to 2*4 per cent. Whilst writing these remarks I have conducted another experiment on the same seeds three or four years old ; but in this case the seeds were first bared of their coverings. The three-year-old seeds give a hygroscopic range of 2-4 per cent., and the four-year-old seeds of 3*5 per cent. So also with reference to the bared seeds of the Acorn. I find 232 STUDIES IN SEEDS AND FRUITS that two-year-old seeds display a range of 2-o per cent., and three-year-old seeds i'6 per cent. Respecting the diminution of the water-contents in time in the case of the kernels of permeable seeds, I will first take those of Quercus Robur. Here the kernel in the course of time assumes a chocolate-coloured, semi-waxy appearance and almost a stony hardness, changes which begin at the periphery. Of kernels fourteen months old about a third were completely affected. In another third half of the kernel had undergone the change ; whilst in the remaining third the transformation was either in its early stage or scarcely noticeable. Here a mixed sample of the kernels proved to contain 11-3 per cent. of water. Every seed in samples two years old and more displayed the change completed, the water-percentage in kernels two years old being 13*2, and in kernels three years old 12*2. So with the old seeds of Msculus Hippocastanum^ it was found that at least for the first three or four years there was no diminution in the water-contents. Seeds bared of their coverings exhibited a water-percentage of 13-6 when eleven months old, 14-4 when twenty-six months old, 12*6 when three years old, and 13*8 after being kept for four years. These seeds, it may be added, do not discolour with time like the seeds of Quercus Robur. As is well known, the seeds of the Oak only retain their germinative capacity for a few months. With those of the Horse-chestnut the period, as shown in Chapter XVIII., is still less. Now and then one comes upon old seeds that are as hard and seemingly as dry as a stone, seeds that might almost be taken for fossils, yet in the oven they prove to contain 10 or 12 per cent, of water. The old seeds of Grias cauliflora, the Anchovy tree of Jamaica, a myrtaceous tree that flourishes in the lower courses of rivers in the West Indies, offer con- spicuous examples. A very remarkable alteration in texture takes place in the seeds after their death, when the conditions are dry, as on beaches, where they are stranded in quantities. FATE OF SEEDS INDICATED BY BALANCE 233 The fleshy, though tough living seed, has the singular structure presented by the seeds of some other myrtaceous trees, such as Barringtonia^ where an enlarged hypocotyl, invested by a thick rind and forming the storehouse of the food-reserve, constitutes the seed. As the seed dries it becomes as hard as a stone, and on section displays the appearance of a fossil fruit, these stone-like seeds being usually i to i^ inches long. Yet on testing the water-contents of one of these old seeds, which must have been at least six or seven years old, and had been almost five years in my possession, I obtained a result of 1 2 per cent., almost all the water lost in the oven being subsequently regained from the air in the course of a few weeks. I may direct the attention of the botanist engaged in microscopical and chemical research to these remarkable changes in the seeds of this plant. Sometimes a portion of the seed rots, whilst the other portions experience the change ; and when such a seed is found in the early stage of transformation, a very puzzling structure is displayed. Similar questions might be raised with reference to the condition of old seeds of Barringtonia speciosa^ which possess the structural features above described in the case of the seed of the Anchovy tree. The seeds do not become quite so hard with time, but are sufficiently altered to cause one to look twice in order to be assured that one is not dealing with some non-vegetable substance. Yet old kernels collected three years before lost 10 per cent, of their weight in the oven, and in a few weeks returned almost to their original weight by replacing the lost water with moisture abstracted from the air. There is, however, an indication in my experiments on impermeable seeds that might seem to point in the direction of the change in seeds as interpreted from Professor Ewart's point of view. In Chapter IV. it is shown in the case of bared kernels of Guilandina bonducella that whilst freshly bared Guilandina kernels added 13-4 per cent, to their weight in five days, this excess was reduced to 7*3 per cent, in three months, and to 3 or 4 per cent, after a year, an excess which was retained bonducella. 234 STUDIES IN SEEDS AND FRUITS during the next twelve months, when the experiment ended. I did not finally test the water-contents, but the fact that in the last year the seeds exhibited a hygroscopic variation of 1*4 per cent, indicates that they must still have held a fair amount of water. But against this indication of Professor Ewart's view must be placed the indications of another experiment on the same seeds, the results of which are also tabulated in Chapter IV, Here I merely filed through the impervious shell, and thus enabled the air to have access to its ultra-dry kernel within. During four months the seed slowly added to its weight as much as 1 1 per cent. ; but at the end of the experiment, which covered two years, the seed was still about 8 per cent, heavier than before its shell was filed through. However, a solution of the difficulty seems to be oflFered by Professor Ewart's remark, when discussing the drying in time of Acacia seeds, that " it is as though the cuticle allowed traces of water to escape externally, but none to enter " (ibid.^ p. 199). From this I am inclined to think that his oven- experiments for testing the water-contents were carried out on the seeds whilst protected by their hard, impervious coats. If so, this explains the whole matter. In my experiments on the impermeable seeds of Entada scandens and Guilandina bonducella, which are described towards the close of Chapter VI., I show that when exposed both in the entire condition and in the broken condition to a temperature of 100° to 110° C. for two hours, the seeds in their hard coverings lost only about a fourth of the water lost by the seeds no longer protected by their coats. It was also elicited that the seeds in their cover- ings subsequently made little or no attempt to regain the moisture from the air and doggedly maintained their imper- meability. Thus Professor Ewart's surmise that the coats of a seed allow the water to escape, but inhibit re-absorption, is certainly applicable to the behaviour of an impermeable seed during and after the oven test. But this does not reproduce the conditions of drying at ordinary temperatures in the course of years. FATE OF SEEDS INDICATED BY BALANCE 235 Nearly all his results relating to the diminution of the water-contents of Acacia seeds when kept for years can be explained on the assumption that the oven-experiments were made on seeds in their coverings. The fact that the fresh air-dried seeds sometimes hold as much as 14 per cent, of water sufficiently indicates that the shrinking process was not always complete, and that, as his general table also indicates, they included some permeable seeds, thus accounting for the greater loss of water in the oven. I am inclined to think that the low water-percentages of his old Acacia seeds were due to the experiments being carried out on seeds in their impervious coverings. The protection a seed receives from its coats in the oven was not only exhibited in other experiments on impermeable seeds, as in the case of those on the seeds of Canavalia obtusi- folia discussed at the close of Chapter VI., but it was well displayed by the differences in behaviour of permeable seeds, when subjected to the oven test in the entire and in the divided condition, as shown in the results tabulated towards the end of the same chapter. After an exposure of two hours to a temperature of 100° to 110° C, the entire seeds of Pisum sativum^ Faba vulgaris^ and Phaseolus multiflorus lost i o or 11 per cent, of their weight, whilst seeds of the same set which had been cut across lost 14 or 15 per cent. In the first case the seeds were completely covered by their coats, and in the last case only in part. The seeds were in all cases eight to ten months old. It is on these grounds that I venture to differ from Professor Ewart. The future inquirer must decide between us. The hygroscopic variation offers a special difficulty in the Permeable examination of the effect of time on the weight of a permeable fn the same seed. Notwithstanding that my experiments extend over g^feTrinl periods of three to four years, it is only safe to say at present the first three that the seeds are still in the hygrometric state which they assumed when first entering the resting stage, exhibiting a variation about a mean between the several weighings of about 236 STUDIES IN SEEDS AND FRUITS 2 or 3 per cent., being lightest in the summer and heaviest in the winter months. Variable seeds, where there is a mixture of permeable with impermeable seeds, display about half this variation, namely, i per cent. But, as has been said, the disturbing influence of the hygroscopic reaction is a great obstacle in detecting small differences in such ex- periments. If there has been a change, it has certainly not involved any increase in the average weight. On the contrary, the indications, such as they are, point slightly in the direction of a diminution ; but it will not be possible to obtain definite results until the experiments have been greatly prolonged and one is able to eliminate the effect of the hygroscopic reaction by comparing the averages for groups of years. In such experiments on hygroscopic seeds it is necessary that they should always be kept in the same room. So sensitive were my seeds to change that a transference from one room to another was sufficient to cause a variation of i or even of i per cent. Several years ago MM. Van Tieghem and Bonnier published some interesting results in a paper in the Bulletin de la Societe Botanique de France (tome xxix. 1882), some of which bear directly on the effect of time on the weight of permeable seeds. They found that after two years in free air, peas had gained about \\ per cent, in weight, haricots about 2 per cent., seeds of a species of Vkia about i per cent., and Ricinus seeds rather more, but the actual increase in the last case is not given. From my own observations it would appear that this rise in weight was within the ordinary hygroscopic range, and that if weighed at another season all the seeds might have displayed a decrease instead of an increase in their weight. As indicated in Note 25 of the Appendix, the variation in weight of completely air-dried seeds of Pisum sativum^ Faha 'vulgaris^ and Phaseolus multiflorus during a period of nearly fifteen months ranged from 2*6 to 3*6 per cent, of the seed's weight. The variation in weight is also there given for the seeds of Ricinus communis in the case of experiments FATE OF SEEDS INDICATED BY BALANCE 237 extending over three years and nearly two years, the amount of the change being n to i-6 per cent, of the total weight. In all cases the seeds were heavier at the beginning than at the end, a result due to their being kept at first in a damp room. The variation is the normal hygroscopic reaction. There is no sign of any permanent increase in weight. If one was guided only by the run of the figures and disregarded the conditions of the experiment, one might infer that these seeds lose weight as they get older. The permeable and variable seeds now under observation The author's with the object of testing the influence of time on their ing^erp^fi-' weight belong to nearly thirty genera, of which one-third i"ents. are leguminous, and include Ahrus^ Achras, Anona^ Casalpinia^ Canavalia^ Citrus, Datura, Erythrina, Faba, Hura, Iris, Luffa, Morinda, Phaseolus, Pisum, Ricinus, Thespesia, etc. In conducting all experiments of this kind it is, as already Whilst with remarked, a matter of necessity that the seed should have s^miarfruits acquired a stable weight, either as an impermeable seed when, ^^^ ^^-^^ with regard to the hygroscopic reaction, it is absolutely inert, complete the or as a permeable seed when it displays a small variation on cesrfefo^re either side of a mean. It is not merely requisite to employ natufaiiy de- resting seeds for the purpose, but they must be resting seeds tached, the with a stable weight. The seeds of the berry and of the legume moist, fleshy when first liberated by the opening or by the decay of the half of*their fruit are, as described in subsequent chapters, in very different ^^jj^* ^^*^'" stages of drying. In both cases the seed may in colour, liberated hardness, and other features have the appearance of a normal resting seed. Yet if it belongs to a berry it has still to lose 40 or 50 per cent, of its weight by drying in air ; whilst if it belongs to a legume it will have already practically completed the drying process. It is true, as shown below, that legumin- ous seeds usually lose slightly in weight after being gathered from the dehiscing pod ; but here we must often be anticipating nature a little. I have not many observations bearing on this point, but they are sufficient for the purpose of illustration. With regard first to the impermeable seeds of legumes, my 238 STUDIES IN SEEDS AND FRUITS observations indicate in the case of those of Entada scandens that resting seeds, weighing about 400 grains when removed from the pod, may lose about 2 grains during the next few weeks whilst exposed to the free air. Most of this small loss is probably connected with surface moisture, since Entada pods break up into closed joints from which the seed has to be removed. In the cases of the pod of Guilandina bonducella^ which dehisces usually in full exposure to the sun, it would be unhkely that seeds after lying in the gaping pod under such conditions would not have completed the drying process. Coming to permeable seeds of dehiscing pods, I will cite the case of those of Canavalia ensiformis^ which in ten days after collection from the pods lost about 8 per cent, of their weight and then entered the stable hygrometric state. Then, again, the seeds of Casalpinia sepiaria^ which are in some cases permeable and in others impermeable, lost about 10 per cent, of their weight after being gathered from the opening pod. In both these instances of permeable seeds it is highly probable that nature was to some degree anticipated, and that, left alone, they would have completed their drying before detachment from the pod. However, the loss of weight experienced by resting seeds or seed-like fruits after they have been collected is often far more considerable than in the leguminous seeds above cited. As Nobbe puts it (p. 382), the process here involved is what the agriculturist would term " sweating " in the case of wheat, and what the forester would call " airing " when gather- ing acorns, chestnuts, and maple fruits for winter storage. My experiments on Acorns {Quercus Robur) and on the seeds of the Horse-chestnut {^/Esculus Hippo castanuni) indicate that they have still much moisture to lose when they are first detached in the " browned " condition from the tree. The Horse- chestnut seed, as it lies on the ground freshly liberated from its fruit, has still to surrender one-third of its weight in moisture to the air before its drying process is complete. The Acorn also, when in the early stage of browning it falls FATE OF SEEDS INDICATED BY BALANCE 239 from the cupule, has yet one-fourth of its weight to lose during its drying. Though leguminous seeds as they escape The after- from the withered pod have practically completed the drying '^^'^S of process, it is very different with the seeds of watery or fleshy fleshy or fruits when they are freed, as must often happen, from the ^^^^""^ ''"^^^' moist fruit. As the result of a number of observations I found that after removal from the ripe fruit, and before enter- ing upon the air-dry condition of the normal resting seed, the seeds lost weight as shown below : — {a) The seeds of the Apple {Pyrus Mains) lost 45 per cent, of their weight. [b) The seeds of the Bread fruit {Artocarpus incisa) lost 55 per cent. of their weight. [c) The seeds of Momordica Charantia lost 30 per cent, of their weight. [d) The seeds of Tamus communis lost 44 per cent, of their weight. [e) The seeds of the Honeysuckle [Lonicera) lost 42 per cent, of their weight. [f) The seeds of Armn maculatum lost 49 per cent, of their weio-ht. [g) The seeds of the Shaddock (Citrus decumana) lost 40 per cent, of their weight. (/;) The seeds of Genipa dusiifolia lost 43 per cent, of their weight. (/) The seeds of Opuntia Tuna lost 45 per cent, of their weight. Before quitting this subject of the drying of seeds after their liberation by nature's means or after their collection by man, I would refer the reader to Note 12 of the Appendix for further details ; but in many ways this stage of the drying process is linked with other processes dealt with in other chapters. SUMMARY (i) For the satisfactory study of the latent life of seeds, says Becquerel, the experiment ought to cover a period far exceeding that of the duration of the seed's germinative capacity (p. 225). (2) The author's investigations into the changes in weight that seeds experience during the first three or four years after their assump- tion of a stable weight in the drying process give the following indications. The impermeable seed preserves its weight and shows no hygroscopic reaction throughout that period ; whilst the permeable 240 STUDIES IN SEEDS AND FRUITS seed remains always in the same hygrometric state, and retains the same average weight, showing only fluctuations of i or 2 per cent, on either side of a constant mean (p. 226). (3) As regards the impermeable seed, it is urged that as long as it preserves its weight and is non-hygroscopic, we may assume that it retains its germinative powers (p, 226). (4) The constancy of the weight of impermeable seeds during a period of three or four years is then illustrated in a tabular form (p. 228). (5) The failures in impermeable seeds are at once detected by the balance. They are instructive in their presentation of permeability as a quality by default, the impermeable seed owing to some defect in its coats gradually gaining weight and slowly assuming the role of a permeable seed (p. 229). (6) Reference is made at some length to the very different view of the ultimate fate of the impermeable seed held by Professor Ewart. In the case of Acacia seeds he considers that in the course of years they become as dry as corresponding inorganic material, and may hold less than i per cent, of moisture. This view is controverted, and an explanation of its origin is suggested (p. 230). (7) It is shown that in all experiments on the weight of permeable seeds extending over some time, the disturbing effect of the hygroscopic reaction, involving as it does a variation of 2 or 3 per cent, of the total weight, presents a great obstacle to the detection of small differences. For this reason, therefore, the experiments should cover many years (P- ^35). . . . . , , (8) It is considered that the increase in weight of i to 2 per cent. recorded by Van Tieghem and Bonnier, in the case of seeds of peas, haricots, vetches, etc., after a two years' experiment, comes within the ordinary hygroscopic range and does not necessarily imply an increase in the average weight (p. 236). (9) With the object of testing the influence of time on seed-weight, the author began four years ago a series of experiments on the seeds of nearly thirty genera, the intention being to continue them for many years (p. 237). (10) The necessity in weighing experiments extending over long periods of first selecting seeds that have completed the drying process is pointed out (p. 237). (11) In this connection it is shown that whilst some seeds, as those of leguminous pods and of similar dehiscent fruits, are almost completely air-dry when liberated naturally from the fruit, others from fleshy or watery fruits have still 40 or 50 per cent, of their weight to give up to the air. Seed-like fruits, as grains of cereals and acorns, have to submit to a "sweating" or "airing" process before storage (p. 238), CHAPTER XI A CLUE TO THE HOMOLOGIES OF FRUITS Some casual observations of the berries of a Berheris in mv garden directed my attention to the fact that the seeds in the ripe fruit were harder, smaller, and lighter in weight than those of the green berry, or, in other words, that seed-contraction had taken place within the moist fruit. This was established by further investigation, as shown by the results tabulated below. The curious circumstance that the seeds of Berberis had undergone shrinkage in the ripening berry gave me a clue for attacking the problem concerned with the homologies in the maturation of different kinds of fruits, especially of the berry, capsule, and legume. It led me to study the conditions of seed-shrinkage and of seed-coloration in fruits generally, and as a matter of course this in its turn led to the investiga- tion of the dehiscence and drying of the fruits with which such matters are closely bound up. It kept me clear of the en- tanglement of the controversy relating to the priority of the Table showing the Contraction of the Seeds of Berberis IN the Ripening Berry. The shrink- age of seeds in the moist berry affords a clue for the comparison of fruits in their ripen- ing stages as illustrated (a) by Ber- beris, Condition of fruit. Condition of seeds. Average weight of a seed. Average length of a seed. Full-sized green berry just beginning to colour Ripe berry Soft and green Harder and brown 0-23 grain 0-19 ,, 4-5 millimetres. 3"5-4 .. The loss in weight of the seed was about 17 per cent. 16 242 STUDIES IN SEEDS AND FRUITS capsule and the berry by finally causing me to regard the baccate condition as one that may be imposed on a variety of fruits, not only on the capsule but on the legume, as in the Tamarind, Acacia Farnesiana, and some species of Cassia, but also on the nucule, as with some Labiatae. The indications in this table are sufficiently evident. Sub- sequently, on investigating this point in the cases of Arum maculatum, Tamus communis, and Passiflora pectinata, I found that there also a marked contraction of the seeds occurred whilst the berry was passing from the green unripe stage into the red, juicy, mature condition. In the instance of Arum maculatum, after a comparison of the full-sized green and red berries on the same spike, and con- taining the same number of seeds for several plants, the follow- ing results were obtained. Table showing the Contraction of the Seeds of Arum maculatum in the ripening Berry. Condition of fruit. Condition of seeds. Average weight of a seed. Average size of a seed. Full-sized green berry . Red berry .... Whitish and un- wrinkled. Reddish and wrinkled. I ■ I grain 0-9 ,, 6 millimetres. 5 The loss in weight of the seed was about 1 8 per cent. It is thus shown that in spite of their immersion in a moist pulp, the seeds in the reddening berry of Arum maculatum underwent a noticeable contraction and loss of weight. To the eye the contrast is greater than appears in the figures of the table, since the change is associated with marked differences in the general appearance and condition of the seeds. On the one hand, the seeds of the green berry are not only larger and heavier, but they are distinguished also by their whitish hue and their unwrinkled surface. On the other hand, the seeds of the red berry, besides differing in size and weight, are reddish, wrinkled, and somewhat harder. The embryo in both CLUE TO THE HOMOLOGIES OF FRUITS 243 cases is rather less than half of the seed's length, the chief difference being in the albumen, which is rather mealy in the seeds of the red berry and more fleshy in the seeds of the green berry. Another indication of the contraction of the seed in the moist ripening berry is to be found in the decrease in the relative weight of the coats. Since the coats form 33 per cent, of the weight of the entire seed in the green berry, and 25 per cent, in the red berry, we see that these seeds follow the principle laid down for shrinking seeds in Chapter IX. The fruits of Tamus communis give us the same indications, the seeds of the ripe red berries being smaller, less heavy, and rather harder than those of the full-sized unripe green berries. There seemed at first to be an intermediate stage, when the berries assumed a yellowish hue, but this proved to be con- nected with the premature withering of the parent stem. In making such observations it is necessary to compare berries growing on the same branch. The table subjoined gives the average of a large number of weighings and measurements, almost all yielding similar results. Table showing the Contraction of the Seeds of Tamus communis in the Ripening Berry. Condition of fruit. Condition of seeds. Average weight of a seed. Average size of a seed. Full-sized green berry Red berry Greenish-yellow Brown and harder o'57 grain 3-9 millimetres. 3-6 „ The loss in weight of the seed was about 9 per cent. The loss of weight (about 9 per cent.) of the shrinking seed in the ripening berry is not great, and a much greater loss is sustained when the seed is exposed to the air, as is shown in the results given a page or two later. The individual differ- ences in weight and size in the seeds of Tamus communis seem small, but they become considerable when forty or fifty seeds are weighed together or measured in a line. The seeds of 244 STUDIES IN SEEDS AND FRUITS the green berry are greenish yellow, whilst those of the red are brown, the " browning " beginning in the green berry. In both stages the seeds are firm and the albumen solid, but the brown seeds are rather harder. From the berries of Passiflora pectinata, a species first described from the Bahamas, the same evidence is obtained. I made a study of these fruits in the island of Grand Turk at the southern end of the group. In the red mature berry the dark purplish crustaceous seeds are enclosed each of them in a moist, pulpy aril, as is characteristic of the genus, the whole interior of the fruit being moist. In the green, full-grown unripe fruit, the seeds are dark green, heavier, larger, and rather softer than in the ripe berry, the interior of the fruit, together with the saccate arils, being relatively dry. The results of my observations may be thus tabulated. Table showing the Contraction of the Seeds of Passiflora pectinata in the Ripening Berry. Condition of fruit. Full-grown, dryish green berry Red, ripe, moist berry Condition of seeds. Dark green and semi-crusta- ceous in dryish arils Dark purplish and crustaceous in moist, pulpy arils Average weight Average length of a seed. of a seed. Average breadth of a seed. o'35 grain 5 '8 millimetres.j 3 "5 millimetres, 0-31 5"3 The loss in weight of the seed was about 1 1 per cent. The shrinking of the seed immersed in the moist pulp of a berry is significant in many ways, and particularly because it supplies, as already observed, a clue by which we can trace the homologies in the maturing and drying stages of very different types of fruits. Or perhaps we would better describe it as affording a datum-mark to which we can reduce for purposes of comparison the various conditions presented by such fruits. CLUE TO THE HOMOLOGIES OF FRUITS 245 It has first to be noticed that the loss of weight which the seeds of these berries undergo in the moist fruit is but a small proportion of the loss which they sustain when subsequently freed by decay of the berry and exposed to the air. If the seed of Tamus communis loses 9 per cent, in the reddening berry, its total loss of weight when dried in free air amounts to about 46 per cent., as shown in the results tabulated below. The seeds of Eerberis^ Passiflora^ and of Arum maculatum^ which give the same indications, are there compared with it. One can recall familiar instances of the shrinking, hardening, and "browning" of seeds in fleshy fruits such as the Apple, the Sapodilla, and the Star Apple ; but here, though the change is evident to the eye, it is not easy to give a numerical value to the difference without a carefully guarded comparison of the average weight of the seeds in a large number of the full- sized unripe and ripe fruits. As the green apple mellows, its soft white seeds become smaller, harder, and brown in colour. The same process is familiar in sapotaceous fruits like the Sapodilla and the Star Apple {Achras Sapota and Chrysophyllum Cainito)^ where, as the fruit ripens, the soft white seeds become hard and brown. In the following table the loss in weight of the seed in the ripening berry is compared with the total loss when the berry dries up. Changes in the Weight of Seeds of Berries during the Ripening and Drying up of the Fruit. Weight of a seed in grains. Relative weight of a seed, taking the seed of the green berry as loo. Green berry. Ripe Dried-up berry. berry. Green berry. Ripe berry. Dried-up berry. Berberis (species of) . Arum maculatum Tamus communis Passiflora pectinata . 0-23 i-io 0-57 o"35 1 1 o"i9 1 o'i2 1 100 0-90 0-55 100 0-52 0-31 100 0-31 0-20 100 w 9' 89 5^ 50 54 57 246 STUDIES IN SEEDS AND FRUITS This table supplies a means of comparing the behaviour during the maturing and drying stages not only of the seeds of other kinds of fruits, but of the fruits themselves, and particularly of the capsule and the leguminous pod. We will first take the Horse-chestnut {^Msculus Hippo- castanurn)^ which almost acquires the baccate habit, though its familiar condition, as it lies open on the ground, is that of a dryish dehiscent fruit. The same preliminary shrinking of the seed, associated with hardening and " browning " of the seed-coverings, takes place in the closed capsule. These changes, however, only occur in the last stage of maturation immediately preceding dehiscence. As the green fruit mellows with maturity it becomes yellowish, and it is during this mellowing stage that the shrinking, hardening, and browning of the soft white seed take place within. If the soft white seed is removed and allowed to dry in the air, its coats rapidly harden and assume the characteristic reddish-brown hue, a change which experiment showed to be associated with a loss of 17 per cent, of the original weight. The hardening and coloration of the coverings were completed in twenty-four hours, when the seed was placed in a warm, dry cupboard ; whilst in diffuse light in a damp room they occupied two or three days, an indication that these changes are the result of partial drying only and do not require the action of light. If we wished to designate the particular stage in the maturation of the capsule of the Horse-chestnut correspond- ing to the ripe berry, we should select the mellowing stage immediately preceding dehiscence, when the green capsule assumes a yellowish tinge. It is then, and we are now indebted to the clue supplied by the seeds of the Berberis berry, that the seed undergoes its preliminary shrinking and that the hardening and colouring of its coverings within the closed capsule occur. Like the seeds of the berry also, it has yet much more water to lose. It has been already implied that as soon as the capsule begins to open it displays a well-browned, hard-coated seed (or seeds), which, as indicated by an experi- CLUE TO THE HOMOLOGIES OF FRUITS 247 ment before described, has already sustained before dehiscence a loss of water to the extent of 17 per cent, of its original weight as a soft white seed. When such seeds after removal from the naturally dehiscing fruit are allowed to dry in the air of a room, the total loss of weight finally amounts to about ^2 per cent. Thus, to take an example, a soft white seed freshly removed from a full-sized green capsule and weighing 300 grains would weigh about 250 grains when first exposed as a brown, hardening seed in the fruit commencing to dehisce. In the air it would rapidly dry, until it ultimately assumed a stable weight, subject only to hygroscopic variations, of about 142 grains. Stated as percentages, these changes in weight in the successive stages of drying would be as follows : — Soft white seed . . . . . .100 Same seed after some hardening and shrinking in ) ^ the closed capsule j ^ Same seed after the drying process has been ) completed in the open capsule J ^' Such are the indications supplied by the Horse-chestnut seeds and by comparing them with the data before given for the berries of Berberis^ Passiflora^ Arum maculatum^ and Tamus communis^ it will be at once perceived that they run well together with the indications of the berry. There is the same preliminary shrinking of the seeds within the closed ripening fruit, and there is the same great loss of weight when the fruit has passed maturity and begins to dry. Whether the seed undergoes the greater part of its drying within a shrivelling berry or exposed in an open capsule, the process belongs to the same stage in the history of the fruit. We can now perceive how the shrinking of the seed in the ripening berry comes to our aid in contrasting other fruits in their several stages of maturation by enabling us to fix on a stage that is common to all. My next example of a capsular fruit will be that of Iris iris Pseud- Pseudacorus. Here, as in ^senilis Hippocastanum^ the Horse- chestnut, the soft white seeds of the green full-sized capsule begin to shrink and harden and commence to " brown," whilst acorus. 248 STUDIES IN SEEDS AND FRUITS the fruit is mellowing and assuming a yellowish hue before dehiscence. Since the soft white seeds of the green capsule lose 60 per cent, of their weight when allowed to dry, and since the brownish seeds in the capsule on the eve of dehiscence lose about 50 per cent, of their weight when exposed to the air, it follows that about 20 per cent, of their weight is lost by the soft unripe seeds when shrinking in the ripening fruit before dehiscence. We thus get for Iris Pseudacorus results very similar to those obtained for the Horse-chestnut. Both display the regime of the berry in the drying of their fruits and seeds, processes quite independent of any distinction that may be drawn between baccate and capsular fruits. Here again the mellowing, greenish-yellow stage immediately preceding dehiscence corresponds to the ripening of the green berry when it reddens in Arum, Tamus, etc. I have compared the results for these two capsular fruits in the table below with the mean result for a berry as supplied by the data for Berberis, Arum maculatum, Tamus communis, and Passiflora pectinata. Comparison of Capsules and Berries with regard to the RELATIVE Weight of the Seeds during Maturation and during the Drying Process, the Weight of the Seed in the Full-sized Green or Unripe Fruit being taken as 100. Soft white seeds in the full-sized green capsule. Seeds in the ripe capsule on the eve of dehiscence. Seeds dried in air after being freed from the dehisced capsule. 47 40 ^sculus Hippocastanum (Horse-chestnut) Iris Pseudacorus . Mean results for a berry (see table before given) 100 100 83 80 Seeds in the full- sized green berry Seeds in the ripe berry Seeds in the shrivelled and dried-up berry 100 86 53 The weights of an average Horse-chestnut seed in the three stages are 300, 250, 142 grains ; and for Iris Pseudacorus, 2*0, i-6, and o"8 grain. CLUE TO THE HOMOLOGIES OF FRUITS 249 Observations of this kind extend over a year or two and require a little patience, since the same locality has often to be visited several times, and much also has to be done at home. I will now take a fruit intermediate between a capsule and Second, a berry, the baccate capsule of Thespesia populnea, a tropical fap^s'Ses beach tree of the malvaceous order. The full-sized yellowish- illustrated by r • 1 J 1 • 1 11 • Thespesia green rruit possesses an abundant, thick yellow juice and white, populnea. softish seeds. In the next stage it becomes a darker green, the juice becomes scanty, and the seeds shrink a little, harden, and assume a purplish tinge. Then the fruit begins to " brown " rapidly, and its sides collapse ; whilst its seeds also turn brown, and, continuing to dry and harden, ultimately lose about half their original weight when the drying of the fruit is complete. Finally, the fruit breaks down and the seeds are freed by its decay. In the figures 100, 87, and 50, which represent the relative weights of the white softish seed, of the purplish seed before drying of the fruit has actively commenced, and of the brown, hard seed in the fruit when the drying has ended, we have stated numerically the essential stages of the capsule and the berry. The actual weights of an average seed in these three stages would be ^' ^^ 4-8, and 2*25 grains. An index of the changes in the fruit is afforded by the changes in the condition of the adherent calyx, which remains moist and green long after the seeds within have begun to shrink and harden, and only begins to wither when the capsule com- mences to " brown " and to lose weight, thus indicating that the first shrinkage of the seed within the still moist fruit, as in the case of the true berry, precedes the active drying of the capsule. With the ordinary dehiscent leguminous pod there is Third, quite another regime. In illustration I will first take that po^£Jas"in"^ of Casalpinia sepiaria. the familiar "Wait-a-bit" of Jamaica. Caesalpmia ^ . ^ ^ , . , -^ sepiana. The full-sized green pod with its white, soft seeds represents the green capsules of the Horse-chestnut and Iris Pseudacorus and the green berries of Berberis^ Arum maculatum^ and Tamus communis after they have attained their maximum size. In the next stage, which corresponds to the ripe berry and the mellow- 2 so STUDIES IN SEEDS AND FRUITS ing capsule before dehiscence, the pod turns yellowish green and the white seeds experience the preliminary shrinkage and hardening and take on a greenish hue. Then active drying of the pod and seeds commences ; but before dehiscence begins the shrinking and hardening of the seeds have been almost completed, so that in the opening pod we find normal dark, mottled resting seeds that will perhaps lose another lo per cent, of the original weight in keeping. The respective weights of the seeds in the green pod, in the pod turning yellowish, and in the dried pod on the eve of dehiscing, are 8*3, 6*2, and 3-8 grains, which stand to each other as 100, 75, and 46, and thus indicate the three stages of the berry. Some of the processes involved in the general shrinking of leo-uminous pods and seeds will be discussed more in detail in the succeeding chapter. Here I will only refer to cases which seem specially suggestive for determining the stages in maturation and in the drying process that are homologous or are truly comparable with those in the capsule and berry. In this connection the pods of Ukx europaus occupied much of my attention. They were the first fruits to which I applied the clue afforded by the Berberis berry. Though much of their behaviour is characteristic of the typical leguminous pod, it is not always that we find the stages so well defined. With Ulex europaus practically all the shrinkage, hardening, and coloration of the seeds are carried out in the closed pod ; and when the pod dehisces it exposes to view the normally contracted hard seeds. Three stages in the maturation and drying of the fruit are distinguished by the colour of the seeds. When the green pod has reached its full size they are soft and bright green, a hue that they owe mainly to the dark green embryo which can be seen through the thin coats. Then follows a stage which corresponds seemingly to the first failure of the nutrient supplies from the mother plant. The soft green seeds turn a greenish yellow as the pod begins to dry and darken. But there is no very evident shrinking of the seed, though the cord withers. That is only detected by careful measurement and CLUE TO THE HOMOLOGIES OF FRUITS 251 by the balance, since it is slight in amount. The colour-change again is largely due to the change in colour of the embryo. The last stage is occupied with the active drying of the darken- ing pod, the yellow seeds contracting and hardening rapidly and adopting the permanent chocolate-brown colour of the normal resting seed. In this stage, however, the change in seed-colour is mainly an affair of the coverings, whilst in the two earlier stages, marked by green and yellow seeds, it was largely concerned with the embryo. This last stage closes with the completion of the drying of the pod in the sun's rays, and the pod dehisces suddenly, giving rise to those curious little clicks that one hears so frequently when standing near a gorse bush on a sunny day. Now all these changes in the seeds of Ulex europaus^ from the soft green state to the hard chocolate-brown condition, are carried on in the closed pod. In fact, all three stages may be observed together in the same pod. The discolora- tion begins at its distal end, and before the blackening process has extended down one-third of the pod's length all the seeds will have changed their hue from green to yellow. When it has affected four-fifths of the pod, the uppermost seeds will be found actively shrinking, hardening, and colouring chocolate brown ; and by the time the whole pod has blackened all the seeds will be in that condition. But although the pod has dried considerably during the blackening process, it is still fairly moist at its completion ; and the seeds, though considerably reduced in size and no longer soft, are far from being as hard as in the normal resting seed, and have yet to decrease in size. Up to this point there has been no opening of the pod, and the completion of the drying process and the ultimate dehiscence are soon affected by exposure to the sun. It has been noted above that the green and yellow seeds have green and yellow embryos, the colouring of the soft seeds depending in these two stages mainly on the colour of the embryos. This change from green to yellow is probably connected with the failing of the nutrient supplies, since it coincides with the commencement of the withering 252 STUDIES IN SEEDS AND FRUITS of the cord or funicle. But the change in the last stage from yellow to chocolate brown is exclusively associated with the drying process. This is indicated by a simple experiment. If a yellow seed is allowed to sink in a glass of water no altera- tion takes place. But if it is allowed to rest on the surface for a day or two, the lower submerged portion preserves its yellow hue, whilst the upper part exposed to the air turns brown. Table showing the Shrinkage or Contraction of the Seeds OF Ulex europ^us (Gorse) in the Pod before Dehiscence. Condition of pod. Condition of seeds. Weight of a seed. Size of a seed. 4-4-5 millimetres, 3-3-5 2-5 Green and full-sized Beginning to discolour and to dry Dried and on the eve of dehiscing Soft and green. Soft and greenish yellow Hard and choco- .late brown 0-25 grain 0*20 ,, 0"1I ,, The loss in weight in passing from the green to the yellow stage is 20 per cent. ; whilst the total loss experienced is 56 per cent. Thecorre- If we wished to determine the correspondence in the between^^a maturing and drying stages between a leguminous pod like legume and ^h^^- of Ulex europaus and a berry like that of Berberis^ we would do so in this fashion : — f Green pod, with soft green seeds of Ulex. ' \ Green berry, with soft green seeds of Berber'is. /■ Green pod of Ulex commencing to darken, with green seeds \ turning yellow and shrinking slightly. ' ] Colouring berry of Berberis^ with seeds turning brown and V. becoming smaller and harder. r Pod of Ulex rapidly drying and blackening, with seeds turn- j ing brown, shrinking greatly, and hardening. ^' j Berry of Berheris rapidly shrivelling, with seeds losing much V water and hardening. r Dried pod of Ulex liberating its seeds by dehiscence. 4. J Shrivelled berry of Berber'is liberating its seeds through the ( decay of its coverings. The correspondence in the fourth stage will perhaps be least expected. But when one reflects that the opening of a pod, CLUE TO THE HOMOLOGIES OF FRUITS 253 however methodical the process may appear to be, depends on a structural character which was originally developed for a special purpose in the living plant and could have had no concern with the liberation of seeds from a dead pod, we must confess that this appearance of method is purely accidental. It is an accident that I am able to avail myself of the binder's crease in tearing the sheets of a book into regular portions ; and to that extent the mode of dehiscence is accidental with the Ulex pod. It will subsequently be shown that if this parallelism between the dry, dehiscing pod and the decaying, shrivelled berry is valid, it ought to have a far-reaching influence on our views of adaptation and seed-dispersal. My readers will in this connection recall those numerous leguminous pods that liberate the seeds only by breaking down through decay, and I shall in a later page point out that the peculiar form of moniliform pods, which is there described, is probably determined by constrictions induced in an earlier stage of the development of the fruit by the abortion of the ovules, and the premature shrinking of the seeds. Comparison of the mean Results obtained for Berries, Capsules, AND Legumes, with Respect to the Relative Weights of the Seeds during the Maturation and Drying of the Fruits, the Weight of the Seed in the Green Unripe Fruit being taken AS 100. In the green berry 100 In the ripe berry S6 In the dried-up, shrivelled berry 53 In the green capsule 100 In the ripe capsule on the eve of dehiscence 82 When dried in air after the opening of the capsule 44 In the green legume 100 In the green legume turn- ing yellow or beginning to blacken 78 In the dried-up legume on the point of dehiscing 45 The results here given are for the berries of Berberis, Passiflora, Arum maculatum, and Tamils commwiis ; for the capsules of .-Escuhts Hippocastanum (Horse-chestnut) and Iris Pseudacorus ; and for the legumes of Ccesalpinia sepiaria and Ulex europaus (Gorse). 2 54 STUDIES IN SEEDS AND FRUITS We see from these results that in the ripe berry only a small shrinkage of the seeds occurs, namely, about 14 per cent., and that the seeds have still much of their water to lose in drying. In the capsule the same small shrinkage takes place before dehiscence, but the principal loss of weight and the greater part of the contraction occur after the seeds have been exposed to the air. In the pod practically all the shrink- age of the seeds is carried out in the closed fruit, and when dehiscence occurs normal resting seeds are exposed. The seeds of a capsule are thus placed at a disadvantage when compared with those of a pod, since in the pod the seeds are protected until sufficiently hardened, whilst in the capsule they are usually exposed in a relatively soft and in a less protected state on account of the dehiscence taking place at a much earlier stage. The parallelism, or rather the correspondence, between the stages of the maturation and the drying of berries, legumes, and capsules, may be put in the following manner :■ — ■ (i) The green berry, the green capsule, and the green legume of maximum size, with their large, soft, white or green seeds, are all in the same stage. (2) The ripe, juicy berry, the mellowing, still closed capsule, and the legume just beginning to discolour represent the next stage, which is characterised by a slight shrinking of the seeds, and by their coloration when white, or by a change of colour if green. Though evident enough in the berry, this stage is transient and often disguised in the capsule and legume, points which will be further discussed in the succeeding chapter. It is early in this stage that the shrinking of the cord or funicle marks in dehiscent fruits the commence- ment of the severing of the seed's connection with the parent plant. (3) The shrivelling berry, the capsule dehiscing and losing much of its water in the air, the drying and blackening but still closed legume, belong to the next stage. Dehiscence occurs at a much earlier stage with the capsule than with the CLUE TO THE HOMOLOGIES OF FRUITS 255 legume. As shown in Chapter XIII, the legumes lose before dehiscing nearly all the water they can yield to the air ; whilst the capsule before it opens may not even have commenced to dry, or loses only a small proportion of the water that it ultimately gives up in the air-drying process. The greater part of the shrinking and of the hardening of the seeds belongs to this stage. (4) This stage characterises only the berry and the legume, since the dried capsule has already completed its stages and is lying widely open with its seeds falling out. The shrivelled berry and the dry pod now in their turn liberate their seeds, the one by decay and the other by dehiscence. Such are some of the principal points brought out in this comparison of the capsule, legume, and berry. Additional evidence will be adduced in support of most of them in the following chapters. There is much that is significant in this correspondence between these three different kinds of fruits, and much that we should bear in mind when we speak of special adaptation The question 'for dispersal of fruits and seeds. Though there is a great tion. deal that may please the eye and captivate the fancy in the mechanisms of a dehiscing pod or capsule, it is doubtful whether they represent anything more in nature than the rotting apple and the shriveUing currant. The exposure of the brightly coloured seed in the opening pod, as in AbruSy Adenanthera^ and Erythrina^ often adds beauty to the plant ; yet, viewed from this standpoint, it is nothing more than one observes in the seeds exposed in the decaying orange and in the rotting Anona fruit as they lie on the ground. If the seeds in the ripe berry, or mature baccate fruit, are in the same stage as those in the mellowing capsule before it opens, or in the closed, full-sized green legume on the eve of drying, then the shrivelled berry represents the dried- up open capsule and the shrunken but still closed legume. In all three fruits the shrivelling and drying, whether 256 STUDIES IN SEEDS AND FRUITS accompanied or not by dehiscence, seem to betoken a stage that was not, if I may use the expression, in the original plan laid down for nature. The biological connection being severed with the parent plant, the fruit dies, but the seed lives. There was a time, as I hold, in the age of vivipary, when uniform climatic conditions prevailed, and the embryo was already well advanted in germination and able to start life for itself before the severance from the parent took place. In later ages climatic differentiation has intervened, and the fruit " dies, leaving the embryo in all stages of arrested growth, with the chances of its future development by no means assured. The embryo is then dependent for its protection on the hardening of its coverings, a process mainly accomplished after the cord has withered, and in a fruit no longer drawing its nutrient supplies from the parent, but practically dead. It is hard to detect anything but a baffled design when we see nature suddenly withdrawing the plant's fostering care over its offspring and leaving all to chance. It is difficult to perceive any evidence of adaptation for dispersal in a rotting apple ; but if my view is correct, the same should be true of dehiscent fruits, the withering and opening of which, though seemingly displaying more of method, are equally determined by external influences of a haphazard kind. SUMMARY (i) The shrinkage of seeds in the moist berry affords a clue for the comparison of fruits in their ripening stages, as illustrated by Berberis (p. 241), Arum maculatum (p. 242), Tamus communis (p. 243), and Passijiora (p. 244). Thus guided, we can trace the homologies in the maturing and drying stages of very different types of fruits. This shrinkage within the moist berry, which involves a loss of weight on the average of from 10 to 15 per cent., represents but a small proportion of the total loss which the seed sustains when exposed to the air or dried with the fruit. (2) It is the shrinking of the seeds in the ripening berry that comes to our aid in contrasting other fruits in their several stages of maturation by enabling us to fix on a stage that is common to all. CLUE TO THE HOMOLOGIES OF FRUITS 257 Thus in the case of capsular fruits, like those of the Horse-chestnut {/Esculus Hippocastanum) and of Iris Pseudacorus, we see displayed the regime of the berry in the preliminary shrinking of the seeds within the ripening fruit and in the subsequent great loss of weight when the fruit has passed its maturity and dries in the air. Here, then, it becomes evident that the green berry and the green capsule are in the same stage, the ripe juicy berry and the mellowing still closed capsule in another stage, and the shrivelling berry with. the drying, dehiscing capsule in a third stage. (3) With the ordinary dehiscent leguminous pod, as illustrated by the iDehaviour of those of C^esalpinia sepiaria and Ulex europceus^ there is quite another regime, since dehiscence occurs much later than in the capsule ; but here also the preliminary shrinking of the seeds and other changes in the fruit enable us to detect the equivalents of the stages of the berry and the capsule in those of the leguminous pod. We thus learn that the berry, the capsule, and the legume, in the full- grown green or so-called unripe condition with large soft seeds, are in the same stage. The ripe, juicy berry, the mellowing, still closed capsule, and the green legume just commencing to discolour represent the next stage in the maturation of the fruit, which is characterised by a slight shrinking and hardening of the seeds. Then the berry shrivels, the capsule dehisces and dries, and the legume darkens and dries rapidly, but still remains unopened, the shrinking and hardening of the seeds being in all cases actively continued. When the last stage arrives the dry capsule has already completed its history and lies gaping widely, with its seeds falling out, whilst the shrivelled berry and the dry legume now liberate their seeds, the first by decay, the second by dehiscence. (4) If the seeds in the ripe berry are in the same stage as those of the mellowing capsule before dehiscence and of the full-sized moist legume on the eve of drying, then the shrivelled berry represents the dried-up open capsule and the dry legume or pod on the point of dehiscing. From this standpoint, therefore, the mechanism of a dehiscing legume or capsule, however adaptive its appearance, does not count for anything more in nature than the rotting apple or the shrivelling currant. If we are not able to detect any signs of adapta- tion for dispersal in a decaying berry, the same should be true of dehiscent fruits, the withering and opening of which, though apparently displaying more of method, are equally determined by external influences of a haphazard kind. 17 CHAPTER XII THE HOMOLOGIES OF FRUITS AS REVEALED IN THE DRYING PROCESS This chapter is concerned principally with the story of the drying fruit, or rather with the indications that it supplies. Here again it is to the balance that we look for our data, and it will be seen that the results obtained are of some interest and often unexpected in their significance. Without further introductory remarks I will at once plunge into the middle of my subject. Indications It is very suggestive that two such different-looking fruits traSbetween ^s those of Barringtonia speciosa and Ribes Grossularia (Goose- the berries of berry\ both of which are characterised by the systematic Barringtonia , ^ <' , . . „ , ,, . ^ . „ ■ speciosa and botanist as berries, give olt when allowed to dry naturally in Grossularia the air the same amount of water, losing in each case about b^n-vT" ^5 P^'' c^^^' of their weight in the ripe condition. The one, a large fruit 8000 to 9000 grains In weight, is described as corticate and fibrous, whilst the other, weighing about 100 grains, is described as succulent and pulpy. Even if the air- dried fruits of both plants were subsequently exposed in the oven to a temperature of 100° C, the relative weight of the water-free residues would not differ greatly in amount, that of the Gooseberry being probably about 10 per cent., and that of Barringtonia speciosa about 13 or 14 per cent, of the moist fruit. For details of the experiments on the Gooseberry see Note 10 of the Appendix. When comparing two such different-looking kinds of 258 THE HOMOLOGIES OF FRUITS 59 berries we are likely to begin with a misconception. The dry, Misconcep- husky, symmetrical fruit oi Barringtonia speciosa^ such as the ftom^he"^ currents disperse over the coral islands of the Pacific, is a dried comparison 1 J 1 • 1 1 1 • , °f ixmts that berry, and as such is only to be compared with the shrivelled are not in the fruit of the Gooseberry. This lack of adjustment is frequently ^*'"^^^*S^- met with in comparing the stages of fruits. Take, for instance, the distinction which the systematist usually draws between a berry, as fleshy and indehiscent, and a capsule, as dry and dehiscent. Here we are contrasting the air-dried capsule with the moist berry, two quite different stages in the history of these fruits. Naturally, the true correlative of the dry dehiscent capsule would be the shrivelled berry, whilst the ripe berry would find its homologue in the full-grown moist capsule as we find it living on the plant. This relation between the berry and the capsule has been already dealt with in Chapter XI. The necessities of the systematist are partly responsible for the incongruities in the comparison of fruits, since he gives a place to the dry fruit that retains its shape, but refuses to recognise as on the same footing the dried-up berry or the shrivelled drupe. But part of the blame must lie with one's natural repugnance to the shrivelling process, seemingly so significant of inutility and death. Let but the form be preserved, even though the life of the fruit has gone, and we become apt to attach importance to a distinction which is purely accidental and in no sense ordinal in character. These remarks do not at all exaggerate the lack of true adjustment which prevails in the general classifications of fruits. It is far from easy to see how this can be avoided in practical systematic botany, but the inconsistency remains. There lies beside me The Handbook of the British Flora, by Bentham and Hooker (5th edit., 1887), and there I read (p. -T^G) that fruits are generally divided into "succulent" and Thedistinc- " dry," the first being usually indehiscent, whilst the second 5y"h/*^" are often dehiscent and open at maturity. The succulent systematist . . J^ ^ between fruits are there typified by the berry and the drupe, and succulent the dry fruits by the capsule, legume, achene, etc. Of fruits.'^^ 26o STUDIES IN SEEDS AND FRUITS the dehiscent capsule and legume it is stated that when ripe the pericarp usually splits into valves. One may note in passing that this can only be said of the capsule. How- ever, each of these statements taken independently has the sanction of experience ; but it is an error to connect them together in a classification of fruits, since they are essentially incongruous. In the first place, as regards the two main divisions into succulent and dry fruits, it is apparent that we are here con- trasting moist fruits that have yet to dry with those that have more or less completed the drying process. As already indicated in Chapter XI, the moist fruit is a living fruit, whilst the dry fruit is a dead one. Strictly speaking, there is no such marked distinction in nature between moist Nature does and dry fruits, except such as is connected with the differ- such^a ^^ce between a living and a dying or dead fruit. To be distinction, convinced on this point we have only to look at the columns of the following tables, though this view has already been established in the previous chapter. Where, for instance, are we to find amongst other types of fruits the stage that is representative of the air-dried capsule of Datura ? If we followed the method of the systematist we should find it in the moist drupe of the Sloe {Prunus communis). A glance at the tables will show that in so doing we should be comparing a dead and dry fruit with a moist and living one. Both the living capsule and the living drupe in these two plants lose about the same amount of water when they die and dry up (70 to 73 per cent, of their weight) ; and if we contrast them we should either compare them when they have attained their maximum size on the plant as moist living fruits, or when they have lost their vitality and have dried up. This is the only valid mode of comparing fruits, and the failure to adopt it leads to erroneous conceptions of the biological significance of the process of dehiscence. The dried dehiscing capsule and the shrivelled drupe go together. THE HOMOLOGIES OF FRUITS 261 The same rule applies when we look for the representative of the Datura capsule amongst the legumes or the berries. A Canavalia pod, an Arum berry, and a Datura capsule contain about the same amount of water in the fuU-orown living state, and can only be compared in the same condition either as moist living fruits or as dry dead ones. We thus come to perceive that all fruits, when they reach maturity on the plant, whether drupes, legumes, berries, capsules, All mature etc., are moist fruits and cannot be distinguished from each are'moiS^*^ other by their water-contents. Nature does not recognise ^'■"**s- the distinction between moist drupes and berries on the one hand and dry legumes and capsules on the other. Group for group, the contrast between capsules, legumes, berries, drupes, etc., as regards their water-contents in the full-grown living condition, is relatively small ; and in each group we find much the same variation in the amount of water lost in drying, namely, between 50 and 80 or 85 per cent. The differences are mainly developed when we allow the drying to take place in all cases, the berry and the drupe to shrivel up, and the capsule and the legume to dry and dehisce, the ultimate contrast between the asymmetry of the one kind and the retention of the regular form in the other being dependent on the nature of the tissues composing the pericarp. Nature makes no deliberate effort to assist the systematist, and inconsistencies of the kind above noted are inseparable from our necessarily arbitrary endeavours to systematise her processes. The error involved above is of course the com- parison of fruits that are not in the same stage. But other inconsistencies are apt to follow. Thus, as already noticed, it is implied in the above statements from The Handbook of the British Flora^ that dry dehiscent fruits like those of the capsule and the pod open when they are " ripe." This might indicate that dehiscence occurs in these fruits in the moist, mature condition. But, as we will see in Chapter XIII, this is only true of the capsule, the legume opening when the fruit 262 STUDIES IN SEEDS AND FRUITS is dry and dead. The typical capsule which opens before drying begins, or in the early stage of drying, is just as much entitled to the designation of "moist" as a berry. I venture, therefore, to think that these remarks indicate the necessity of renovating our prime conceptions of the differences between fruits. The Loss of Weight of Mature Fruits, including their Seeds, WHEN Dried in Air under ordinary Conditions of Tempera- ture. (The full-grown moist fruits before drying or shrinking begins are here employed.) I. Legumes. Loss of weight after drying in air for Char- weeks or months. Number acter of . Stated as a of seeds. dry Stated in grains. percentage. Stated as fruit. water-loss or water- Moist Dry Moist Dry percentage. weight. weight. weight. weight. Pisum sativum (Pea) 7 250 50 100 20 80 Guilandina bondu- 2 400 95 100 24 76 cella Phaseolus multi- 4 300 75 100 25 75 florus (Scarlet- runner Faba vulgaris s 750 19s 100 26 74 (Broad Bean) Mucuna urens 3 1000 280 100 2S 72 Canavalia obtusi- 6 400 112 100 28 72 folia Leucsena glauca 24 100 30 100 30 70 Entada polystachya H 800 240 TOO 30 70 Csesalpinia sepiaria 5 100 32 100 32 68 Cassia fistula 95 Woody 5000 1650 ICO 33 67 Vicia sativa . 10 ... 15 5-6 100 37 63 Cajanus indicus . 4 40 16 100 40 60 Acacia Farnesiana 20 150 60 100 40 60 Andira inertnis I Woody 135 54 100 40 60 Dioclea reflexa 4 Woody 1800 738 100 41 59 Vicia sepium 3 or 4 6 2"S 100 42 58 Ulex europreus 4 or 5 i'S I'l 100 44 56 Poinciana regia 40 Woody 3500 1575 100 45 55 Csesalpinia Sappan 4 Woody 260 130 1 00 50 50 THE HOMOLOGIES OF FRUITS 263 II. Capsules. Momordica Charantia Blighia sapida (Akee) Scilla nutans . ^sculus Hippocastanum (Horse-chestnut) Iris foetidissima Ipomoea tuba . Iris Pseudacorus Gossypium barbadense Primula veris (Primrose) . Datura Stramonium . Allium ursinum Thespesia populnea . Swietenia Mahogani (Ma- hogany) Hura crepitans Hypericum Androssemum Viola tricolor . Bignonia (near sequinoc tialis) Arenaria peploides . R a V e n a 1 a madagascari ensis Aquilegia (species of) Canna indica . Char- acter of dry fruit. seeded 3 seeded Inde- hiscent Woody Woody Inde- hiscent Siliqui- form Woody Folli- cular Loss of weight after drying in air for weeks or months. Stated in grains. Stated as a percentage. Stated as water-loss or water- Moist Dry Moist Dry percentage. weight. weight. weight. weight. 500 75 100 IS 85 1730 346 100 20 80 10 2*4 100 24 76 700 i68 100 24 76 170 42-5 100 25 75 100 25 100 25 75 250 65 100 26 74 100 26 100 26 74 4 I '2 100 30 70 360 108 100 30 70 2-4 0-8 100 33 67 230 76 100 33 67 5800 2030 100 35 65 3200 1152 100 36 64 3 I "I 100 37 63 3 1*2 100 40 60 1250 500 100 40 60 8 3*3 100 41 59 700 294 100 42 58 10 4-5 100 45 55 100 47 100 47 53 264 STUDIES IN SEEDS AND FRUITS III. Berries, Drupes, etc. Loss of weight when dried in air Family. Fruit. for weeks or months. 1 Stated in grains. Stated as a percentage. Stated as water- loss or water- Moist Dry Moist Dry percent- weight. weight. weight. weight. age. Pyrus Malus (Apple) Rosacese Berry 900 ,.6 100 14 86 Ribes Grossularia Ribesiacese J 120 18 100 15 85 (Gooseberry) Tamus communis Dioscorese ,, 14 2 '5 100 18 82 Opuntia Tuna Cactese it 800 144 100 18 82 (Prickly Pear) Citrus aurantium Aurantiacea; ,, 2400 460 100 19 81 (Orange) Sambucus nigra Caprifoli- ,, 3'3 07 100 21 79 (Elder) acese Lonicera Pericly- Caprifoli- ,, 4-8 I '2 100 25 75 menum (Honey- acese suckle) Monstera pertusa Aracese 5"o i-o 100 20 80 Arum maculatum ,^ 6-1 1-8 1 00 29 71 Hedera Helix (Ivy) . Araliacese 5 2 100 40 60 Quercus Robur (Oak) AmentacejE Nut 60 24-0 100 40 60 Prunus communis Rosacece Drupe 30 8 100 27 73 (Sloe) Sparganium ramosum Pandaneas ,, I 0-45 100 45 55 Barringtonia speciosa Myrtacese Berry 9000 1350 1 00 15 85 Areca Catechu . Palmace^ ^ 250 So 100 32 68 Cocosnucifera (Coco- ,, Drupe 60,000 18,000 100 30 70 nut) Acrocomia lasiospatha Arenga saccharifera . J, ,, 550 200 100 36 64 J, Berry 600 300 100 SO 50 Mauritia setigera ,, ,, 1000 560 100 56 44 Cocos plumosa . „ Drupe 100 63 100 63 37 Oreodoxa regia jj Berry 15 10 'o 100 66 34 Hyophorbe Vers- \\ ,, '5 "5 6-2 100 40 60 chafftii JVoie. — The total amount of water in these fruits can readily be ascertained by applying a small correction to the loss sustained when dried in air. As a rule the air-dried fruits would lose between 10 and 15 per cent, of their weight when exposed to a temperature of 100° C. The corrected result for the air-dried berry of Tamus communis, assuming that it lost 12 per cent, of its weight in the oven, would be as follows : — Moist weight Air-dried weight Oven-dried weight . . .16 Loss when air-dried ... 82 Loss when oven-dried . . 84 THE HOMOLOGIES OF FRUITS 265 The contents of the foregoing tables raise a number of interesting points, and it would be easy to devote some chapters to the details of the experiments, if space allowed and necessity required it. Indeed, not a few of these points will come under our notice when we discuss the relation of parts in the living or moist and in the dead or dried fruit. There are, however, some matters that call for immediate notice. It has already been explained in a note to the tables that a small minus-correction applied to the air-dried weight will give approximately the total water-contents, such as would be indicated by the loss of weight of the fresh fruit when exposed to a temperature of 100° C. The water remaining after drying in air is the water of hygroscopicity, which fruits possess in common with all other vegetable substances, whether living or dead (see Chapter VII). Another point here claims attention. It is remarkable how The large much water fruits described as woody in the dry state contain ^?er"in° in the full-grown living condition on the plant. Amongst the Jj^^mg, ^^^^^ fruits that in the dried state specially merit the designation of " woody," one would certainly include the long pod of Cassia fistula^ the large capsule of the Mahogany tree {Swietenia Mahogant)y and the polycoccous capsule of Hura crepitans (the Sandbox-tree). Yet each of these fruits, as will be seen in my tables, loses about two-thirds of its weight when allowed to dry in free air in the moist, green, full-grown condition, the water-loss, denoted by the decrease in weight, being respectively 67, 6^^ and 64 per cent, of the original weight. Although the woody fruits lose less water when dried in air than fleshy fruits, a glance at the tables will show that the difference is usually not great. If fleshy fruits may lose between 70 and 80 per cent, of their weight, woody fruits may lose between 60 and 70 per cent. But many disturbing influences come into play and prohibit Disturbing any precise general statement until their effect is determined. affeSgW This is at once made evident when we perceive that the woody ^f^Ji^gg" *"" fruits of Cassia^ Swietenia, and Hura lose about as much 266 STUDIES IN SEEDS AND FRUITS water when dried as does the Acorn (^Quercus), the capsule of Viola, and the pods of Vicia and Cajanus. Then, again, it is apparent that several very different causes have combined to produce the same result in the Coco-nut, the Acorn, and the pod of Cassia fistula, all of which lose a similar amount of water, namely, from 60 to 70 per cent, of their weight. Some of these causes will be considered when we come to deal with the relation of parts in a fruit. The development of sugars in the ripening berry makes a material difference in the weight of the fruit after it has been dried in air. Elder berries {Samhucus nigra), before the sugars are formed, lose about 87 per cent, of their weight, but with the production of sugar their weight during drying is diminished by only 78 or 79 per cent., the saccharine materials being especially hygroscopic and preventing the complete drying of the fruit. The berries of the Honeysuckle {Lonicera Periclymenum), which, when the sugars are formed, lose about 75 per cent, of their weight, behave in a similar fashion. In the same way the ripe fruits of the Gooseberry {Ribes Grossularia) cannot be dried properly in air on account of the abundance of the sugars. The congealed juice that encrusts the surface of the air-dried berry is very hygroscopic. The ripe fruit loses about 85 per cent, of its weight ; but if the sugars are removed by washing, the air-dried materials make up only about 5 per cent, of the weight of the moist berry (see Note ioa of Appendix). The same behaviour is displayed by the berries of Opuntia Tuna (Prickly Pear), which, when air- dried, lose 82 per cent, of their weight, but if the sugars are removed by washing, the dry residue of 1 8 per cent, is reduced to 12 per cent, (see Note iob of the Appendix). The seeds of such sugary fruits often remain moist and sticky and require washing for their complete drying. Those of the Pomegranate (Punica Granatum), for instance, never dry properly unless previously washed. A good example of the influence of the sugars on the air-drying of fruits is afforded by the different behaviours of the husky coverings of the ripe THE HOMOLOGIES OF FRUITS 267 drupes of the Coco-nut {Cocos nucifera) and of Cocos plumosa. In the first case there is a loss of quite 80 per cent. In the second case, where the coverings contain a good deal of sugar, the loss is only about ^^ per cent. As with the sugars of berries, the presence of oil in the pericarp of fruits greatly retards the air-drying process. This explains why, in my experiments, the drupaceous berries of Oreodoxa regia (Palmaceae) lost only 22 per cent, of their weight. In the same way, fragments of the pericarp of the Cashew- nut {Anacardium occidentale)^ which contain a caustic oil in abundance, dry but slightly, losing less than 15 per cent, of their weight when exposed to the air. In connection with the loss of weight sustained by fruits Immature 1 1 ... . . • 1 • • r 1 fruits contain when dried m air, some curious considerations arise rrom the more water fact illustrated in the table below, that immature fruits contain J^Jg™**"''® more water than mature fruits. Although the data there given refer almost exclusively to immature fruits of nearly the maximum size that are characterised by incompletely developed seeds, they illustrate a process of change that runs through nearly the whole of the fruit's life-history, from the time of its occurrence as a young fruit, until, with maturity passed, the fruit dries up and loses its vitality. But in this respect, namely, in the progressive decrease of the water-contents as they pass from youth to maturity, and thence to the loss of vitality, fruits share the fate of all vegetable substances. If we regard only the percentage of water in the whole or in this re- in the part of a plant, whether stem, leaf, root, fruit, or seed, fiiustratea we can construct a scale beginning with the young growth, P^arac-* containing, we will say 70 or 80 per cent, of water, and ending teristic of the with the air-dried dead substance that holds only the water world, of hygroscopicity, amounting only to 12 or 15 per cent., the water which it derives from the air and which it gives up in the oven. Between the initial and terminal stages of this scale there is an ever-progressive decrease in the proportion of water that the living plant-substance yields up when drying to the air. But this progressive decrease in the proportion of the 268 STUDIES IN SEEDS AND FRUITS water-contents differs in character in the earlier and later parts of the period covered by the scale. In the first portion, where we are concerned with the living plant or its part, the decrease in the percentage of water is merely due to the fact that during the building-up processes involved in growth the solids increase more rapidly than the liquid constituents. In the second portion, when the plant or its part is dying and drying, there is an actual loss of water, and this goes on until the plant- substance, like all other dead vegetable materials, ceases to give up water to the air and retains only the water of hygro- scopicity. Such is the role played by water in the life of a plant, either entire or in part, as stated in terms of the decrease in the proportion of the water-contents. In active life this decrease, as just observed, is only relative, and is due to the more rapid increase of the solids. When the plant dies it is absolute, and involves the loss of all the water required for the processes of vitality. But between the period appropriated by the living plant or its part and the period associated with death and desiccation, there often seems to be an interval of varying length characterised by repose. This is the rest-period that appears to be claimed by all vegetable life, by the plant in its entirety, and by the plant in its smallest constituent parts. Though it is difficult to point to any plant or any part of a plant that does not seem to undergo this so-called rest-period, I am inclined to think that nature often merely cloaks, but does not suspend the processes of growth. With fruits such a period of repose, if it exists at all, must be very brief ; and with the great majority of seeds, which soon lose their vitality on being dried, I should be disposed to believe that the period between the cessation of active growth and the com- mencement of loss of vitality must be very short. Such are some of the considerations that present them- selves when we reflect that the immature fruit holds more water than the ripe fruit. They illustrate the great significance that lies behind all experiments even of the simplest nature, THE HOMOLOGIES OF FRUITS 269 such as are represented in the following two tables. The story of the Acorn, viewed from the standpoint offered by the role taken by the water-contents, seems to be particularly suggestive, especially in connection with the tendency to vivipary at times displayed, a subject discussed in Chapter XIX. Some of the most fascinating problems bound up with plant-life lie behind the phenomena of the drying fruit ; and none are more important than that connected with germination on the plant. In this respect the story of the decrease in the water- contents of the Ivy berry {Hedera Helix)^ as it grows steadily from the autumn through the winter, finally dropping to the ground in the spring, and often with one or more of its seeds germinating, is particularly interesting. This progressive decrease is clearly shown in one of the following tables. But Table showing the Difference in the Water-Contents of Im- mature AND Mature Fruits, as indicated by their Loss of Weight when dried in Air under ordinary Conditions. (By mature fruits are meant those that are full-grown and moist, contain ripe seeds, and show no signs of drying. By immature fruits are usually meant those that have attained nearly the full size and weight, but have seeds with contents not set or incompletely developed. For the berries of Sambucus nigra another explanation is required, as is given below. ) Average weight of Loss of weight when dried moist fruit m grams. in air stated as a percentage. Type of fruit Immature. Mature. Immature. Mature. Iris Pseudacorus . 200 250 85 per cent. 74 per cent. Capsule. ,, foetid issima 130 170 83 n ,, ^sculus Hippocastanum 600 700 82 ,, 76 „ ,, (Horse- chestnut) Hura crepitans 3000 3200 86 „ jj Phaseolus multiflorus . 270 300 82 „ Legume. Faba vulgaris (Broad Bean) Guilandina bonducella . 650 750 85 „ >' 340 400 81 ,, Hedera Helix (Ivy) 17 S"5 79 f. Berry. Quercus Robur (Oak) * . 10 60 77 .. 68 „ Nut. Sambucus nigra (Elder)t 3 3 87 „ 78 Berry. * The browning acorn is taken as the mature stage (see Chapters XIV and XIX). t In the case of Elder berries the difiference is due to the formation of sugars in the ripening fruit, as explained in an earlier page of this chapter. 270 STUDIES IN SEEDS AND FRUITS the data there supplied only illustrates one of the features in a process that in the Ivy berry often terminates in germination on the plant. Matters more directly relevant to the subject of vivipary in this plant, especially those concerning the growth of the seed and its embryo, will be dealt with in detail in Chapter XIX. The decrease in the water-contents of a growing fruit naturally involves the increase of the solid constituents. How fruits gain in solids as they grow is well brought out in th^- general table immediately preceding these remarks, and with Table showing the gradual Decrease in the Water-Contents of Acorns (Quercus Robur) as indicated by their Loss of Weight IN different Stages of their Development when dried in Air under ordinary Conditions. (The cupule is not included.) Date of collection. Condition of fruits. Average weight of a fruit in grains. Loss when dried in air, taking the moist fruit as 100. Aug. 24, 1 910 \ Firmly attached by living / tissue to cupule 10 7 J per cent. Sept. 13, ,, 32 74 ,, 20, ,, Still firmly connected 46 72 „ 27, „ Connection a little looser, nuts browning. SO 65 „ Oct. 4. „ do. 55 61 Connection slight 51 48 ',', is', ',', Browned ; fall at a touch 57 48 After two months, when 30 00 ,, air-drying complete. Sept. 4, 1908 Firmly attached to cupule 56 75 )> '7) >> Still firmly attached 62 68 „ 30' )> do. 62 54 Oct. 6, „ Connection slight; nuts browning. 71 51 „ 14, ,. Browned ; fall at a touch 64 43 ,, After two months, when air- 40 00 ,, drying complete. This table is only intended to illustrate the decrease in the water-contents. At the same time a rough idea can be formed of the progressive changes in the average weight of an acorn, which is all that the method of the experiment will allow. The acorns of each series were obtained from the same locality. It will be noticed that the acorns of 1908 were considerably heavier and larger than those of 1910. Ten fruits were used in each case. For other data relating to the ripening and drying of the fruit the reader should consult Chapters XIV, XIX, etc. THE HOMOLOGIES OF FRUITS 271 special detail for the Acorn and the Ivy berry in the two additional tables. These tables speak for themselves, and there is no necessity to push the subject further here. Table showing the gradual Decrease in the Water-Contents of Ivv Berries (Hedera Helix) as they Developed and Matured at Redland, Bristol, during the Winter i 908-1 909. (The collections were made by my sister, Mrs H. Mortimer, from the same plant and weighed by her at once, the samples containing from forty to sixty berries. They were weighed again by the author some months after.) Date of Condition of Changes of weight of an average berry when dried in ordinary air-conditions. collection. Stated in grains. Stated as a percentage. Nov. 9 „ 18 All green Moist. Dry. 1-69 0-36 2-86 o-6i Moist. Dry. 100 21 loo 21 Dec. 3 Two - thirds \ 4-02 0-98 100 24 >> 17 green, rest V black J 4-96 1-46 100 29 Jan. 9 All black 5 "44 1-93 100 35 ., 24 Feb. 21 " 5*50 2'o6 4-10 1-64 100 37 100 40 Mar. 19 " 4-02 1-58 100 39 Observation at Salcombe, Devon. May 16 " 4-60 2-00 1 100 43 NoU.—The reader is referred to Chapter XIX for other details respecting the growth of the seed and its embryo. SUMMARY (i) Some of the most interesting problems connected with plant- life lie behind the phenomena of the drying fruit. (2) But a comparative study in this direction brings to the front preliminary considerations of importance, more especially those concerned with the lack of true adjustment which prevails in the general classification of fruits and with the comparison of fruits in the diiferent stages of their history (p. 259). (3) Observation on the drying of fruits shows us that when the systematist speaks of a berry as fleshy and indehiscent and a capsule as dry and dehiscent, he is contrasting a living with a dead fruit (p. 260). (4) We come also to discover the fallacy that may lie in the distinction between succulent and dry fruits, especially when it implies 272 STUDIES IN SEEDS AND FRUITS a contrast between a nioist living fruit that has yet to dry and a fruit that has more or less completed the drying process and is to a greater or less degree devitalised (p. 260). (5) The tabulated results of the author's observations on the drying of fruits in air indicate that nature does not recognise this distinction between moist and dry fruits, all mature living fruits being moist fruits. The contrast which the systematist draws between the fleshy drupe of Prunus and the dehiscing, dried, or drying fruit of Datura is not the contrast nature offers. Nature as interpreted through the balance tells us that the full-grown moist and living capsule of Datura Stramonium contains just as much water as the ripe drupe of Prunus coynmunh^ and that the dried and dead open capsule of the one could only be compared with the dead and shrivelled drupe of the other (p. 261). (6) These misconceptions lead to others. Thus, it is usually implied that dry dehiscent fruits, like typical capsules and legumes, open only when they are ripe, an assumption that involves us in much confusion between the maturing, dehiscing, and drying stages of fruits. How much we may err in this respect is indicated in Chapter XIII, where it is also shown that ripe capsules and ripe legumes are all moist fruits as far as their water-contents are concerned, and that whilst the capsule dehisces in the living, moist state, the legume opens in the dried and dying condition (p. 262). (7) Amongst the points brought out in the tables are the large amount of water in full-grown, living, woody fruits, and the manner in which the drying of fruits is retarded by the presence of sugars and oils (p. 265). (8) Some curious considerations also arise from the fact that immature fruits contain more water than mature fruits. But this progressive change in the water-contents as the fruit passes from immaturity, when it contains, we will say, 80 per cent, of water, to maturity, when the amount would be about 70 per cent., and thence on to the drying stage accompanying its loss of vitality, when it retains only the water of hygroscopicity, probably about 15 per cent., is characteristic of all vegetable matter (p. 267). (9) At first during active life this decrease is only relative and is due to the more rapid increase of the solids. In the latter stage, when the plant dies, it is absolute and involves the loss of all the water required for the active processes of vitality. The occasional vivipary of the acorn on the Oak and of the seeds of the Ivy berry [Hedera) on the plant is in part an expression of the principle that the mature fruit contains less water than the immature fruit (p. 269). (10) The increase in the solid constituents of a growing fruit as the water-contents decrease is well brought out in the tables, in one table for a variety of fruits, in another with special detail for the acorn and the Ivy berry (p. 270). CHAPTER XIII THE DEHISCENCE OF FRUITS One is apt to associate the process of dehiscence with dryish fruits, neither very soft nor very hard, and justly so, because many openins: fruits belong; to this category. Both the capsule The capsule 111 1 J J r •. J .u dehisces and the legume are classed amongst dry truits ; and the before dry- implication often is that their dehiscence is connected with |e|i^e*after the relief of strain produced by unequal contraction during drying, the drying of the fruit. My observations indicate that this applies more especially to fruits like legumes, that often only dehisce after they have been considerably dried, and that as a rule it does not concern capsules. The incorrect conception seems to be due in the case of capsules to the tendency to regard as one and the same process the loosening of the cohesion between the valves or carpels, which may take place when the green fruit begins to mellow, and their subsequent drying, when the fruit is rapidly losing its vitality. The sudden relief of tensions generated by drying in the latter part of the dehis- cence, resulting as it sometimes does in the forcible expulsion of the seeds, produces effects that often emphasise the influence of drying in the latter stages of the process, and is apt to favour the idea that dehiscence is merely a matter of desiccation. Thus, in Fiola, the breaking down of the cohesion between Capsules the valves is one thing, whilst the subsequent folding inwards ^^^^^^ moist of the edges as the valve dries up is another. Whether the ^^^^^°^^_ "nipping" and forcible propulsion of the seeds are purposive ingispre- or merely accidental can only be determined after an extensive 273 18 274 STUDIES IN SEEDS AND FRUITS comparative study of different fruits in all their stages. That which concerns us here is the circumstance that Viola capsules will dehisce under moist conditions without any drying what- ever, and without any of the display of the results of elastic tension which that process engenders. If we place a detached, full-grown green fruit in wet moss, the first stage of dehiscence, namely, the loosening of the cohesion between the valves, will be accomplished without any drying, whilst the curling in of the edges of the valves will be inhibited. This is what happened in my experiments on the fruits of the Garden Pansy {Viola tricolor) ; and one may contrast with such results, where dehiscence takes place without any drying, those results for the same plant where drying plays a prominent part in the opening of the capsule, as given below. The Drying and Dehiscence of a Detached Ripe Capsule of Viola tricolor, weighing Four Grains, the Results being stated in Percentages. Full-grown green fruit (4 grains). Beginning to de- hisce (2-8 grains). Valves lying back (2-6 grains). Valves folded on the seeds and dry (i '6 grain). 100 70 66 40 Again, if we take a number of full-grown but still closed capsules of Msculus Hippocastanum (Horse-chestnut) and of Iris PseudacoruSy and place some in wet moss and others exposed to the free air on a table, we obtain the following results. In a few days several of those on the table will be found to be opening after losing 25 or 30 per cent, of their weight ; whilst those in the wet moss which have not dried at all but have probably added to their weight will be also dehiscing. The detached capsules of Iris fatidissima illustrate the same thing more forcibly, since they dehisce in wet moss, but fail altogether to open when allowed to dry. Drying, therefore, though it develops strains, the relief of which ends in the dehiscence of fruits, is not a necessity for their opening. It is THE DEHISCENCE OF FRUITS 275 essential in the latter stages, if the parts of the fruit are to acquire the elasticity concerned in the forcible expulsion of the seeds, but it is not necessary for seed-liberation. Seeing that most fruits liberate their seeds without forcibly discharging them in this manner, it may be doubted whether we should specially regard such violent propulsion as purposive, determined as it is largely by the degree of dryness of the air. However this may be, it is evident from the above experi- ments that other factors than those concerned with drying go to determine the dehiscence of fruits. Though the phenomena are physical in origin, writes Pfeffer Prof. Pfeffer {Physiology of Plants^ iii. 146-153), the development of the hiscence. requisite physical conditions is a physiological problem. The required instability, when mechanical agencies of external or of internal origin may release the dehiscing organism, is produced by growth, the requisite tissue-strains and the conditions for their release being prepared by the vital activity of the organism. The distinction which he draws between internal and external agencies in the opening of dehiscent fruits coincides with one of the principal differences between the modes of dehiscence of legumes and many capsules, as indicated by my observations. Whilst in the one case the opening of the capsule is often brought about through the distension of its walls by the growth of its seed-contents, in the other the dehiscence of the legume is usually caused by the strain generated in the drying process. The author's T,^ 1111- 1 ..^°'' J observations. in the capsule the dehiscence, however arismg, corresponds with the maximum growth of the fruit. In the legume it happens at a much later stage, namely, after its biological connection with the parent has been more or less severed, and when it has lost the greater part of its water by drying, and in consequence its vitality. Regarding the difference The capsule between the two processes from merely a physical standpoint, jj.jgs The we can say that whilst the capsule dehisces and dries, the ^^53,is"el. legume dries and dehisces. By removing a single letter we can at the same time express the biological distinction by say- ing that whilst the capsule dehisces and dies, the legume dies 276 STUDIES IN SEEDS AND FRUITS and dehisces. With the capsule, therefore, the mechanism of dehiscence is concerned with a living fruit. In the legume it has to do with a dead one. The pre- In both the green capsule and the green legume there is mellowmg generally a preliminary mellowing stage, corresponding, as stage. shown in Chapter XI, to the ripening of the berry ; but it may be transient or disguised by other changes, more especially with the legume. It marks the completion of active growth and indicates seemingly the commencement of the severance of the biological connection between the fruit and the parent. In this mellowing stage the green fruit often assumes a yellowish tinge or a paler hue, and its tense, turgid appearance gives place to one of relative flaccidity. Its firm, rigid walls become softer and more yielding, and the cohesion along the sutures is loosened. This is best exemplified in the capsule, though the same may be noticed in legumes, as in the Pea. But whilst with the capsule the immediate result is the dehis- cence of the fruit, with the legume no such effect is produced at that stage, and dehiscence occurs at a much later stage as a relief from the tension produced from outside by the drying up of the pod. Observations I will now refer to some of my observations on the opening hiscenceof of capsules on the living plant, and will begin with Iris theliifng" fattdissima. This plant, growing as it often does in more or plant. less shady woods, offers one a better chance of eliminating the drying factor than does Iris Pseudacorus^ which frequents more exposed situations. Though the mellowing stage is not so easily recognisable as in the species last named, we can detect the approach of dehiscence in the lessened turgidity of Iris foetid- the green fruit and in its paler hue. The capsules, when they display the earliest signs of their opening (a slight separation of the valves at and near the apex) are quite moist fruits and show no signs of drying. But, once begun, the process is rapidly completed by drying, and in a few days the valves stand well back, exposing the bright scarlet seeds. That the first rupture is due to some cause issima. THE DEHISCENCE OF FRUITS 277 acting from within, such as the pressure of the seed-contents, is highly probable. Full-grown fruits placed in water and in wet moss commenced to dehisce in a few days, whilst others left to dry on a table made no effort to open after prolonged drying. With Iris Pseudacorus careful observation of the living Iris Pseuda- plant convinced me that dehiscence as a rule occurs in moist, *^*"^^* mellowing fruits, that is to say when the green capsule assumes a yellowish tinge. Although, as already observed, the detached fruits will dehisce in wet moss, this does not exclude altogether the participation of the drying factor in the plant's natural station by the water-side, where it is fully exposed to the sun. If fruits dehisced in my experiments without drying, some of them also opened after they had lost about 25 per cent, of their weight exposed on my table. Nature does not follow formularies in such matters ; and though internal causes mainly determine the early dehiscence, we cannot entirely disregard the influence of external conditions. The stages in the history of the dehiscence of fruits normally maturing in September would seem to be as follows : — (i) The full-grown green capsule, firm, full, and turgid ; (2) The capsule mellows, becoming yellowish and rather softer, which results in loosening the cohesion between the valves ; (3) The dehiscence begins, determined by the pressure of the seed-contents, but aided by exposure to the sun and by buffeting in the wind ; (4) The rapid drying of the fruit and the full exposure of the seeds. The results of one method of proving that capsular fruits A proof that dehisce in the moist condition are given in the subjoined dehfscehi table for Iris and for the Horse-chestnut. Here we find the moist , , r . . r • condition, that the fruit showing the first signs of dehiscence loses but slightly less in weight when dried in air than does the full-grown moist fruit with matured seeds that has not begun to open. 278 STUDIES IN SEEDS AND FRUITS Table showing the Loss of Weight when dried in Air of Full- sized, Moist, Capsular Fruits, both before Dehiscence and IN the earliest Stage of Dehiscence, the Dehiscing Fruits BEING gathered IN THAT CONDITION FROM THE PlANT. Full-sized fruits before dehiscence. Fruits beginning to dehisce. Iris Pseudacorus Iris foetidissima ^sculus Hippocastannm . 74 per cent. 11 :: 70 per cent. 73 69 „ Note. — The average was taken of from five to ten fruits in each case. But it is often very difficult, by observing capsular fruits on the plant, to eliminate the drying factor ; and the most we can often say in such cases is that dehiscence takes place in the living fruit before the drying is very evident. In the case of Scilla nutans., for instance, we find the full-grown green capsule in a condition of strain, not, however, on account of the pressure of the enclosed seeds, since they only partly fill the cells, but through the turgidity produced by active growth. Though green and moist, it ruptures with a " pop " when squeezed between the fingers. A little later the fruit becomes paler, looks a little dryer, and its turgid appearance has dis- appeared. If we press it gently, there is no longer an elastic resistance, and the valves, though still in position, are seen to be partially disconnected. The fruit has dehisced, although still green and fairly moist, and only the pressure of the finger reveals what has occurred. Here it seems impossible to separate the dehiscence from the early stage of drying ; and yet the loosening of the cohesion between the valves was probably effected in the mellowing stage when the firm, turgid green fruit became softer and almost flaccid. On physio- logical grounds I would suppose that the dehiscence of a green fruit in active vitality could never be normally produced either by internal or external causes, and that dehiscence could only occur after the biological connection with the parent begins to be severed. This I take to be the mellowing stage of fruits, THE DEHISCENCE OF FRUITS 279 pronounced in berries, much less evident in capsules, and often more or less disguised, or very slight, in legumes. It is likely that the behaviour of the fruits of Scilla nutans is typical of many capsules but partially filled by their seeds. We will take the case of Stellaria Holostea. Here, as with Stellaria fruits in general, the development of the fruit is far in advance ° °^ ^*' of the seeds. In its early stage, when about the size of a small pea, the capsule is little more than an empty bladder, since the young seeds do not occupy one-tenth part of the cavity. But when the green fruit is full-grown, it is loosely but not entirely filled by its soft white seeds, so that there is no pressure on the capsular walls from within. Afterwards, as the fruit dries, the seeds shrink and only half fill it. The valves, though still in position, become disconnected at the edges, as a slight pressure of the finger will show ; but there seems to be no reason why they should be sundered in the drying process, since they remain in position ; and it would appear probable that the coherence between their edges was broken down, as in Scilla nutans^ when the biological connection of the fruit with the parent first began to fail. Assuming that the first preparation for dehiscence is accomplished when the capsule ceases active growth and begins to mellow, then we perceive that the next cause of the actual disconnection of the valves may vary according as the seeds completely or only partially fill the cavity. In the first case, TheMehis- as with Iris, the capsular walls, owing to the loosening of the completely connection between the valves, are no longer able to respond to the pressure of the seed-contents by increased growth. They yield at the weakened sutures and the fruit dehisces, the valves as they wither and dry falling back and exposing the seeds. On the other hand, when the capsule is not full of seeds, so that there is no distension or pressure on the walls from within, the cohesion between the valves is still loosened in the mellowing stage, but they remain in position during the early part or most of the drying. A few remarks may now be made on some other capsules. 28o STUDIES IN SEEDS AND FRUITS Those of Arenaria peploides behave like the fruits of Stellaria Holostea above described. Although detached capsules of Datura Stramonium lost i8 per cent, of their weight in my experiments before they dehisced, it is highly probable that, like those of /m, when kept in wet moss they would have opened without any drying at all. In the same way the moist mature fruits of the Primrose {Primula veris)^ after lying one night on my table, were found to be dehiscing and to have lost about 20 per cent, of their weight. But observation of the capsules on the living plant led me to consider that the first stage in dehiscence begins still earlier in the drying process. It is singular that the normal opening of the Primrose capsule at the top may be prevented by making a hole in its base. Appended are the results of my observations on the dehiscence of detached mature capsules of Primula veris and Datura Stramonium^ when allowed to dry on my table. Though obviously such experi- ments are not carried out under nature's conditions, their results will serve to illustrate the early opening of capsular fruits. The Drying in Air and Dehiscence of Detached Ripe Capsules of Datura Stramonium and Primula veris. Average weight in grains of a ripe moist fruit. Loss of weight, the ripe un- opened fruit being taken as 100. Before dehiscence. When de- hiscence begins. After the drying is complete. Datura Stramonium Primula veris 3SC-0 rs 100 100 82 80 30 30 In a species of Aquilegia growing in my garden the opening of the follicles took place shortly after the period of maximum growth. The green follicles are completely closed ; but as they ripen they acquire a purplish tinge and gape open at the base before normal dehiscence occurs. In such mature fruits the cohesion at the ventral suture becomes very slight, and a slight increase of the tension due to some external cause would THE DEHISCENCE OF FRUITS 281 bring about the separation. I found that the follicles just beginning to dehisce were usually the heaviest. Dehiscence may take place in the most watery of capsules, Contrast in as with those of Momordica, where 95 per cent, of the fruit contTnt*sof (excluding the seeds) consists of water, and in the hardest and capsules, most ligneous of capsules, as with those of Mahogany (Swietenia Ma/iogani), where the woody walls hold only about 66 per cent. of water and are 10 millimetres or nearly half an inch thick. The dehiscence of the fruits of a species of Momordica observed Momordica. by me in Jamaica (seemingly a cross between M. Charantia and M. Bahamind) was quite regular, and took place when the fruits were ripe and moist. After the seed-contents were removed, a ripe fruit not yet beginning to open lost 94 per cent, of its weight when dried in air, whilst a similar fruit just beginning to dehisce lost 91 per cent. The first stage In dehiscence seems to be due to the tension produced in the walls of the softening fruit by its contents. With the woody fruits of Mahogany the exact stage at which dehiscence occurs Mahogany, on the tree is not easy to determine ; but one can get an approximate idea. The full-grown green fruits, which seem to average about four-fifths of a pound in weight (5600 grains or 363 grammes) lose about two-thirds of their weight when detached and allowed to dry for several months, but they begin to open when they have lost about one-fourth of their weight. Drying in Air of a Detached, Green, Full-grown Capsule (not YET dehiscing) OF MAHOGANY (SwiETENIA MaHOGANi), INCLUDING Seeds. Condition of fruit. Weight in grains. Percentage. Green, full-grown, not yet dehiscing . The same fruit beginning to dehisce on my table The same fruit after drying for months 5900 4370 2030 loo-o 74"i 34"4 Dehiscence will probably occur also if drying is prevented, as shown below. These results for the drying of a detached fruit of Ma- hogany are, as far as dehiscence is concerned, very much the 282 STUDIES IN SEEDS AND FRUITS same as those obtained for detached fruits of Iris Pseudacorus and jEscuIus Hippocastanum (Horse-chestnut), which dehisced in my room after losing 25 or 30 per cent, of their weight. It has, however, been shown that in both these cases dehiscence occurred in wet moss when drying was prevented, and it is very probable that Mahogany capsules would behave in the same way. In the above experiment the fruit was dried in the entire condition. Closely similar results were obtained in drying a fruit that had been taken to pieces. The original total weight of the green fruit was 5576 grains, and the weight after the drying-in-air was completed was 1996 grains, the proportions being as loo'O to 35'8. The dehiscence of the fruits of Ravenala madagascariensis (Traveller's Palm), as observed by me in plants growing near the rest-house on the Grand Etang in Grenada, presents another type of the opening of capsular fruits. We are here concerned with a sub-drupaceous capsule displaying loculicidal dehiscence. Within the outer moist husk is a bony endocarp or " stone," which even in the fresh fruit requires a heavy blow to break it, and is 5 or 6 milHmetres thick. The dehiscence of this fruit raises a number of questions which are dealt with in other chapters. Not the least important of them are the propor- tion of parts and their several water-contents, and especially the failure of the young seeds, which is here quite phenomenal. With regard to the failure of the young seeds, reference will also be made to the subject in Chapter XVI. Here I will merely allude to it in connection with the fruit's dehiscence. The full-sized seeds vary from 9 to 1 5 millimetres in length ; but there is only room in each of the double spaces of the three compartments for from three to five, making a possible total ranging from eighteen to thirty seeds. But even this is often never reached. The fruit figured in Schumann's monograph below referred to probably did not contain more than fifteen seeds. Amongst those fruits examined by me there were not uncommonly only one to three full-grown seeds in each valve, or from three to nine or ten in all. All the rest of the fruit-cavity THE DEHISCENCE OF FRUITS 283 is occupied by numerous aborted seeds, 3 to 5 millimetres long, and varying from twenty or thirty to forty or fifty in number. This tremendous waste, not of ovules, but of seeds that could find no room for further development, plainly indicates the existence of a great tension within the living fruit. If the stony walls by their pressure are able to prevent the proper growth of all but a very few seeds, they are subjected in their turn to the opposing strain of the expanding seeds ; and in the end the seeds are successful in rupturing the walls, though too late for the maturation of most of them. This is evidently what happens on the tree, since it is the moist fruits that are found dehiscing. The first stage in the dehiscence of Ravenala fruits is therefore due to the expansive force of the growing seed, and drying does the rest. One may note in passing that the genesis of the thick, tough walls of this and other capsular fruits may lie in its being a response to the expanding pressure of the growing seeds. That drying is a potent factor in the completion of the process is shown in the behaviour of the " stones " of Ravenala when allowed to dry after being removed from ripe, unopened fruits. They begin to split loculicidally when they have lost about 15 per cent, of their weight, but some also split septi- cidally at the apex of the valves, the loculicidal dehiscence ultimately prevailing. The loss of water during the drying in air of the entire fruit may here be given, though the subject is also dealt with in tables given in Chapters XII and XIV, Results of Drying in Air a Fresh Fruit of Ravenala madagascariensis. Fresh fruit. Air-dried fruit. Loss of weight in drying. Weight in grains. Percentage of entire weight. Weight in grains. Percentage of entire weight. Skin and husk Stone with seeds , Entire fruit 268 452 720 37'2 62-8 lOO'O 49 252 301 i6-3 837 100 'O 817 per cent. 44-2 .. 58-2 „ The loss of weight of the stone without the seeds would be about 40 per cent. 2 84 STUDIES IN SEEDS AND FRUITS Excellent illustrations of the fruits of Ravenala are given by Schumann in his monograph on the Musaceae (Das Pflanzenreich^ iv. 45). He, however, figures only the dry fruit, as his description of eine holzige Kapsel would also imply. The ripe fruit before dehiscence is yellowish and moist, and it is only when it dehisces, dries, and turns brown that the woody texture is disclosed. The contrast above drawn in the case of such different kinds of capsular fruits as Momordka^ with 95 per cent, of water. Mahogany, with about 66 per cent., and Ravenala^ with about 62 per cent., leads one to compare these extreme examples with other capsules as regards their water -contents. The subject is discussed for fruits in general in the following chapter. The data tabulated below are merely intended to illustrate the great variation in the amount of water held by capsular fruits when full-grown, and before they begin to dehisce and dry, the seeds being for the convenience The Water-Contents of Full-grown Capsules before Dehis- CENCE AND DRYING COMMENCE, THE Seeds being EXCLUDED. Water-contents stated as a percentage. Weight in grains of full-grown Total, including capsules, Loss when dried the water driven excluding in air without off at 100° C. seeds. heat. (Estimated : see below.) Momordica Charantia . 600 94-0 per cent. 95 "o per cent. Scilla nutans .... 7 93'o >. 94'o ,, Datura Stramonium . 220 87-0 ,. 89-0 ,, Iris Pseudacorus .... 130 86-4 „ 88-4 „ Ipomoea tuba .... 35 86-0 ,, 88-0 ,, Arenaria peploides 17 84-6 „ 87 'o ,, ^sculus Hippocastanum (Horse- 440 84-0 „ 86-3 ,., chestnut) Primula veris (Primrose) 17 83-8 „ 86-0 „ Canna indica .... 36 82-6 „ 84-8 „ Iris foetidissima .... 70 787 „ 82-0 ,, Aquilegia (species of) . 17 76-8 „ 8o-o ,, Thespesia populnea ISO 76-3 „ 79-3 ., Swietenia Mahogani (Mahogany) 4800 62-6 ,, 66-3 „ Ravenala madagascariensis * 700 58-2 „ 62-5 „ * The seeds are here included, but their influence on the total result is small. THE DEHISCENCE OF FRUITS 285 of comparison excluded, except in one case there explained. In spite of the contrast between the fruits of Momordica and Scilla on the one hand, and those of Swietenia and Ravenala on the other, the driest of these mature fruits contain a large amount of water. Notes on the above table. — The total water of the fruit as given in the last column is estimated by applying a correction to the air-dried residue. Although the result is only approx- imate, the limits of error, as will be seen, are small. The water driven off in the oven after living vegetable substances have been air-dried is the water of hygroscopicity possessed in common by both living and dead matter (see Chapter VII). According to my observations, this varies usually from about 10 per cent, for air-dried, stony fruits to about 15 per cent. for loose-textured, air-dried fruits, such as ordinary legumes, capsules, and nuts, the seeds being excluded. Thus, in the case of Momordica^ the air -dried residue of 100 grains of fresh material would weigh 6 grains. In the oven, exposed to a temperature of ico° C, this residue would at the most lose 1 5 per cent, of its weight and would be reduced to nearly 5 grains, so that the total water in the fresh material would amount to about 95 per cent. In the same way, if, as is probable, the air-dried Mahogany capsule lost 10 per cent, of its weight in the oven, the air-dried residue of 37*4 grains out of 100 grains of fresh material would be reduced to 337 grains, so that the total water held by a living Mahogany capsule, excluding the seeds, would amount to 66-T^ per cent. It will have been gathered from the preceding remarks, as well as from the indications afforded in the table just given, that when we speak of a capsule as a dry fruit, we have usually in our mind dehiscing capsules, which have been more or Dehiscing less completely severed from the parent as far as the biological beSfgdead connection is concerned. Dehisced capsules now appear as '^J^^^^^^i dead or dying- fruits : and although even the toughest and to the V , , ij -111 . c herbarium, most ligneous among them hold a considerable amount or 286 STUDIES IN SEEDS AND FRUITS water when full-grown and before dehiscence on the plant, as dehiscing fruits they give up the greater part of it to the air, only retaining what is common to both dead and living organised vegetable substances, the water of hygroscopicity. All dehiscing capsules, whether they originally possessed in the full-grown, unopened condition on the plant as much as 94 or 95 per cent, of water, as in Scilla and Momordica^ or as little as 62 or 66 per cent., as in Ravenala and Mahogany, should be classed with dry fruits when they present themselves in the act of freeing their seeds on the plant. It is very questionable whether the expression " dry fruit " has any significance except for the herbarium. Consistently applied, it has no biological value, since the living connection with the parent plant has been severed. Let us take the dry, dehiscing fruits of Canna^ Iris, and Datura, which, as we observe them on the plant, certainly deserve that appellation, though they have lost their vitality. When full-grown on the parent and ready to dehisce, they contain (excluding the seeds) from 80 to 90 per cent, of water ; and their soft seeds, as shown by actual experiment on those of Iris Pseudacorus, are able to proceed at once with germination without the interruption of a rest-period. Such are the fruits with which the student of the living plant is chiefly concerned. The dry, dehiscing capsule belongs only to his herbarium. So it is with legumes, as will subsequently be shown, and so it is with the shrivelling berry. All that is purposive ends when a fruit has passed its prime. The fruit dies, let it be a capsule, a legume, or a berry ; and the mode of liberation of its seeds depends on structural characters that were developed when it was a part of a living plant, and could have had no possible concern with the escape of the seeds from a dead fruit. All appear- ances of adaptation to seed-dispersal in fruit-dehiscence are delusive and based on a one-sided view of the subject. We observe all those cases where Nature seems to give her aid and ignore the multitude of others that she seems THE DEHISCENCE OF FRUITS 287 to leave alone. Nature, as we should read her story, is indifferent to all. The distribution of seeds by dehiscing fruits thus presents itself as determined by the laws con- trolling the disintegration of dead organised matter ; and in this disintegration the loss of the water necessary for the fruit's vitahty occupies an early stage. It is with the drying of fruits biologically severed from the parent plant that the discharge of seeds by capsules, legumes, and other similar fruits is usually connected. As previously pointed out, the typical dehiscence of Thedehis- , '' , , r 1 J • cence of a legumes occurs at or near the close or the drying process, legume is the (The opening can be easily prevented by placing the fruit in a^dea?fnfit°^ wet moss, the valves ultimately falling apart through the decay of the connecting tissues.) If, then, dehiscence takes place in a capsule in a living fruit, it takes place in a legume in a dead fruit ; and all the objections urged in the case of a capsule against regarding the propulsive liberation of seeds as a special adaptation apply even more forcibly to the legume. Late dehiscence is evidently characteristic of all those numerous legumes, with which the reader will be familiar, where the dry valves spring apart suddenly (throwing the seeds often some distance) and then coil up spirally. It is likely, as in the capsule, that the first loosening of the connection between the valves takes place when the fully developed green legume begins to soften or mellow, a stage marking the beginning of the severing of the biological connection with the parent and the ushering in of the drying process. But though such a change is often more or less disguised in legumes, it may be recognised at times in the paler green colour, and more conspicuously in those cases where, as with C^salpinia sepiaria, the green fruit assumes a yellowish tinge. The drying pod generally darkens or blackens, as in Ficia, Lathyrus^ C<£salpima^ Ulex^ etc. ; but there are whole groups, as with Canavalia^ where the fruit as it dries becomes lighter in colour and ultimately has a nondescript, parchment- like appearance. 288 STUDIES IN SEEDS AND FRUITS As illustrating the dehiscence of typical legumes, I will take the behaviour of the pods of Vicia saliva, Ulex europ^us, and Casalpinia sepiaria, as determined by the balance. The plan adopted was similar to that followed in the case of capsules like those of Viola. The total loss of weight by drying of the detached, full-grown green fruit was first ascertained and then the loss of weight during parts of the process. Thus with Viola capsules I found the loss of weight before and after dehiscence began. With the legumes I checked the total loss of weight during the drying of the green fruit by ascertaining the average loss of weight after and before the pod had blackened. By combined observation on these lines of the detached fruit and of the fruit on the plant it is not difficult to obtain an approximate result. With Vicia and Ulex the final loss of water in the closing stage of the dehiscence immediately resulting in the elastic opening of the pod was determined by placing the darkened or blackened, nearly dry pods, under a glass in the sun, when they would all dehisce in an hour, and the weights before and after the experiment were then compared. The results given for the three typical legumes in the table below are closely similar. We there see that the full- grown moist pod on the plant loses on the average 59 per cent, of its weight before dehiscence occurs, and that the subsequent loss of weight is small, the total loss in drying amounting to about 62 per cent. Table illustrating the Stage in the Drying Process at WHICH the Dehiscence of Leguminous Pods occurs. Average weight of a green pod in grains. Changes in weight during drying, the full-grown moist pod being taken as loo. Green. After turn- ing black. Still closed. After dehiscence. Vicia sativa Ulex europaius Caesalpinia sepiaria . 15-0 iio-o 100 100 100 50 50 39 "o 45-0 38-0 37-6 43 -o 33"o THE DEHISCENCE OF FRUITS 289 One might mention a number of other legumes where, although the balance was not employed in this connection, it was evident that dehiscence occurred when the drying in air was far advanced, such as Guilandina honducella and Poinciana regia, the last tardily dehiscent. The transverse dehiscence of legumes into closed joints or The trans- articles also takes place towards the close of the drying process. clnS of '^' I am most familiar with this form of opening in the case of Entadapods. Entada scandens and E. polystachya. In this genus, to employ the description employed by Grisebach in his Flora of the British West Indian Islands^ the legume is " flat-compressed, the joints separating from each other and leaving a persistent, continuous border, the replum." The pods, when the drying is far advanced, break up on the plant into closed joints, each joint en- closing a single seed. With Entada polystachya this is generally preceded by the scaling off of the epidermis. If the epidermis is persistent, as happens at times, the pod remains entire. The ulti- mate liberation of the seed is affected by the decay of the joint. As far as the mode of dehiscence is concerned, the remark- The dehis- able polycoccous capsular fruit of Hura crepitans might be ^ura almost described as composed of a number of single-seeded crepitans, legumes arranged around and attached to a central axis. The rupturing of the cocci takes place in the last stage of the drying of the fruit. A fruit, seemingly dry, but displaying the earliest signs of the splitting of the cocci, was placed in a box and left in a warm corner. After the dehiscence was complete I found that the fruit had lost about 8 per cent, of its weight in the process. This fruit, which contained fifteen cocci, weighed 1233 grains, and I was able to estimate its weight in the green condition from the following data : — Actual weight of a green fruit with 14 cocci = 3062-5 grains. 16 » =3281-2 „ Estimated „ „ 15 » =3172-0 „ This reliable estimate of its weight in the green, full-grown condition enabled me to complete the history of its dehiscence 19 290 STUDIES IN SEEDS AND FRUITS by means of the balance. As shown in the results below tabulated, the stage in the drying process at which dehiscence occurred corresponds closely with that obtained for typical leguminous pods of the genera Vicia^ Ulex, and Casalpinia^ as before given. Drying and Dehiscence of a Fruit of Hura crepitans. (See above for explanation.) Full-grown green fruit (estimated). On the point of dehiscing. After dehiscence. Weight in grains. Percentage. Weight in grains. Percentage. Weight in grains. Percentage, 3172 100 1^33 38-9 1137 3SS But the resemblance between a coccus of Hura crepitans and a leguminous pod is not merely concerned with the stage at which dehiscence occurs, but also extends to the mode of the dehiscence. Each woody coccus splits along the back more or less completely into two valves, whilst at the same time it detaches itself with violence from the central axis and carries the seed away for many yards. After the rupture each valve displays a very slight spiral twist, thus indicating that the mechanism of dehiscence is similar to that of the legume, which after its sudden opening shows two spirally twisted valves. SUMMARY (i) One is apt to associate the process of dehiscence with dry fruits, and both the capsule and the legume are usually classed among dry fruits ; but the author's observations indicate that this association applies more especially to fruits like legumes that usually only dehisce after they have almost completed the drying process, and that as a rule it does not concern capsules. Whilst the legume dehisces after drying, the capsule dehisces before drying begins. (2) The author's results bring him into line with the view expressed by Professor Pfeffer that whilst the phenomena of dehiscence THE DEHISCENCE OF FRUITS 291 are physical, the development of the requisite physical conditions is a physiological problem. (3) Yet the different behaviour of capsules and legumes illustrates the difference betvi^een external and internal causes in the dehiscence of fruits. Whilst with the capsule dehiscence takes place in the ripening fruit as a relief to the tissue-strains developed by growth, in the legume dehiscence usually presents itself as a relief to the tensions developed by drying. Whilst the capsule dehisces and dries, the pod dries and dehisces, the mechanism being concerned in the first case with a living fruit and in the second case with a dead one. Amongst the capsular fruits especially studied in this connection were those of /Esculus^ Arenaria, Datura, Iris, Primula, Scilla, Stellaria, and Fiola. (4) Dealing particularly with capsules, it is considered that the first step in the relief of the strain produced by active growth is promoted by the loosening of the cohesion of the valves affected in the mellow- ing stage of the full-grown moist fruit. It is held that the normal dehiscence of an actively growing fruit is physiologically impossible, and that dehiscence could only occur after the biological connection with the parent begins to be severed in the mellowing process. The nature of the next stage depends on whether the seeds completely fill the fruit cavity, as in Iris, or only partially fill it, as in Scilla and Arenaria. In the first case the capsular walls, owing to the loosening of the connection between the valves, are no longer able to respond to the pressure of the seed-contents and gape widely during the subsequent drying process. In the second case the drying completes the loosening begun in the mellowing stage, but the valves remain more or less in position. (5) Dehiscence may occur alike in the most watery of capsules, as in Momordica, where the fruit-case holds 95 per cent, of water, and in the hardest and most ligneous of capsular fruits, as with Mahogany (Swietenia), where the water-percentage is 66, the dehiscence being carried on in each case on the same regular plan as in Iris and in the Horse-chestnut [Msculus). The dehiscence of the Mahogany fruit is especially described, as well as that of Ravenala, another type of woody capsule, in which last special questions are raised. (6) The contrast just drawn between the water-contents of the pericarp or fruit-case of fleshy and woody capsules leads to the discussion of a number of observations on diff'erent fruits, and stress is laid on the point that even the driest-looking and most ligneous of capsules hold more than 60 per cent, of water in the full-grown living state. When, therefore, we speak of a capsule as a dry fruit, we really have in our minds the dry dehisced fruit that has lost its vitality. Dehisced capsules thus appear as dead or dying fruits ; and the expression " dry fruit " has in their connection no biological significance 292 STUDIES IN SEEDS AND FRUITS for the student, the dry dehiscing capsule belonging only to his herbarium. It is futile for him to look to the structural characters of a dead capsule for evidence of adaptation to the dispersal of seeds. The living fruit alone should be his study. The fruit dies, and the mode of liberation of its seeds depends on structural characters that were developed when it was part of a living plant and could have had no possible concern with the ultimate escape of the seeds from a dead capsule. All that is purposive ends when a fruit has passed its prime. (7) In reference to the late dehiscence of legumes at or near the close of the drying process, as compared with the early dehiscence of capsules which occurs at or near the commencement of the same process, it is pointed out that if dehiscence takes place in a capsule in a living fruit, it occurs in a legume in a dead fruit, and that all the objections urged in the case of a capsule against regarding the propulsive liberation of seeds as a special adaptation apply even more forcibly to the legume. (8) The late dehiscence of legumes is then illustrated by the typical cases of Vkia^ Ulexj and Ct^salpinia. (9) Finally, the transverse dehiscence of some legumes is briefly referred to ; and the dehiscence of the polycoccous capsule of Hura crepitans is described in detail, the behaviour of the opening fruit being that of a number of single-seeded legumes around a central axis. CHAPTER XIV THE PROPORTION OF PARTS IN FRUITS The relation in weight between the pericarp and the seeds in the different stages of a fruit's history now claims our attention. This involves not merely a comparison of parts in the various states, but a detailed examination of the shrinking process which the moist, full-grown fruit has to undergo in entering the air-dried state. Amongst the first questions that offer themselves in an Therespec- investigation of this kind is that concerned with the respective o7morst"and values of the moist and dry fruit for such comparisons. The ^^7 fruits, moist condition is naturally the most important, since the fruit-covering or pericarp and the seed are still actively functioning, whilst in the dry condition the fruit-case is dead and the seed has its vitality suspended. It is true that the systematist often employs the last-named condition of the fruit ; but he has been under the whip of necessity ; and if by so doing he has at times confused the issues as regards the homology of fruits, he has been constrained by the circumstances of his investigation. Yet it behoves us all the more to keep the living fruit always in our mind. By so doing we can best hope to avoid those false analogies and deceptive contrivances which are so apt to be accepted as adaptive when we deal indiscriminately with dead and living fruits. The subject, however, is a very complex one, and Nature herself does not always aid us by bringing the several processes concerned in the maturation of fruits and seeds into a final relation with each other. Thus, as already pointed out in 293 294 STUDIES IN SEEDS AND FRUITS The method of investi- gation as illustrated in the case of Chapter XL, the seeds shrink in the ripening berry. Then, again, we shall see later on in this chapter how in the Acorn {Quercus) the seed often continues its growth after the fruit- shell has ceased to add to its weight and has begun to dry ; whilst with the Coco-nut, when the husk of the drying fruit is losing pounds in weight, the hard shell and the albumen increase considerably in amount. The method of investigation usually adopted may be best illustrated in the case of the large husky fruits of Barringtonia speciosa^ the experiments covering many weeks. (The materials, Barringtonia \x. should be remarked, were allowed to dry in my room in speciosa. . •'. ^ Grenada, except in the early stage of the drying process, when they were exposed for a few hours daily in the sun.) The first step consisted in determining the shrinking ratio of the moist, full-grown fruit as shown by the difference in weight in the moist and dry conditions. There were three ways of obtaining this result : (i) By comparing the average weights of moist and dry fruits ; (2) By drying the moist fruit in the entire state ; (3) By drying separately the parts of the moist fruit, namely, the husky pericarp and the seed. By employing these three methods the following results were obtained for Barringtonia speciosa : — The shrink- age of the fruit in the entire condition. Method. Shrinking ratio, taking the moist fruit as 100. Average of moist fruits (15 to zi ounces), and of dry fruits (3 to 5 ounces) Drying the entire fruit .... Drying the fruit in parts 100 22 100 16 100 13 On account of the considerable variation in the size and weight of the fruits of Barringtonia speciosa the results supplied by the first method could only be regarded as roughly approximate, and in consequence they were used merely as a check. The drying of the fruit in parts was deemed to give speciosa. THE PROPORTION OF PARTS IN FRUITS 295 results in excess of what would happen under natural conditions which would be best imitated in the method of drying the entire fruit. One or two reasons led me, when accepting the result of the second method, to reduce it slightly, and the shrinking ratio of 100 to 15 for the fruit was finally adopted. The next step was to determine the separate shrinking Theshrink- ratios for the pericarp and the seed of the moist fruit. In sqfarate^ spite of its husky appearance, the pericarp, like the husk of the P|gjf "f^ ^^ coco-nut, contains a very large amount of water. Two plans Barringtonia were followed here, as below described. (i) The relative weights of the husk and seed of the moist and dry fruits were obtained, and the results were applied to the shrinking ratio of the entire fruit as above ascertained. In this manner it was found that in the moist mature fruit the weight of the pericarp constituted about 80 per cent, of the weight of the entire fruit, whilst in the dry fruit it amounted to about 50 per cent. Since the fruit attains its full size far in advance of the seed, it was necessary to select moist fruits where the seed had attained the maximum weight. Barringtonia speciosa. Weight in grains. Shrinking ratios. Relation of parts. Moist. Dry. Moist. Dry. Moist. Dry. Pericarp . Seed Entire fruit 8000 zooo 10,000 750 750 1500 100 100 100 9 "4 37-5 15-0 80 20 100 50 100 (2) The actual shrinkage or loss of weight was obtained separately for the pericarp and seed of the moist fruit with the following results : — Ratio of shrinkage for the pericarp 100 to 10. „ „ „ seed 100 „ 40. Supplementary observations on individual fruits led me to the opinion that in the results of the first method the weight of 296 STUDIES IN SEEDS AND FRUITS the moist seed was too great and its estimated shrinkage exces- sive, whilst it also appeared that in the dry fruit the pericarp is as a rule rather heavier than the seed. The requisite correc- tions were not great, but they brought the various results into harmony, and the final statement accepted was as follows : — Barringtonia speciosa. Shrinking ratios. Relation of parts. Moist. Dry. Moist. Dry. Pericarp Seed .... Entire fruit . 100 100 100 9'5 40*0 15-0 82 18 100 11 Such is an example of the method that has been usually employed, alike for the green coco-nut, weighing some sixty or seventy thousand grains, and for the small berries and pods of the Elder {Sambucus) and the Gorse ( Ulex), that weigh only two or three grains. Still, as I have before remarked, these results are all concerned with the drying fruit. The more I handle these " drying " data, which bulk very largely in my note-books and have taken up a considerable portion of the time occupied in the preparation of this work, the more my interest in them dwindles. Nature offers to us the living fruit, and it is there that the real biological interest lies. If she presents us also with the dead fruit — I am of course referring more particularly to the pericarp exclusive of the seeds — we ought to regard it much as a physician would regard a patient dying from natural decay, a process which in the fruit we should term " drying up." The relative I will now proceed to deal with the results of my observa- perifarp and tions on the weight-relations of the pericarp and seeds in various seeds^in^^_^ tyP^^ °^ mature fruits before any withering or loss of weight before drying through drying: occurs. In the following: table the entire fruit begins. ..^ 1 -i-iru • ^ IS taken as 100, the proportional weight or the pericarp alone being given, that of the seeds representing the complement. This plan has been adopted with the object of letting the table tell its story by the aid of a single set of figures at the same time THE PROPORTION OF PARTS IN FRUITS 297 arranged in numerical sequence and grouped according to the type of the fruit. If the reader desires particulars relating to the average weight of a fruit, number of seeds, etc., he will find them at the end of the chapter in the table containing the elements for the determination of the drying regime of fruits. Comparison of the Weight-Relation of the Pericarp in different Types of Full-grown Fruits before Drying begins, the Weight of the Entire Fruit being taken as 100. (The fruit is here regarded as made up of pericarp and seeds. The only families indicated are Leguminosos by L. and Palmaceje by P.) P." P. I: ... L. U L. L. Relative weight of the moist pericarp, the entire fruit being taken as loo. Legume. Capsule. Berry. Drupe. Miscellaneous. Pyrus Malus (Apple) Citrus Aurantium — (a) Mandarin Orange . (d) Common ,, Citrus decumana (Shaddock) . Cocos nucifera (Coco-nut) Prunus communis (Sloe) , Achras Sapota (Sapodilla) Ribes Grossularia (Gooseberry) Sambucus nigra (Elder) . Acrocomia lasiospatha Cocos plumosa Arenga saccharifera Psidium Guajava (Guava) Momordica Charantia Opuntia Tuna (Prickly Pear) . Tamus communis . Ravenala madagascariensis LoniceraPericlymenum( Honey- suckle) Cassia fistula .... Sparganium ramosum Swietenia Mahogani( Mahogany) Hura crepitans (Sandbox-tree) Barringtonia speciosa Poinciana regia Entada polystachya Csesalpinia Sappan . Theobroma Cacao . 85 80 75 75 89 87 99 It 97 95 95 93 90 89 87 86 82 75 96* 95 92 92 90 85 * The pericarp-proportion of 96 per cent, refers to the green coco-nut only when the husk has attained its greatest development, whilst the albumen and shell are but partly formed. If we imagined a fruit where the seed and pericarp reach their greatest develop- ment together, the pericarp-proportion would be about 80 per cent. ; but nature, as shown later on in this chapter, does not supply such fruits. 298 STUDIES IN SEEDS AND FRUITS Comparison of the Weight-Relation of the F eric akp— continued. Relative weight of the moist pericarp, the entire fruit being taken as loo. Legume. Capsule. Berry. Drupe. Miscellaneous. Areca Catechu (Areca-nut) P. 74 Thespesia populnea 72 Ulex europaeus (Gorse) . i'. 70 ... Scilla nutans .... 70 ... ^sculusHippocastanum (Horse- 70 chestnut) Monstera pertusa . 66 ... Hyophorbe Verschafftii . p." 65 Oreodoxa regia p. 64 ... Caesalpinia sepiaria . L. 64 Canavalia obtusifolia L. 62 ... Mucuna urens L. 61 ... Andira inermis L. 61 Pisum sativum L. 60 Phaseolus multiflorus L. 60 Arum maculatum . 60 Hedera Helix (Ivy) 59 • • Datura Stramonium 58 __ Allium ursinum 56 ... ... Bignonia (near ffiquinoctialis) . 56 ... Iris Pseudacorus 55 Vicia sativa .... L. "50 Faba vulgaris (Broad Bean) . L. SO Guilandina bonducella . L. 50 Vicia sepium .... L. 49 Artocarpus incisa (Bread-fruit) ... ... 49 Dioclea reflexa l! 47 Cajanus indicus L. 47 ... ... Leucsena glauca L. 47 ... ... '.'.'. 1 Primula veris (Primrose) . 46 • • Mauritia setigera p! ... 46 Acacia Farnesiana . L. 46 Aquilegia (species) . 45 (fol'licle) Iris foetidissima ... 43 Canna indica .... ... ... 39 ... Ipomoea tuba .... ... 35 Arenaria peploides . 25 ... A 48 Quercus Robur (Oak)* . ^ B 35 (nut) C25 * The peculiarity in the growth of the acorn is described later on in this chapter. Here it is sufficient to observe that A represents the pericarp-relation at the time when the seed and the fruit-shell cease to grow together. After this the growth of the pericarp is arrested and the seed alone increases in weight, so that the relative proportion of the pericarp decreases, as in B and C, until the growth of the seed is in turn arrested and the maximum weight of the fruit is attained. If in the A stage the fruit weighed 50 grains, in the B and C stages, when the shell would be hardening and losing its vitality, the weight would be increased to 55 and 60 grains respectively. the table. THE PROPORTION OF PARTS IN FRUITS 299 This table illustrates the relative proportions of the pericarp Remarks on and the seeds in the weight of the full-grown living fruit, or, as we might term them, the moist relations. We should be handling a subject bristling with difficulties if we attempted at this stage of the inquiry to draw any inferences except such as are of a loose general nature from these data. Yet, scanty as they may seem, these numerical results represent a great amount of labour, since in several cases the ground was made secure by methodical observation of the fruit in its several stages. For instance, some scores of the fruits of the two species of Iris were examined, and some dozens of visits of observation were made, in the different seasons of three successive years, before I was satisfied with my investigation of the fruits and could safely fix upon the full-grown moist condition. An experience thus gained could be extended to fruits of a similar type ; but it would be unwise to make such an examination of the relative proportion of parts in a fruit without some acquaintance, either direct or indirect, with the fruit in its various stages on the plant. The more one is acquainted with the fruit and its parts and with the different states of its development, the more secure will be the ground on which to base a general conclusion. One notices that the sixty-four fruits here named princi- pally consist of legumes, capsules, and berries, the drupes being not so well represented. About one-sixth comprises fruits, all of them either berries or drupes, where the weight of the pericarp exceeds 90 per cent., or the weight of the seeds is less than 10 per cent of the entire fruit. The bulk of the fruits, where the seed-weight ranges from 10 to 60 per cent., and that of the pericarp from 90 to 40 per cent, of the entire fruit, is mostly made up of legumes, capsules, and berries ; and there is not much to choose between them in their arrangement in the scale. With drupes and berries " size," as interpreted here by " weight," does not appear to count for much in determining the place in the scale ; whilst with legumes and capsules the largest and 300 STUDIES IN SEEDS AND FRUITS the heaviest fruits have usually the smallest seed-weight. Thus amongst the drupes, the Sloe {Prunus communis)^ weighing about 30 grains, and the Coco-nut {Cocos nuciferd)^ weighing 60,000 grains, have much the same proportions. Coming near together amongst the baccate fruits are those of the Elder {Sambucus nigrd)^ weighing 3 grains, the Gooseberry (Ribes Grossularid)^ weighing 100 grains, and the Shaddock {Citrus decumana)^ weighing 14,000 grains. Although in this respect the capsules behave mostly like the legumes, the largest fruits having usually the greatest proportion of pericarp, one can point to cases where fruits widely different in size and weight have the same proportion in their parts. Thus in the capsules of the Blue-bell {^^cilla nutans)^ weighing 10 grains, and in that of the Horse-chestnut {Msculus Hippocastanum\ weighing 700 grains, the proportional weight of the pericarp is the same. With the legumes those fruits possessing the greatest proportion of pericarp, to wit, those of Cassia fistula^ Poinciana regia^ and Entada polystachya^ are certainly the largest ; and this circumstance seems to be associated with the presence of much ligneous tissue. In fact, legumes like those of Pisum sativum (Pea) and of Faba vulgaris (Broad Bean), which have a thick, fleshy pericarp, are not conspicuously high in the scale. The same indication is supplied in a general way by the capsules, since the two largest and heaviest amongst them, those of Swietenia Mahogani and Hura crepitans^ are not only the most ligneous, but rank amongst capsular fruits with the highest proportional weight for the pericarp, namely, 85 and 84 per cent, respectively. It should again be observed that the weights and other particulars concerning the fruits in this table will be found in the last table in this chapter. The history of the proportional and absolute weights of the pericarp and seeds of a fruit during its early growth, maturation, and drying now demands our attention. It is a familiar fact in the history of fruits and seeds that the pericarp or fruit-case is as a rule in its growth far in advance of the THE PROPORTION OF PARTS IN FRUITS 301 seed. This often comes under our notice in green leguminous pods, as in the Pea (Pisum sativum)^ where, although the pod may be of full size, the immature seeds within are very small and quite out of proportion to the fruit containing them, the disproportion being subsequently removed by the rapid growth of the seeds in the ripening pod. The circumstance that the earlier history of a fruit's develop- ment is mainly concerned with the fruit-case and the later with the seed is treated with some detail in a later page of this chapter. I will now therefore allude to that critical period in this sequence of events which may be pronounced the turning point in the history of the seed, that period when it has to make the choice between entering the resting state or germinat- ing on the plant. There are good grounds for holding that in most fruits Both the the seeds, which, as before remarked, are far behind the fruit- pericarp case in the earher stage of their growth, ultimately attain JJ^^^^^'Jj.y^^'^*^ maturity about the same time as the fruit. In order that the about the pericarp and the seeds may reach their full development about the same time, it is necessary that in the ripening fruit the pericarp should considerably diminish and the seeds consider- ably accelerate the rate of their growth. But, as indicated by the proportion of the parts of a fruit in different stages, as determined by the balance, there are evidently cases where the seed proceeds with its development after the pericarp has not only completed its growth but has commenced to dry. In other words, the fruit-case begins to lose its vitality before the seed enclosed has attained its maximum development. This is shown in the Acorn {Quercus) and in the Coco-nut, and, as my observations suggest, probably in the fruits of Barringtonia speciosa. It would thus promise to be not in- frequent with one-seeded fruits of these types. Before giving the data on which these general inferences are based, I should remark that this subject only came into prominence during the elaboration of my data. It was the comparison of the results obtained for coco-nuts and acorns 302 STUDIES IN SEEDS AND FRUITS that first opened my eyes to its importance. Although I can only claim to have broken the ground, the contents of the following table ought to be of interest. If figures can tell a story, these data plainly show the varying rates of growth of the pericarp and the seeds in the development of the fruit, besides illustrating their history in the drying stage when the pericarp has ceased to grow and begins to die. From this standpoint we have two types of fruits displayed in the table. The first, which is probably by far the commonest, is repre- sented by the capsules of Iris and Msculus and by the legumes of Faha^ Phaseolus^ and Entada. Here the time of the maximum growth of the seed roughly corresponds with that of the pericarp, the seed entering upon its rest-period when the fruit-covering begins to dry and lose weight. The second is represented by such closed fruits as the acorn or nut of the Oak {Quercus Rohur)^ and by the berry of Barringtonia speciosa and the drupe of the Coco-nut Palm, the two last possessing husky pericarps. Here the seed continues to add to its weight and size after the pericarp has ceased to grow and has begun to dry. The point in the case of the fruit of Barringtonia^ however, needs further investigation ; but the indications are very suggestive. Thus, in the table given below it is shown that the drying fruit, weighing 4000 grains, has a heavier seed than the full-sized moist fruit, weighing 9000 grains, that has not yet begun to dry. The same thing is brought out in the table illustrating the history of the fruit of Barringtonia speciosa given in Note 1 1 of the Appendix. It will be found there remarked under J that the seed has probably increased its weight whilst the husk has been drying. It is quite possible that future investigators will discover that the differences between the two types of fruits represented in the following table are more in degree than in kind, and that even in the prevailing type the seeds may continue to add to their weight for a little while after the fruit -case has begun to lose its vitality and to dry. There are distinct THE PROPORTION OF PARTS IN FRUITS 303 Table showing the Proportions by Weight of the Pericarp and Seeds DURING the different StAGES OF FrUITS, INCLUDING THE IMMATURE, Mature, and Dried Conditions. (The weights are in grains. In the percentage columns the entire fruit is taken as loo.) Plant-name. Parts. Green fruits with immature seeds. Ripe fruits with mature seeds. Less than half size. More than half size. Full size before drying. Early drying stage. Drying in air completed. •1 1 1 i c 1 .1 f c } i a 2 1 c it 100 Iris fcetidissimaj (capsule) 1 Iris Pseudacorus \ (capsule) 1 Pericarp Seeds Entire fruit •" 73 67 140 100 78 102 180 43 57 lOO 25 75 100 25 75 100 IS 30 45 Pericarp Seeds Entire fruit 77 100 77 23 100 98 42 140 70 30 100 13s "5 250 54 46 100 86 180 48 52 100 16 25 75 100 yEsculus Hippo- i castanum (cap- < sule, I -seeded) ( Pericarp Seeds Entire fruit 218 42 260 84 i6 100 468 132 600 78 22 100 490 210 700 70 30 100 400 200 600 66 34 100 68 100 168 40 60 100 Entada polysta- chya (legume) Pericarp Seeds Entire fruit 425 75 500 85 15 100 600 200 800 75 25 100 149 91 240 62 38 100 Faba vulgaris (5- J seeded legume) 1 Pericarp Seeds Entire fruit 170 30 200 85 IS 100 455 245 700 65 35 100 400 400 800 50 50 100 240 1': 40 60 100 48 152 200 24 76 100 52 48 100 Barringtoniaspe- ( ciosa (baccate, - I -seeded) (^ Pericarp Seeds Entire fruit 1980 20 2000 99 100 4800 200 5000 96 4 100 7380 1620 9000 82 18 100 1880 2120 4000 47 53 100 702 648 1350 Cocos nucifera, C Coco - nu t-| (drupaceous) ( Pericarp Seeds Entire fruit 9,900 1,100 11,000 90 10 100 57,600 2,400 60,000 96 4 100 19,000 7,400 26,400 72 28 100 14,040 3,960 18,000 78 22 100 '9 8i 100 Quercus Robur,) Oak (nut) 1 Pericarp Seeds Entire fruit 13 7 20 65 35 100 18 14 32 57 43 100 20 26 46 44 56 100 17 1 30 40 i 70 57 I 100 5 21 26 Phaseolus multi- f florus, Scarlet- J runner (4-j seeded legume) { Pericarp Seeds Entire fruit 133 7 140 95 5 100 200 50 250 80 20 100 i8o 120 300 60 40 100 no 55 90 1 45 200 1 100 27 48 75 36 64 100 No^e. — Although in selecting the fruits to form the same series those similar in character were chosen, minor inconsistencies occur, but the general trend of the data is to be relied on. 304 STUDIES IN SEEDS AND FRUITS indications of this in the data given for Faba vulgaris and Phaseolus multiflorus. In order to throw light on this matter, as concerning the coco-nutj I will give some of the results of observations made in Jamaica. Though the data tabulated below do not present a continuous record, the intervals can be readily filled up ; and it may be added that the general trend of results illustrated in these tables is confirmed by indications supplied by a number of other fruits in addition to those for which the record is here given. It will be seen from the tables that the drying of the full-grown green fruit is practically the drying of the husk alone, since it is likely that mould and other causes of decay usually come into play in nature before the completely air- dry condition is attained, as exemplified in column D. Indeed, planters hold that fruits kept too long do not usually dry up, but rot and decay. Whilst the drying of the husk is proceeding on the plant, remarkable changes take place in the shell and in the kernel. In a full-grown green fruit, as is well known, the shell is thin and the kernel soft and almost creamy. During the drying process the maturation of the seed proceeds. Whilst the husk is losing pounds in weight, the shell is becoming tougher and thicker, and the kernel solidifies and increases in quantity. But the increase of the kernel is much greater. Though in the green fruit its weight is rather less than that of the shell, it becomes 50 per cent, heavier as the seed ripens in the drying fruit. When, how- ever, after many months of drying, the fruit has yielded all its water to the air, except the water of hygroscopicity, which, according to the principle laid down in Chapter VII., is common to both living and dead vegetable matter, the weight of the kernel is only about the same as that of the shell. Such a completely air-dried condition, as has been observed above, would be rarely attained in nature. This last stage is more fully discussed in the explanatory remarks that follow the tables. THE PROPORTION OF PARTS IN FRUITS 305 Tables showing the Relation of Parts by Weight in Different Stages of the Coco-nut (Cocos nucifera). (Table I. is in grains; Table II. in grammes.) I. Grains. (Data obtained from individual nuts in different stages.) A. B. c. D. Full-grown green fruit. Fruit in the early stage of drying on the tree. Fruit after dry- ing for 2 to 3 months on tree (a ripe fruit.) Fruit completely air-dried (see explanation of tables). Parts. Weight. Per- centage of entire fruit. Weight. Per- centage of entire fruit. Weight. Per- centage of entire fruit. Weight. Per- centage of entire fruit. Husk . Shell . Kernel Water. 46,750 2,640 2,200 3,410 4-0 6-2 31,185 2,618 3,080 1,617 8i-o 6-8 8-0 4-2 i3>44o 3,744 5,952 864 56-0 15-6 24-8 3-6 9410 3480 3570 57-2 21*1 217 Total . 55,000 lOO'O 38,500 loo-o 24,000 loo-o 16,460 lOO'O Relation of the total weights 100 70 44 30 II. Grammes. (The proportions given in Table I are here applied to a fruit assumed to weigh origin- ally 4000 grammes, or nearly 9 lbs. ) Husk . Shell . Kernel Water. 3400 192 160 248 4-0 6-2 2268 190 224 118 810 6-8 8-0 4-2 986 275 436 63 56-0 15-6 24-8 3-6 686 254 260 57-2 2I-I 217 Total . 4000 lOO'O 2800 lOO'O 1760 lOO'O 1200 I00"0 Relation of the total weights 100 70 44 30 3o6 STUDIES IN SEEDS AND FRUITS Though these tables largely explain themselves, a few explanatory remarks are necessary. The data in Table I were obtained from individual fruits, excepting those in column D, those fruits being selected which came from the same palm and gave results similar to those supplied by others in the same stage. In Table II the percentages shown in Table I are applied to a green, full-sized fruit of average weight. It is important to notice that whilst the first table is in grains, the second is in grammes. All the observations were made during the winter 1907-8 in Jamaica. The contents of column D call for special remark. They represent the results of different experiments on the ordinary drying in air of the separate husk, shell, and kernel, applied to the fruit in the stage at which they reach their greatest development. Since the husk attains it in the green unripe fruit of full size, the drying experiment was made on the husk at this stage. In the same way and for the same reason the shell and kernel of the ripe fruit were subjected to the drying test. The following results were obtained : — Loss of weight when dried in air of husk of green coco-nut, 80 per cent. Loss of weight when dried in air of shell of ripe coco-nut, 7 per cent. Loss of weight when dried in air of kernel of ripe coco-nut, 40 per cent. To produce the completely air-dried coco-nut these losses had to be applied to the husk of the green fruit in column A, and to the shell and kernel of the ripe fruit in column C. The fruit typified in column D is therefore a coco-nut that has given up all its water to the air, or rather all the water that the air can absorb, the fruit retaining only what it would hold independent of its vitality. This completely air-dried fruit represents, accordingly, rather the result of a laboratory experiment than a process of nature, since, as previously remarked, we should only expect to find a coco-nut in this state under exceptional conditions. Such a fruit would most THE PROPORTION OF PARTS IN FRUITS 307 probably have lost its germinative capacity, as may be inferred from my observations on the embryo of the coco-nut recorded in Chapter XVIII. But although these data are concerned with a dead fruit, they bring the completely air-dried fruit of the Coco-palm into line with other fruits in the same devitalised condition, such fruits only holding the water of hygroscopicity which is common to both living and dead organised vegetable matter. If we were asked what stage in the drying of a coco-nut corresponds to that of the dried- up apple, cherry, and currant as they hang in dry weather from the tree, we should point to the completely air-dried fruit of column D. The trade husked coco-nut, such as is sold in the English markets, would be drier than the ripe fruit of column C, but moister that that of column D. If kept under dry atmospheric conditions and secured against the attacks of mould, it would lose more water. It would thus appear that what the planter calls a "ripe" A "ripe" coco-nut is a fruit that has lost in the ripening process rather '^°*^°"""*- over half its weight as a green nut. This process may be completed on the tree, or it may be continued in the stored, detached fruit which has been gathered in the early ripe condition. When we speak of the fruit as maturing its seed whilst it dries on the palm, we are not adopting the phraseology of the planter. For him a ripe nut is a fruit with a thick, solid, oily kernel ; and in practice he is less concerned with the mode of ripening than with the characters indicating his ripe fruit. Thus, the statement in Simmond's Tropical Agriculture that the seed-nut must be " fully ripe and not aged " would be full of meaning to the planter, but it might be misinterpreted by one not familiar with the Coco-palm in its home. However, the theoretical view advanced above that the The seed of seed grows whilst the fruit is drying is in accord with practical grows^whilst experience. Ripe nuts, it is advised in the work above named, iJdrSng**^^ should be allowed to dry for not less than a month after gathering before they are planted. If intended for "copra" they should not be broken up for four or six weeks after 3o8 STUDIES IN SEEDS AND FRUITS gathering. This implies that the maturation of the seed is continued after the early ripe fruit has been detached. It is during the period of storage, covering usually one or two months, that the oil in the kernel increases considerably in amount. The proportions of water and oil in the kernel vary Decrease of in an inverse relation, the water gradually diminishing^ and the the water -i j n • • i r • t • and increase Oil gradually mcreasmg as the rruit matures. In an mterestmg of the oil. compilation entitled All about the Coco-nut Palm^ which was published at Colombo in 1885, the following are given as the proportions of water and oil in the kernels of young and ripe fruits as obtained by M. Lepine of Pondicherry : — Young coco-nuts contained 90*3 per cent, of water and 2*3 per cent, of oil. Ripe coco-nuts contained 53*0 per cent, of water and 30*0 per cent, of oil. This increase of the oil in a stored coco-nut is a point emphasised by different writers in the book just quoted ; and it is one on which stress is laid in a letter to me by Mr H. Matthes, a planter of " Bacolet," Tobago. The use of the term " young " deserves a word of explan- ation. It is applied to the full-sized green fruit with shell and kernel but partially formed. The term " ripe " is applied to a fruit with a hard shell and a thick solid kernel. During the ripening process the husk loses a large amount of water, and in consequence the ripe fruit is much lighter in weight than the green fruit, though to the inexperienced eye its appearance may not be greatly changed. As confirming my method, I have compared below the proportions of parts for young and ripe fruits as given in All about the Coco-nut Palm with my own results. The data Observations are apparently all derived from M. Lepine, but they are epine. concerned with a smaller variety of coco-nut than that experi- mented upon by me in Jamaica. Since the husk and shell are not distinguished by him in the young nut, I have followed the same plan in the subjoined table. THE PROPORTION OF PARTS IN FRUITS 309 Comparison of the Proportions of Parts in Coco-nuts as obtained by lupine and guppy. Young fruit (Lepine). Ripe fruit (Lepine). Percentages obtained by Guppy. Weight in grammes. Percentage of entire fruit. Weight in grammes. Percentage of entire fruit. Green. Ripe. Husk and shell Kernel , Water . 1760 90 300 8i-9 4-2 13-9 766 434 250 53-0 30*0 17-0 89-8 4-0 6-2 71-6 24-8 3-6 100*0 Total . 2150 loc-o . 1450 lOO'O lOO'O A. B. C. D. E. F. After allowing for variation in the proportion of parts in coco-nuts of different localities, it is easy to recognise in the young and ripe fruits of Lepine the green and ripe fruits of my tables. In both sets there is a marked increase in the absolute weight and in the relative amount of the kernel in the green (young) and ripe fruits. M. Lepine's fruits were evidently selected at random ; but it is to be noted that his young fruit is about 50 per cent, heavier than his ripe fruit. The question cannot be further discussed here, but it should be added that the total amount of water in the coco-nut is dealt with in Note 27 of the Appendix, and that the subject of the embryo is treated in Chapter XVIII. As another example of those cases where the seed continues The case of to increase in size and weight after the pericarp has ceased to (Quercus" grow and is beginning to lose its vitality, I will take the fruit ^o'^"''), of the Oak {Quercus Rohur). Here the subject is bound up with that of the occasional vivipary of the acorn, or, in other words, with its germination on the tree, a matter dealt with in Chapter XIX, My observations were made at Salcombe in Devonshire in the successive autumns of 1908-19 11. The 3IO STUDIES IN SEEDS AND FRUITS method employed consisted in making observations system- atically during the five or six weeks preceding the fall of the acorn from its cupule in October. On each occasion a number of fruits were gathered from the same two or three trees, and ten were selected for examination and experiment. The vital connection with the parent plant is maintained by the attach- ment of the base of the fruit to its cupule. When the acorn begins to " brown " this attachment to the cupule begins to loosen, the result evidently of the drying of the pericarp or shell. The " browning " and drying of the shell proceed until the biological union with the cupule is severed, when at a touch the acorn falls to the ground. With regard to the table on opposite page it may be observed that the acorns of the experiment in 1 908 were larger and heavier than those employed in 19 10, a difference that will explain the divergencies in the absolute weights. The results of many observations are embodied and stated numerically ; but there is much that of necessity finds no expression in the figures. A careful examination is needed before the data here tabulated can be used legitimately, and especially is it requisite that those making use of them should know a little of the acorn and its ways. As far as is consistent with its being a tabular statement, the author has endeavoured to make it as self- explanatory as possible. But he can hardly expect his readers, whilst perusing the dry array of numerical results, to invest them with the interest they created in his mind as they gradually disclosed their story in the course of a fascinating piece of investigation. That interest they can only acquire by going to the Oak themselves and by appealing to the balance in an inquiry that should at least cover two seasons. The secret of vivipary will lie behind the results of their observations. To show how this table is to be employed I will take one of the entries, that of September 27, 19 10. It is here indicated that in a freshly gathered acorn, weighing 50 grains, the shell or pericarp weighed 19 and the seed 31 grains. In other words, as stated in the next two columns, taking the weight of THE PROPORTION OF PARTS IN FRUITS 311 Table illustrating the Growth of the Acorn (Quercus Robur) AS revealed by the Balance, during the six Weeks preceding its Detachment from the Cupule, from Observations made BY THE Author at Salcombe, Devonshire, in 1908 and 1910. (The cupule is not included in these observations.) Relative Loss of weight Average weight weight of after drying in air Date Condition and state of in grains of a the pericarp at the ordinary when attachment of the fruit to single fruit and and seed. temperature, stated gathered. the cupule. its parts (lo nuts taking the as a percentage of in each case). entire fruit as 100. the weight in the moist fruit. 1910. Total. Peri- Seed. Peri- Seed. Entire fruit. Peri- Seed. carp. carp. carp. ^Firmly attached by liv- ■) Sept. 13 \ ing tissue to the P^ i8-2 13-8 57 43 74-0 76 'O 70-0 „ 20 1 cupule ; pericarp 2 ( mm. thick and moist r 21-6 24-4 47 53 72-0 76-4 67-8 ., 27 1 Attachment looser ; be- 1 ginning to turn brown (so 19-0 31-0 38 62 64-4 72-8 59-2 Oct. 4 iss 19-3 357 35 65 6i"o 68-7 56-8 )) II Easily detached, but still a slight biological connection ; brown- ing pericarp thinner and drier 51 i5'3 357 30 70 47 "4 62-6 41 -o „ 18 Falls at a touch from the cupule ; well browned 57 17-1 39"9 30 70 47 7 57'5 43*5 „ 18 After keeping for some time and no longer losing weight, pericarp very thin, 0-3 mm. 26 19 8i 1908. Sept. 4 » 17 f Firmly attached to cupule ; pericarp thick and moist I 62 24-6 29-1 3i'4 32-9 44 47 56 53 25-0 32-3 » 30 Oct. 6 Attachment loosening ; browning ; pericarp thinner and drier 1" i8-i 52-9 25-5 74*5 45'5 48-6 ., 14 Vital connection severed ; falls at a touch from the cupule 64 ... 57-4 ,. H After keeping for some months and no longer losing weight 35 ... 18 82 the fresh fruit as lOO, the shell makes up 38 and the seed 62 parts of the whole. When this fresh acorn is allowed to dry in ordinary air, until it ceases to lose weight and exhibits merely the usual hygroscopic variations regulated by the 312 STUDIES IN SEEDS AND FRUITS atmospheric humidity, the shell or pericarp loses 72*8 percent, of its moist weight, the seed 59*2 per cent., and the fruit in its entirety 64*4 per cent. Such are the indications given in the columns of this table for the acorn gathered at this date. When we compare them with those fruits collected earlier and later, we find that during the acorn's growth the seed steadily increases its weight and decreases its water-contents long after the pericarp has ceased to grow and has begun to dry. This table will be noticed in different connections, but especially in relation to vivipary or germination on the plant in Chapter XIX. With these explanatory remarks I will now proceed to refer more in detail to the particular lesson which the data furnish us here respecting the develop- ment of the acorn on the tree, namely, the continued growth of the seed after the fruit-shell has begun to lose its vitality. This is not only the tale of the balance ; but it is the story that the acorn, as we handle it, conveys to us plainly enough in the increase in size, weight, and solidity of the seed, whilst the shell is becoming thinner and drier in the " browning " process. The tendency of a seed in some cases to continue its growth after the fruit-case or pericarp has begun to lose weight and dry, in other words, to die, finds its final expression in the germination of the seed on the plant. To put it in another way, it is a step towards vivipary. It is not by a mere coin- cidence that I am enabled to bring into touch with the viviparous habit all the three plants that have before been mentioned as illustrating the normal growth of the seed after the pericarp has begun to dry and to lose its vitality. In the cases of the Oak and the Coco-nut Palm, the connection is more or less direct, whilst with Barringtonia speciosa the implica- tion is only indirect. Thus in Chapter XIX I have dwelt upon the tendency to vivipary displayed by the nuts of the Oak {Quercus Robur), as observed by me during successive years at Salcombe in Devonshire. That the coco-nut does occasionally " sprout " on the palm came under my notice in Fiji {Plant Dispersal^ by THE PROPORTION OF PARTS IN FRUITS 313 H. B. Guppy, p. 472) ; and there is to be cited in this connection the well-known habit in the Pacific of suspending the ripe fruits from a tree by a strip of the husk and leaving them exposed to the weather until they germinate. Although there is no evidence of vivipary in the case of the fruits of Barringtonia speciosa^ there are grounds for believing that the fruits of an allied species {B. racemosd)^ which are frequently found germinating in the floating drift of the Rewa estuary in Fiji, begin to germinate whilst hanging from the trees that abound at the water-side {ibid.^ pp. 564, 575). The discovery of this peculiarity in the growth of the Thedis- acorn was made in this way. My attention was first directed theSdody ■ THE PROPORTION OF PARTS IN FRUITS 321 Pericarp and Seed. — Table of Loss in Weight — continued. Plant-name. Hura crepitans (Sandbox-tree) , Ravenala madagascariensis Swietenia Mahogani (Mahogany) Acacia Farnesiana Caesalpinia sepiaria . Cajanus indicus Canavalia obtusifolia Cassia fistula Dioclea reflexa . Entada polystachya . Faba vulgaris . Guilandina bonducella Leucsena glauca Mucuna urens . Phaseolus multiflorus Poinciana regia Ulex europseus . Vicia sativa ,, sepium mean result PaLMACE/E. Acrocomia lasiospatha Areca Catechu . Cocos nucifera (Coco-nut) * Cocos plumosa . Hyophorbe Verschafftii Mauritia setigera Oreodoxa regia Type of fruit. Woody cap- sule Woody cap- sule Woody cap- sule Legume Drupe Berry Drupe Berry Loss in weight during the drying process, the moist condition being taken as loo, the dry weight being then given. Pericarp. 100 34 100 39 100 37 [oo 35 [CO 24 100 38 (00 20 100 31 [00 31 100 25 [OO 12 '5 [oo 14 100 22 [OO 14 [oo 15 [OO 45 100 41 [00 32 too 34 [00 23 [00 21 [OO 63 [OO 15 100 55 58 Summary of some of the Results. Average for fleshy and pulpy berries ,, typical fleshy capsules Typical fleshy drupe (Prunus) . Average for legumes . ,, woody capsules Palm fruits Six kinds Eight kinds One kind Fifteen kinds Three kinds Seven kinds Seed. 100 45 100 65 100 26 100 S3 100 63 100 87 100 57 100 57 100 51 100 53 100 43 3« 100 67 This result applies to a dry, waterless coco-nut. 322 STUDIES IN SEEDS AND FRUITS The regime The regime of the drying fruit as indicated by the loss of of the drying • ^ r ^ • i i i i • n fruit. weight or the pericarp and the seeds can be numerically formulated for nearly sixty plants from the data given in the following table. If we wish to extend the inquiry by dis- tinguishing in the case of the seed between the seed proper and its coverings in this drying process, the data for several of these plants will be found in Chapter IX. Should one desire to go further and differentiate in these respects not only between the seed-coverings and the seed proper, but also in the case of albuminous seeds between the behaviour of the albumen and the embryo, the requisite data for plants like Cassia fistula^ Poinciana regia^ etc., will be found in the same chapter. In the first place, regarding the fruit as made up of pericarp and seeds, we can extract from the table all the elements necessary for determining its drying regime, namely : {a) The weight of the moist fruit ; {b) Its loss in the drying process ; (<:) The proportion of pericarp in the moist and dry fruit. In the same way, for the seed we should require : {a) The weight of the moist seed ; (b) Its loss in the drying process ; (c) The proportion of the coats in the moist and dry seed. The subject can only be illustrated here by a few examples. If the reader desires to work out the regime of other plants, he will find in the table the requisite data. THE PROPORTION OF PARTS IN FRUITS 323 Illustrations of the Drying Regime of Fruits and Seeds on the Plant. (The data concerning the proportion of the pericarp to the seeds will be found in the next table, whilst those relating to the proportions between the seed and its coverings are given in Chapter IX. By referring to the table the elements for other plants will be found.) Weight in Loss of weight Relative weight, taking grams. the entire fruit or seed as 100. Ripe moist. Dried. Ripe moist. Dried. Ripe moist. Dried. 30 f Pericarp 200 28-4 100 14-2 50 Legume-^ Seeds (2) 200 66-3 100 33'i 50 70 Guilandina I Entire 400 947 100 23-7 100 100 bonducella 1 Seed - Coats 61 177 100 29 61 58 Kernel 39 15-6 ICO 40 39 42 Entire 100 33-3 100 33'3 100 100 f Pericarp 248 50-4 100 20-3 62 45 Legume - Seeds (6) 152 61-6 100 40-5 38 55 Canavalia Entire 400 II2-0 100 28-0 100 lOO obtusifolia Coats "•^ 27 100 24 45 27 Seed Kernel 137 7'3 100 53 55 73 '^ Entire 25 iCo 100 40 100 100 f Pericarp 137-5 i6-3 100 1 1 -9 55 25 Iris Capsule -{ Seeds(6s) II2-5 48-7 100 43'3 45 75 Pseudacorus I Entire 250-0 65-0 100 26-0 100 100 (see footnote, Seed J Coats 072 0-14 100 20 40 20 P- 324 Kernel I -08 0-58 100 54 60 80 i Entire I -80 0-72 100 40 100 100 Primula veris 1 n , / Pericarp 1-84 o*3 100 16-3 46 25 (Primrose) 1 Capsule A Seeds 2-i6 0-9 100 417 54 75 [ Entire 4 00 1*2 100 30-0 100 100 r Pericarp 490 67-2 100 13-8 70 40 Capsule - Seeds (i) 210 100 -8 100 48-0 30 60 yEsculus Entire 700 168 100 24*0 100 100 Hippocastanum 1 c Coats 73*5 28 100 38-1 35 28 i Seed A Kernel 136-5 72 100 52-8 65 72 Entire 210-0 100 100 47-6 100 100 324 STUDIES IN SEEDS AND FRUITS Illustrations of the Drying Regime of Fruits, etc. — continued. Weight in grains. Loss of weight stated as a percentage. Relative weight, taking the entire fruit or seed as 100. Ripe moist. Dried, Ripe moist. Dried, Ripe moist. Dried. Ipomoea tuba - Capsule \ Seed - Pericarp Seeds Entire Coats Kernel Entire 100 7-3 9-2 i6-s S 20 2-3 ^'? 5-0 100 100 100 100 100 100 30'3 25-0 31-5 29-4 30-3 100 44 56 100 20 80 100 46 54 100 Swietenia Mahogani (Mahogany) Capsule - Seed -j Pericarp Seeds Entire Coats Kernel Entire 5100 900 6000 8-4 5-6 14-0 1869 231 2100 074 2-96 370 100 100 100 100 100 ' 100 37 26 35 9 ii 85 '5 100 60 40 100 89 II 100 20 80 100 Cassia fistula < Legume - Seed \ Seed \ Pericarp Seeds Entire Coats Kernel Entire Coats Albumen Embryo Entire 4250 750 5000 2-63 7-37 10 "GO 2-63 S"37 2 "00 10 "00 1320 300 1620 0-6 3 "4 4-0 o*6o 2-64 076 4-00 100 100 100 100 100 100 100 100 100 100 311 40*0 32-4 22-8 46*1 40*0 22-8 49-2 38-0 40*0 85 15 100 26-3 737 loo-o 26-3 537 20'O lOO'O 80 20 100 '4 100 '5 66 19 100 Arum \ T>„ maculatum | ^"^^^ " Pericarp Seeds (3) Entire 1 4*2 o'6i 2-8 1-42 7-0 1 2-03 1 100 100 100 i4'5 507 29*0 137 5i"4 14-0 60 40 100 30 70 100 Pyrus Malus (Apple) Berry - Pericarp Seeds (10) Entire 993 136-4 7 ! 3*6 1000 i40'o 100 100 100 99-3 07 lOO'O 97 '4 2-6 lOO'O Note that in some cases, as in /rz's Pseudacortis, there is a slight difference between the loss of weight of the entire seed, as given under the fruit and under the seed. This is due to the aborted seeds being included in the first case. THE PROPORTION OF PARTS IN FRUITS 325 Table containing the Elements for the Determination of the Drying Regime of Fruits and Seeds as illustrated in the Pages preceding. (The only natural orders indicated are the PalniaceK by P. and the Leguminosse by L, before the plant's name. Further details will be found in Note 28 of the Appendix for those plants where N. follows the name.) Weight- Loss of Average relation of weight of weight Number the pericarp. fruit when T,pe ofa of taking the air-dried. moist seeds entire fruit taking the fruit. fruit in in fruit tested. as 00. moist fruit as 100, the grains. dry weight Moist Dry being then fruit. Fruit. given. Pyrus Malus (Apple), N. Berry 900 10 99 '3 97 "4 100 14 Citrus Aurantium (Common ,, 2400 S 99 '3 98-3 100 19 Orange), N. Citrus Aurantium( Mandarin ,, 1900 5 99 "0 Orange) Citrus decumana (Shad- ,, 14,000 80-85 97 dock), N. P. Cocos nucifera (Coco- Drupe 60,000 I 96 78 100 30 nut), N. Ribes Grossularia (Goose- Berry 120 40 95 80 100 15 berry), N. Prunus communis (Sloe), N. Drupe 30 I 95 90 100 27 Achras Sapota (Sapo- dilla), N. Sambucus nigra (Elder), N. Berry 1000 5 95 ,, 3"S 3 or 4 93 60 100 21 P. Acrocomia lasiospatha, N. Drupe 55° I 92 85 100 36 P. Acrocomia sclerocarpa I 89 ... P. Cocos plumosa, N. 100 I 9i'6 91*6 100 63 100 50 P. Arenga saccharifera, N. ,, 600 I 90 80 Psidium Guajava (Guava) . Moniordica Charantia, N. . Berry 600 300 90 Capsule 500 15 89 51 100 15 Opuntia Tuna (Prickly Berry 800 90 89 70 100 18 Pear), N. Tamus communis ,, 14 4 or 5 87 50 100 iS Raven a la madagasca- Capsule 700 10 or 12 87 80 100 42 riensis, N. Lonicera Periclymenum, N. Berry S S 86 66 100 25 (Honeysuckle) L. Cassia fistula Legume 5000 95 85 80 100 3Z*4 Sparganium ramosum, N. . Drupe I 85 84 100 45 Swietenia Mahogani Capsule 5800 60 85 89 100 35 (Mahogany), N. Hura crepitans (Sandbox- ,, 3000 14 84 80 100 36 tree), N. Barringtonia speciosa, N. . Berry 9000 I S2 52 100 15 L. Poinciana regia, N. . Legume 3500 40 80 80 100 45 L. Entada polystachya, N. ,, 800 14 75 62-5 100 30 L. Caesalpinia Sappan . ,j 260 4 75 75 100 50 Theobroma Cacao, N. Berry 7000 45 75 326 STUDIES IN SEEDS AND FRUITS Table containing the Elements, etc. — continued. Average Weight- relation of Loss of v^eight of fruit w^hen vireight Number the pericarp, air-dried. ^T ofa moist of seeds taking the entire fruit taking the moist fruit fruit. fruit in in fruit tested. as 1 00. as 100, the dry weight grains. Moist fruit. Dry fruit. being then given. P. Areca Catechu (Areca- Berry 250 I 74 53 100 32 nut), N. Thespesia populnea . Baccate capsule 230 13 or 14 72 45 loo 33 L. Ulex europseus . Legume 3 4 or 5 70 66 100 43 Scilla nutans Capsule 10 20 70 20 100 25 /Esculus Hippocastanum, ,, 700 I 70 40 100 24 N. (Horse-chestnut) Monstera pertusa, N. Berry S I 66 30 100 20 P. Hyophorbe Verschafftii ,, iS'5 I ''S 24 100 40 P. Oreodoxa regia, N. . ,, 15 I 64 56 100 66 L. Csesalpinia sepiaria . Legume 100 5 64 48 100 32 L. Canavalia obtusifoHa . ,, 400 6 62 45 100 28 L. Canavalia gladiata, N. ,, 6 ... 45 L. Canavalia ensiformis, N. . ,, 14 44 L. Mucuna urens . ,, 1000 3 or 4 61 31 100 28 L. Andira inermis . ,j 135 I 61 53 100 40 L. Pisum sativum (Pea) . ,, 250 7 or 8 60 20 100 20 L. Phaseolus multiflorus ,, 300 4 60 36 100 25 (Scarlet-runner) L. Phaseolus (tropical species) ,, 3 ... 34 Arum maculatum Berry 7 3 60 30 100 29 Hedera Helix (Ivy) . 1! 5 3 59 52 100 40 Datura Stramonium . Capsule 300 600 58 24 loo 30 Bignonia, near aequinoc- Siliqui- 1240 55 56 38 100 40 tialis, N. form capsule Allium ursinum Capsule 2-4 3 56 29 100 33 Iris Pseudacorus ,, 250 80 55 =^5 100 26 L. Vicia sativa Legume 15 10 50 40 100 37 L. Vicia sepium ,, 6 3 01^4 50 40 100 42 L. Faba vulgaris (Broad ,, 800 5 50 24 100 26 Bean), N. Artocarpus incisa (Bread- Aggregate 12,000 60 50 fruit) L. Guilandina bonducella Legume 400 2 50 30 100 24 L. Dioclea reflexa . ,, 1800 4 47 36 100 41 L. Cajanus indicus ,, 40 4 47 45 100 40 L. Leucsena glauca, N. . >) 100 24 47 35 100 30 Primula veris( Primrose), N. Capsule 4 62 46 25 100 30 P. Mauritia setigera, N. Berry 1000 t 46 45 100 56 Aquilegia, N. . . . Follicular capsule 10 100 46 24 100 45 L. Acacia Farnesiana Legume 150 20 46 40 100 40 Iris foetidissima . Capsule 180 40 43 34 100 25 Canna indica, N. )> 100 24 39 II loo 47 Ipomoea tuba . 100 4 35 20 100 25 Arenaria peploides . ,, 8 10-12 25 9 100 41 Quercus Robur (Oak), N. Nut 60 35 19 100 40 THE PROPORTION OF PARTS IN FRUITS 327 Table supplementary to the Preceding and Relating only to the Pericarp-Relation of the Air-dried Fruit, that for the Moist Living Fruit being not ascertained, except in the Case of COCOS, WHERE TWO SpECIES FROM THE PREVIOUS TaBLE HAVE BEEN ADDED FOR THE PURPOSE OF COMPARISON. (See the preceding table for explanation, N. =Note 28 of the Appendix.) Weight-relation Average weight of the pericarp, Type of fruit. of a dry fruit taking the in grains. entire dry fruit as 100. L. Abrus precatorius (4 or 5 seeds) . Legume 8 25 L. Albizzia Lebbek (7 or 8 seeds), N. ,, 18 55 Anacardium occidentale (Cashew) Nut 95 66 P. Bactris (species of), N, Drupe 170 67 L. Bauhinia(i8 seeds) . Legume 300 78 Cakile aequalis, N. . . . Lomentaceous siliqua o'7 71 ,, maritima, N. . do. 0-9 70 P. Caryota (species of) with 2 Drupe ? 80 26 seeds, N. P. Cocos schizophylla ? N. . „ 500 65 „ nucifera (Coco-nut), N. . ,, 16,000 78 ,, plumosa, N. . . . ,, 63 91-6 L. Cynometra (species of) with i Legume 200 26 seed P. Elaiis guineensis, N. . Drupe 120 85 L. Erythrina corallodendron (7 or 8 Legume 40 41 seeds), N. L. Erythrina indica (10 seeds) . . ,, 140 35 Gossypium barbadense (i6 Capsule 31 28 seeds) N. Gossypium hirsutum (23 seeds) N. ,, 50 24 Hibiscus esculentus ( 1 carpels) N. 100 30 Kleinhovia hospita (5 seeds) _, 1-6 50 P. Licuala grandis .... Berry 10 42 P. Livistonia (species of) ^5 49 Moringa pterygosperma ( 1 5 seeds) Siliquiform capsule 170 62 P. Prestoea montana Berry 18 30 Ricinus communis (Castor-oil), N. Coccous capsule 12 50 Rumex Nut o-o8 30 P. Sabal umbraculifera . Berry 10 44 Saccoglottis amazonica (2 seeds) Drupe 330 94 Scirpus maritimus Nut 0-07 50 Terminalia Catappa, N. Drupe no 95 Viola tricolor (40 seeds) Capsule 15 50 32 8 STUDIES IN SEEDS AND FRUITS SUMMARY (i) This chapter deals with the weight-relations between the peri- carp (fruit-case) and the seeds in the different stages of a fruit's history, and it involves a special inquiry into the drying of the mature fruit. (2) After pointing out that for purposes of comparison the fruit in the moist living condition is far more important than in the dried dead state, the author illustrates his method of investigation by taking the fruit of Barringtonia speciosa (p. 294). (3) The relative weights of pericarp and seeds in sixty-four mature fruits before drying begins are then discussed, and the results of the author's observations are tabulated (p. 297) (4) With regard to the influence of the type of fruit on this relation, it appears that fruits possessing the greatest proportion of pericarp, that is to say, where the weight of the fruit-case is more than 90 per cent, of that of the whole fruit, are either berries or drupes (p. 299), (5) In determining the weight-relation, " size " counts for little with drupes and berries. Thus about 2000 drupes of Prunus communis (Sloe) make up the weight of a single green fruit of the Coco-nut Palm, and nearly 5000 Elder berries [Sambucus nigra) are required to weigh down an average fruit of the Shaddock [Citrus decumana). Yet the proportions of pericarp and seeds are much the same in the two cases. With legumes the largest and heaviest fruits have the smallest seed- weight, a rule usually but not always applicable to capsules also (p. 300). (6) The history of the weight-relations between the fruit-case and the seeds in different stages of the fruit's development is then dealt with, and it opens up a study of the growing, ripening, and drying fruit (p. 301). (7) It is observed that whilst in the younger fruits the growth of the fruit-case is far in advance of that of the seeds, both fruit-case and seeds usually reach maturity about the same time. (8) Yet there are fruits where seeds continue their growth after the fruit-case has begun to dry. This is exemplified in the fruits of the Coco-nut Palm [Cocos nucifera\ of the Oak [Quercus Robur)^ and probably also of Barringtonia speciosa. The cases of the two first- named are discussed with much detail, the results largely of the indications of the balance. That of the coco-nut, which was studied by the author in the West Indies, is first dealt with, and it is shown that the seed grows markedly whilst the husk is drying. That of the acorns of the Oak, which was investigated by the author in Devonshire, is treated at length, and it is established that the seed of the acorn continues its growth after the shell has begun to dry (pp. 301-314). THE PROPORTION OF PARTS IN FRUITS 329 (9) This occasional tendency of a seed to continue its growth after the fruit-case or pericarp has begun to dry, or, in other words, to die, finds its final expression in the germination of the seed on the plant. To put it in another way, it is a step towards vivipary. (10) Reference is then made to the principle established by the results tabulated in this chapter, that as the fruit grows, matures, and dries, the proportion by weight of the pericarp decreases and that of the seeds increases. This is merely stating in other words the familiar fact of observation that the earlier history of a fruit's development is mainly concerned with the fruit-case and the later history with the seed. As illustrating this principle for the legumes, the pods of the Scarlet-runner [Phoseolus multiflorus) and of the Broad Bean {Faba vulgaris) are specially discussed, the fruits of Iris Pseudacorus being taken for the capsules, whilst the acorn is also referred to. Small departures from this rule are noticed in the cases of the coco-nut and of the fruits of Iris fcetidissima and Barringtonia speciosa (p. 315). (11) Attention is then called to the significance of the fact that in the drying fruit the pericarp loses far more water than the seed, the result being that whilst the fruit-case dies the seed Hves. Whilst on the average the pericarp of the moist fruit loses between 75 and 80 per cent, of its weight, the seed as a rule loses only half its weight (p. 318). (12) When we compare the difi:erent fruit-types we find that it is in the behaviour of the pericarp in the drying process that most varia- tion is displayed, the seeds being much more constant in this respect. Thus on the average fleshy capsules and pulpy berries lose 86 per cent, of the weight of the moist pericarp when drying on the plant, fleshy drupes 75 per cent., legumes 73 per cent., woody capsules 63 per cent., and palm fruits 62 per cent (p. 318). (13) A method of numerically formulating the drying regime of fruits is then described and illustrated, and in a special table are given the elements of this determination for nearly sixty plants (p. 322). CHAPTER XV THE RELATION BETWEEN THE NUMBER OF SEEDS AND THE WEIGHT AND SIZE OF THE FRUIT The relation The relations between the number and weight of the seeds number of o^ the one hand and the total weight of the fruit and the weight"oVa^ proportional weight of the pericarp on the other offer an fruit, interesting stydy. Although we shall at first treat the subject on its own ground, it will soon be perceived that in so doing we are ignoring important determining influences. Foremost among such influences stands that connected with the abortion of ovules before and after fertilisation, this distinction in time with regard to fertilisation being pregnant with results as regards the future of the fruit. But the subject of the failure of ovules and seeds is dealt with in Chapter XVI., and here it will be only incidentally noticed as we proceed with the discussion. (a) in the The living fruit, that is to say, the green, moist, full-grown iving fruit, fj-uit with large, soft, uncontracted seeds, first claims our attention. This is but natural, and indeed the main interest of the withered or air-dried fruit should chiefly lie in its ability to aid us in our studies of the living fruit. Although the data below tabulated are scanty, their acquisition has often involved a good deal of labour, and a large amount of material had usually to be gone over to get a few results. It is necessary, for instance, to select only full-grown moist fruits that show no signs of drying ; but of these a large number have frequently to be rejected on account of defective seed-development, or of a lack of uniformity in the size of the seeds. 33° SEED-NUMBER AND FRUIT-SIZE 33"^ Table showing the Relation between the Number of Seeds in Capsules and Legumes, and the Proportional Weight of the Pericarp in Full-grown Fruits before Drying begins. Proportions of Ranges. pericarp and JNuniDcr Average seeds, taking Aver- Number of seeds. of fruits ex- amined. total weight fruit in grains.* the weight of entire fruit as 100. age weight of a seed in grains. Pericarp pro- portions. Total weight of fruits in Peri- Seeds. grains. carp. ^sculus Hip- pocastanum (capsule). 2 8 6 650 920 72 67 28 33 182 150 64-78 61-75 450- 900 600-1140 65 46 54 27 Canna indica 85 41 59 2-6 (capsule). 100 39 61 27 25 108 37 63 27 Iris f 37-50 161 62-5 37*5 I "4 6o'2, 647 I37,"i86 Pseudacorus 67-77 265 54 '0 46-0 I "9 50-60 246-288 (capsule). U4-86 353 55-5 44 "5 I '9 Si-4,59"5 35o> 357 Iris i '7 132 46 "0 54-0 2-6 foetidissima \ '^ 172 44-6 55 '4 27 ... (capsule). ill 196 207 44-1 41-1 55-9 58-9 2-8 2-8 ... Csesalpinia i 3 216 74 '5 25-5 18-3 Sappan 1 4 240 75-6 24-4 14-8 (legume). I 5 272 74'9 25-1 136 Diocleat i reflexa \ ^ 1341 46-8 53"2 238 (legume). 1 ' 1814 47-3 527 239 Guilandina bonducella { ' 260 68-0 32-0 81 -o (legume). 1 ' 330 56-0 44 -o 73-0 i S 199 61 39 15 '4 Pisum X ^ 167 64 36 lO'O sativum 6 138 58 42 97 ... (legume). 7 198 62 38 io"9 I 9 290 57 43 13-9 * In the case of single fruits the actual weight is given. t The average length of the 3 -seeded pods is 127 millimetres, and of the 4-seeded pods 152 millimetres. + Two sets of pods are here represented, the first and last belonging to one season, and the others to another season. For capsules with a few large seeds, the behaviour of that of Capsules and D •' legfumes. the Horse-chestnut {jEscuIus Hippo castanuni) would probably be typical. Out of the original six ovules, not more than two or three develop into mature seeds ; and fruits with a single 332 STUDIES IN SEEDS AND FRUITS seed are not infrequent. The capsules with one and two seeds are best suited for comparison in this respect. In passing from the single to the double-seeded fruit of full size and showing no signs of dehiscence or drying, fruits with the seed or seeds in soft white coverings and the embryo normally developed, we find (i) An increase of about 40 per cent, in the total weight of the fruit ; (2) A decrease in the proportional weight of the pericarp from 72 to 67 per cent ; (3) A decrease of about 1 8 per cent, in the weight of each seed. With many-seeded capsules, as with those of Canna and /nV, we also find that a marked increase in the total weight of the fruit and a gradual decrease in the relative weight of the pericarp accompany the additions to the number of seeds, but the average weight of a seed seems usually to remain unchanged. All these are merely indications, and appeal for confirmation will be made subsequently to results estimated from the dry fruits. The data for the living legume are too scanty, and will have to be supplemented by results obtained from dried fruits. We now come to the question of the use of the dried or withered fruit in determining the relation between the number of seeds and the proportional weight of the pericarp. Naturally such an investigation is far easier with dried or withered fruits than with moist mature fruits that have not begun to lose weight in drying. We can furnish ourselves from the plant with abundant materials in all stages of drying, and where further drying is needed it can be readily accomplished at home. But the case becomes very different when we make use of moist fruits. Here it is necessary to select only those fruits which a previous study has shown to have reached their maximum growth, but have not yet begun to dry ; and this is not always so easy as it seems. Then, again, the question as SEED-NUMBER AND FRUlT-SlZE 333 to which of the two sorts of fruits offers the best materials, the moist or the dry, has to be answered with another query- as to whether the two results would be really comparable and would possess a biological importance similar in degree and in kind. As far as the dry fruit is concerned, it is requisite to remember that we are here actually determining the relation between a resting seed with its vitality suspended and a dried- up and dead fruit-case. Whether it is a shrivelled berry or a shrunken drupe, as in Ribes and Prunus, or a dried dehiscing pod, as in Ficia, or a withered capsule, as in Iris and Canna, or a woody dry capsule, like that of the Mahogany {Swie tenia) ^ that on account of the abundance of ligneous tissue retains the form of the moist fruit, makes no difference. The mere retention of form in some dried fruits, as in certain kinds of legumes and capsules, and its complete loss in others, as in most drupes and berries, are merely accidents in the history of the fruit. The investigator does not recognise the distinction between moist and dry fruits in the living condition. For him all fruits are moist in the living state ; and if, after the drying up and death of the fruit-case, the form of the living fruit is to some degree preserved, he will avail himself of the circumstance only in so far as it assists him in his studies of the living condition. It would therefore appear that the moist and the dry fruit i he data are not mutually comparable, and that the only comparison of drled-up ^ any biological value is one which enables us to reconstitute the ^^^^^^ ^^ living fruit, when the only materials at our disposal are its service when dried-up remains. The loss in weight which the pericarp and the living seeds of the living fruit undergo when dried in ordinary air- condition, conditions can be ascertained by experiment, and the results can be applied to the dry fruit. These shrinkage ratios, being constant for the same species and independent of the size of the fruit, do not, when applied, interfere with the progressive scale of the weight-relations between the pericarp and the seeds. In this sense, therefore, the data supplied by the dry 334 STUDIES IN SEEDS AND FRUITS fruit can be utilised ; and as long as the means of converting them are available, the actual conversion may be at times dispensed with. The validity of this use of the dry fruits is brought out in the two following tables, which contain the results of observa- tions in Grenada on a considerable number of the dry legumes of heuc^ena glauca and Albizzia Lebbek. In the first table the results for the dry legumes are alone given. In the second table these results are compared with those for the moist fruits as far as the relative weights of the pericarp and seeds are concerned. It should, however, be added that whilst the shrinking ratios for the legumes of Leucana glauca have been ascertained by experiment, those for Albizzia Lebbek have been -Table comparing the Relation between the Number of Seeds and the Weight, Length, and Pericarp Proportions OF the Dried Legumes or Pods of Leuc^na glauca and Albizzia Lebbek. Proportional weight Average weight Aver- age weight of a seed in grains. of pericarp and Number Number Average m grams. seeds, takmg the of seeds in a pod. of pods ex- amined. length of a pod entire pod as 100. in milli- metres. Entire pod. Peri- carp. Seed con- tents. Peri- carp. Seeds. Entire pod. IO-I2 6 114 mm. 127 4 '9 7-8 072 38-6 61-4 ICO Leucsena \ 13-17 II 130 >, 14-8 5-2 9-6 0-68 35-1 64-9 100 glauca 1 20-21 7 173 ,. 24-6 «7 15-9 077 35-4 64-6 100 22-26 9 19s .. 29-0 lo-i 18-9 0-78 34-8 65-2 100 I 10 142 ,, 8 -So 6-52 2-28 2-28 74-1 25-9 100 2 14 180 ,, 1370 911 4-59 2-29 66-5 33-5 100 3 13 193 M 1778 11*22 6-56 2-19 63-1 36-9 100 4 13 218 „ 2275 13-99 8-76 2-19 bi-5 38-5 100 5 12 228 „ 27*45 16-80 10-65 2-13 6i-2 38-8 100 Albizzia J 6 10 236 „ 3i"52 18-99 12-53 2-09 602 39-8 100 Lebbek 7 8 251 ., 35'2o 19-58 15-62 2-23 55-6 44 '4 100 8 8 262 ,, 3977 22-30 17-47 2-18 5^-1 43-9 100 9 9 254 „ 40-07 22-05 18-02 2-00 55-0 45-0 100 10 11 290 ,, 49-39 26-92 22-47 2-24 54-5 45-5 100 II 8 282 „ 51-26 28-50 22-76 2-07 55-6 44-4 100 12 10 294 „ 59-42 34-43 24-99 2 -08 57-9 42-1 100 SEED-NUMBER AND FRUIT-SIZE 335 B. — Table comparing the Relation between the Number of Seeds and the Weight of the Entire Fruit and its Parts IN the Moist and Dry Condition for the Legumes of Leuc^na glauca and Albizzia Lebbek. (The shrinking ratios for the last-named have been estimated as explained in the remarks below.) Relative weights, taking the Average weight entire pod as 100. Number of seeds in grains of entire pod. in a pod. Pericarp. Seeds. Moist. Dry. Moist. Dry. Moist. Dry. J IO-I2 41-8 12-7 53-4 38-6 46-6 61-4 Leuc«na glauca < 13-17 47-6 14-8 49-6 35-1 50-4 64-9 20-21 79-3 246 49-8 35'! 50-2 64-6 ( 22-26 93-2 29 49-2 34-8 50-8 .65-2 I 3178 8-80 82-1 74-1 17-9 25-9 2 i^'K 13-70 76-1 66-5 23-9 33-5 3 61-28 17-78 73-2 63-1 26-8 36-9 4 77-86 22-75 71-9 61-5 28-1 38-5 S 93-82 27-45 71-6 6i-2 28-4 38-8 Albizzia Lebbek • 6 107-28 3152 70-8 60-2 29-2 39-8 7 117-37 35-20 66-8 55-6 33-2 44-4 8 135-87 39-77 65-7 56-1 34*3 43-9 9 133-25 40-07 66-2 55-0 33-8 45-0 10 163-85 49-39 657 54-5 34-3 45-5 M 170-90 51-26 66-7 55-6 33-3 44-4 12 200-19 59-42 68-8 57-9 3. -2 42-1 The shrinking ratios employed for Leucczna glauca are 100 - 22 for the pericarp, and 100-40 for the seeds; the pericarp thus losing 78 per cent, of its weight in the drying process, and the seeds 60 per cent. The ratios used for Albizzia Lebbek are 100-25 for the pericarp, and 100-60 for the seeds. estimated, as I had no opportunity of experimenting on the living fruits. On the average, in moist mature legumes of this character the pericarp loses about 75 per cent, of its weight during the drying process, whilst the weight of the seeds is diminished by about 60 per cent. ; and these are the ratios applied to the dry Albizzia pods. The estimate for the moist fruit thus obtained cannot be far wrong, and since any error would uniformly affect all the results, the relations between them would be unaffected. 1,^6 STUDIES IN SEEDS AND FRUITS Inferences to We find the answer to the question suggested by the from^the" columns of Table A as to the relative values of the data tabulated afforded by moist and dry legumes in the contents of Table B. given. Here we see that although the relative proportions by weight of pericarp and seeds are on different planes, the pericarp being proportionately lighter and the seeds proportionately heavier in the dry than in the moist fruit, the progressive changes of relation are the same in both. We thus arrive at the following conclusions with regard to the relations between the number and weight of the seeds, the weight and length of the pod, and the relative proportions of the pericarp, in the living legumes of Leuc^na glauca and Albizzia Lebbek. (a) It is only in the few seeded pods of each plant that the legumes follow the principle of capsules, where not only a marked increase in the weight and size of the entire fruit, but a gradual decrease in the relative weight of the pericarp accompanies the additions to the number of seeds. (^b) But this principle has a limit in each case. In Leucana glauca it is restricted to the shorter pods with less than a dozen seeds. In Albizzia Lebbek it is confined to the shorter pods with less than seven seeds. Beyond these limits in both cases, as the seeds increase in number and the pods increase in length, the relative proportions of the pericarp and the seeds remain about the same. (c) The average weight of a single seed varies but little, whatever may be the number of seeds or the length of the legume. A slight increase of weight is indicated in the case of the longer pods of Leucana glauca ; but I do not imagine that this small difference would have been sustained if the materials had been more abundant. Theindica- The weight-relations of the pericarp in some other dried dried ° °* ^^ legumes, as shown by those of Vicia^ Ulex^ and Erythrina in legumes. |-|^g following table, give no very definite results, the tendency of the decrease in the relative weight of the pericarp, as the fruits increase in size and weight and the seeds in number, being but slight. However, in Guilandina bonducella^ which has. SEED-NUMBER AND FRUIT-SIZE 337 usually one or two large seeds in each pod, there is a decided repetition of the behaviour of capsules in these respects, thus supporting the conclusion derived from the legumes of Leucana glauca and Alhizzia Lebbek that it is the pod with few seeds that is most likely to follow the principle of the capsule. But even in such a case this seems only to apply to legumes with a few large seeds. More often it would be difficult in small pods containing only a few seeds to discover any such relation. Thus there is certainly but little to be made out of the results for Abrus precatorius given in the following table, in the columns of which data for larger legumes, like those of Canavalia obtusifolia^ will be found, which are equally indeterminate in their indications. Doubtless the scantiness of the materials is partly responsible for this ; but we might have looked for some more consistent results than are given for dry legumes in the subjoined table. There are evidently some disturbing influences at work that cause this irregularity in the relations between fruit and seed in these legumes. These influences, as will subsequently be shown, are connected with the failure of seeds and the abortion of ovules. One disturbing cause is needlessly brought into action when we employ indiscriminately legumes that have dried on the plant and those that have dried in an experiment. A serious eflfect may be thus produced, a matter discussed in Note 16 of the Appendix. This influence has, however, been avoided in my own results by using for each plant only fruits dried in the same way. Coming to the evidence of dried capsules, like those of Iris Pseudacorus and Iris fix tidissima^ we find the features of the moist capsule reproduced in the following table. As the fruit increases in size and weight and in the number of its seeds, the proportion of the pericarp steadily decreases, but, unlike the moist fruits, there is an increase in the weight of the individual seed. Other types of capsules, such as those illustrated by the siliquiform fruits of Moringa pterygosperma^ seem to follow the same rule. 22 338 STUDIES IN SEEDS AND FRUITS Table showing the Relation between the Number of Seeds and THE Proportional Weight of the Pericarp in Dry Legumes and Capsules. (In the case of all the legumes and of the Moringa capsules the fruits had dried on the plant ; whilst the Iris capsules were allowed to dry slowly in my room.) Vicia sepium (legume). Ulex europseus (legume). Erythrina corallo- dendron (legume). Guilandina! bonducella (legume). * Iris Pseu- dacorus (capsule) Iris foeti- dissima (capsule). Moringa pterygo- sperma (siliquiform capsule). Abrus pre- catorius (legume). Canavalia obtusifolia (legume). Number of seeds in a fruit. Number of fruits amined, Average length of a fruit in milli metres. 30 35 '47 228 266 Average weight in grains. Entire Peri- fruit. carp. 2-36 0-93 3*99 '"55 I '23 0-83 1-24 0-82 1-47 o"93 13*3 57 27 "3 117 41-6 17-0 53*3 88-0 20-8 27-2 36-2 14-2 43*3 62-3 14-6 i6-3 70 'O 80 -o 17-0 20 -o 93*3 23-0 32-4 587 12-6 17-9 ;^s 84 no rs 2*0 8-0 2-2 7*2 7'3 :? 9-0 27 9-0 2-1 133 54 123 156 146 1^ 63 Seed con- tents. I 43 2-44 0*40 0*42 0-54 7-6 15-6 24*6 32*5 5o-8 22 "o 287 46 'O 53-0 60 'o 70-3 19-8 40-8 3-6 5-8 5-55 5 '5 6-3 6-9 79 66 96 83 Aver age weight of seed in grains, o 33 0-27 O'OS 0*09 32*5 30 '4 o'6o 0-57 070 072 071 073 072 o 90 40 4*3 14 I "4 1*4 1*3 I "4 13-2 II'O I2'0 10-4 Proportional weight of pericarp and seeds. Peri- Seeds. carp. 39*4 606 38-9 6i-i 67-6 32-4 66-1 33"9 63-6 36-4 42-9 57-1 42-9 57"i 40-9 59-1 39*0 61 -o 30-9 69'i 39-2 6o-8 337 66-3 26-2 73-8 24-3 757 25-0 75 "o 247 75-3 38-9 6i-r 30-5 69-5 64-1 35 '9 6i-8 38-2 357 64-3 27-5 72-5 22-9 77-1 247 75'3 30-0 70*0 23-3 767 40 6 59 '4 4<'-3 537 38-5 6i-5 43'i 56-9 Entire fruit. 100 100 100 100 100 100 100 These pods were all gathered in the dried dehiscing condition at the same time from the same plant. SEED-NUMBER AND FRUIT-SIZE 339 Summing up the indications of the influence of the number Summary of seeds on the proportion of parts in capsules and legumes, tionsofthe we find that the capsule, as illustrated by the fruits of /m, j.he"number Canna. and ^sculus. in response to the augmentation of the of seeds on t r 1 • 1 .• 1 r u. • T-u the fruit m number ot seeds acquires a relatively lighter pericarp. 1 he the case of fruit increases in size and weight ; but this increase is due ffgumes.*" more to the seeds than to the fruit-case. The legume, as typified by the fruits of Leucana glauca and Albizzia Lebbek^ follows the principle of the capsule in the few-seeded pods ; but in the many-seeded fruits it preserves a fairly constant relation between the weight of the pericarp and the seeds, the pod increasing regularly in length and weight as the seeds increase in number. But the other results obtained for legumes often give no definite clue to any such relations, except in the case of pods with a few large seeds, as in Guilandina bonducella^ where the principle of the capsule is indicated by both the moist and the dry fruits. This lack of relation is partly due to insufficiency of materials ; but in the case of small pods with a few seeds, like those of Abrus and Ulex^ there is evidently some disturbing cause which largely counteracts the display of a connection between the number of seeds and the size of the pod. The circumstance of the growth of the legume being linear, rather than tangential, as in the capsule, may help to explain why the first-named is more irregular in its behaviour. For this reason also the legumes would be more liable to be affected by the abortion of ovules and the failure of seeds. As regards the alterations in the average weight of a seed The degree with the accession to the number of seeds and with the increase of the weight in the weight and size of the fruit, the results seem to justify JruJJo^f^'" the following conclusions : — different size {a) In the case of many-seeded capsules, like those of Iris and Canna^ we find that as the fruit increases in size and the seeds in number, the seeds of Iris increase their weight, whilst those of Carina remain unchanged. But when a capsule matures only one or two large seeds, as with the Horse- 340 STUDIES IN SEEDS AND FRUITS chestnut {Msculus Hippocastanum)^ the seeds of the two-seeded fruits are smaller and lighter than those of the single-seeded fruits, in this case the decrease in weight being about i8 per cent. (^b) With regard to legumes the indications supplied by the 126 dry pods of Albizzia Lebbek are particularly valuable, all of them being obtained at the same time from the same tree. Here the seeds range in number from i to 12 ; and for each number of seeds from 8 to 13 pods were employed. Here it will be seen that whilst the legume doubles its length, and increases its weight sevenfold, the average weight of a seed changes but little. The extreme range of the variations amounts to only about 13 per cent, of the average weight of the seed, the variations themselves being evidently fortuitous in character. The testimony of the legumes of other plants named in the tables cannot carry much weight, because the materials are insufficient. But I would gather from the cases of Ulex europaus and Abrus precatorius that with small pods the seeds keep their weight as they increase in number and the pods increase in size. The study of And now, before leaving this subject of the relation between the between the number of seeds and the weight and size of seSsand^the ^^^ ^^^^^ ^^'^ ^^^ parts, it is necessary to point out that, proportions however suggestive the indications may be, we have not yet of the fruit ,. , SS . , , , , £ ,. 11 ,/ opens up a discovered an end or the thread or this tangled problem. problem. Yet these figures may serve to direct our efforts in the right direction. Thus the first question they will lead us to put will be the very pertinent one relating to the causes and nature of the variation in the number of seeds in the fruits of the same individual plant. If the relative size of the fruit is determined on fixed principles by the number of seeds, it would be natural to inquire what determines the number of seeds. But several subsidiary questions arise when we peruse the columns of the tables. Why, for instance, does a single- seeded pod differ from other many-seeded pods on the same SEED-NUMBER AND FRUlT-SlZE 341 plant in the exceptional proportion of its pericarp ? In the single-seeded fruits of Albizzia Lebhek the large size of the pod in relation to the solitary seed is very conspicuous. But those interested in the subject will call to mind other legu- minous plants displaying fruits of the same character. Let us take, for example, the pods of Cytisus Laburnum and of Sophora tetraplera. Occasionally they contain only a single seed ; and in such cases it will not be necessary either to measure or to weigh them in order to perceive that, as compared with the typical many -seeded fruits, the size of the pod is quite disproportionate. However, if we rip open one of these single-seeded legumes, and examine the interior carefully, we make a discovery. There is, it is true, only one seed, but we can discern with a lens the remains of all or most of the missing ovules that existed in the ovary before pollen was applied to the stigma. Here, then, is the clue. But this opens up another subject for inquiry which is A problem discussed with some detail in the next chapter, namely, the ^th*^"^ abortion of ovules and the failure of fertilised ovules or questions re- r . lating to the or young seeds. iVIuch depends in the history of the fruit abortion of on whether the original pollination of the stigma resulted thefSlure in the fertilisation of the ovules, or merely served to stimulate °^ ^^^*^^' the growth of the fruit. In the first event there would be produced a normally seeded fruit, but in the second event only a seedless one. It is, however, the partial failure of the ovules or seeds that will afFord the most suggestive materials for study. If some of the ovules only are fertilised, important alterations in the form of the fruit may result ; but the character of the change in shape will be determined by the situation of the aborted ovules, whether at the extremities of the fruit, or in the midst of the other ovules. Changes in shape much less marked will occur, if the fertilised ovules or young seeds fail early in their development. If the seed fails at a later stage in its growth, but little effect is produced on the fruit. 342 STUDIES IN SEEDS AND FRUITS SUMMARY (i) The relation between the number of seeds and the weight of the fruit, more especially for legumes and capsules, is then investigated, increase of size being connoted by increase of weight. In this connec- tion the question is raised as to the stage in which the fruit offers the best materials for such a study, whether as a moist, mature living fruit on the plant, or as a dried fruit of the herbarium. The answer supplied is that dried-up fruits are only of service when referred to the living condition, since a fruit with living pericarp and actively functioning seeds cannot be compared with one where the fruit-case is dried up and dead, and the seeds are in a state of suspended vitality. The comparison can only be made with dried fruits by reconstitutmg the Hving condition. (2) To determine this relation, however, we are often compelled by the whip of necessity to appeal to the dry fruit ; and the author tabulates the results of a large number of observations on dried as well as on moist fruits. (3) Summing up the indications of the influence of the number of seeds on the proportion of parts in capsules, like those of Msculus^ Canna, and Iris, he finds that the fruit, in response to the increase in the number of seeds, whilst becoming largei- and heavier, acquires a relatively lighter pericarp. The legume, as typified by the pods of Albizzia Lebbek and Leuaena glauca, on which extensive observations were made, follows the principle of the capsule in few-seeded pods. However, in many- seeded pods, as the seeds increase in number and the fruit increases in length and weight, the legume preserves a fairly constant relation between the pericarp and the seeds. (4) The indications afforded by other legumes often give no very definite results. This is due in part to the insufficiency of the materials, but partly also to the influence of some disturbing cause which specially affects small pods with few seeds, like those of Abrus and IJlex, and seems to make each plant a law to itself with regard to the size and weight of the legume and the number of the seeds. This influence is apt to upset all small sets of observations. (5) It is in the abortion of ovules and failure of seeds that the cause of this disturbing influence is to be found ; and it is pointed out that the legume would be more likely to be affected than the capsule in this respect, since its growth is mainly in one dimension, while with the capsule the growth is tangential rather than linear. (6) As regards the alterations in the average weight of a seed as the seeds increase in number and the fruit increases in size and weight, the indications for the legume are that the seed's weight is but little SEED-NUMBER AND FRUIT-SIZE 343 changed. For capsules the results obtained vary. Thus, with Iris^ we find that as the fruit increases in size and the seeds in number, the seeds add to their weight and size, whilst with Carina they remain unchanged. But when a capsule matures only one or two large seeds, as with the Horse-chestnut [Msculus\ the seeds of the double-seeded fruits are smaller and hghter than those of the single-seeded capsules. (7) However, the study of the relation between the number and size of seeds and the proportions of the fruit opens up a difficult problem. Questions concerned with the variation in the number of seeds in the" same plant at once present themselves, and these cannot be answered without an inquiry into the abortion of ovules and the failure of young seeds. Much depends in the history of the fruit on whether the original pollination of the stigma resulted in the fertilis- ation of the ovules or merely served to stimulate the growth of the pericarp. These matters are dealt with in the succeeding chapter. CHAPTER XVI THE ABORTION OF OVULES AND THE FAILURE OF SEEDS I FIRST took up this subject quite accidentally in Tobago whilst determining the proportional weight of the pericarp in the dry beaded pods of Erythrina corallodendron. Since the pronounced moniliform habit presented a disturbing influence, I was led on to examine its nature, and thus the inquiry commenced. A few weeks afterwards I made a detailed investigation in Grenada of the failure of ovules in the dry legumes of Albizzia Lebbek, and the inquiry developed. The investigations were continued in England and subsequently in Turks Islands. From the beginning my usual plan of following indications was adopted, forming crude hypotheses as I went along and dropping them as soon as they had lost their usefulness. Many points of course remain undetermined, and the contents of the present chapter can only be offered as a contribution to the study of a difficult but highly interesting subject. Each fruit examined told its own story in its own way and threw new light on some point of the subject. Thus, after I had first learned from the legumes of Vkia that all the ovules begin to respond to the fertilisation of the ovary, the capsules of Primula gave the same testimony, but in a different fashion, and further elucidated the matter. The pods of Albizzia and the capsules of Iris and Allium afforded valuable data relating to the influence of the early failure of ovules and of very young seeds on the form of the fruit, the first named giving me the 344 THE ABORTION OF OVULES 345 hint that in the form of a fruit we have the history of the ovule rather than of the seed. Regarding the beading of legumes many fruits supplied evidence both direct and indirect. Not only was special appeal made to characteristic moniliform pods like those of Sophora and Erythrina and to the less marked, though normal, contractions of the legumes of Albizzia^ but help was also received from legumes like those of Poinciana regia^ where the contraction of the pod is the exception and not the rule, and from pods like those of Vicia and Ulex, where the symmetry of the fruit is not afFected by the failure of ovules and of very young seeds. The indications again of fruits of other types, like those of /m, gave valuable data in this connection. Scarcely a fruit examined failed to assist in the elucidation of the problem. One of the earliest notes in my journal relating to this subject was to the effect that there was evidently a very early stage of ovule-abortion in the case of pods of Erythrina corallodendron which expressed itself in the narrowest constric- tions of the fruit, but left no trace of the ovule itself. This has served to lighten up the background of the inquiry from the commencement to the end. I made a special study of the legumes of Vicia sativa and The failure V. sepium as concerns the abortion of ovules and the failure vi^iTsltiva of young seeds. In the first case the legumes are lonp- and ^"^ . T 1 J 1 1 , , ■ 1 V- sepium. narrow. In the second case they are short and relatively broad. The pods of Vicia sativa average 43 or 44 millimetres in length and about 5 millimetres in breadth, and possess as a rule eleven mature seeds and one aborted ovule or un- developed seed. The pods of Vicia sepium average about 3 1 millimetres in length (range 25 to 2>^ millimetres) and 6 millimetres in breadth, four or five seeds being usually matured. In both species the average number of ovules in the flower is about the same, namely, twelve, the range being ten to sixteen in both plants. With both species all the ovules begin to enlarge and to turn green after fertilisation of the ovary. At first about 0*3 346 STUDIES IN SEEDS AND FRUITS millimetre in size, they all attain a size rather less than a millimetre. It is after this stage that the differences between the two fruits are displayed, and in describing them I will make use of my average results. In the case of Vicia septum four of the original twelve ovules now begin to fail and shrivel, the rest proceeding with their growth until about i'5 milHmetre across, when three more fail, and ultimately only five become mature seeds. On the other hand, with Vicia sativa nearly all the twelve ovules give rise to mature seeds, only one as a rule failing either in the early or later stage above described. I may add that such a study should be made on moist green pods in different stages, since the dry fruit could tell us but little of the history of the ovules. Now this conspicuous difference in the history of the ovules is associated with considerable contrast in the growth of the pods, the pod of Vicia sativa, where nearly all the ovules form mature seeds, becoming long and narrow, and that of Vicia sepium, where many_ of the ovules fail, becoming short and broad, so that the ripe fruits differ greatly in appearance. At first sight one would be inclined to connect the shorter pod of Vicia sepium with the greater failure of the ovules, since in both species the original number of ovules is the same. But on closer investigation we find that the contrast in length cannot be so easily explained. In the first place, the ripe pod of Vicia sepium is but partially filled by its seeds, whilst that of V. sativa is nearly or completely filled. If the pod of the first named had been quite full of seed, the argument would have had some cogency ; but as it happens, large unfilled spaces occur at the two ends of the seed-cavity due to the failure of the ovules chiefly in those situations. We thus get an indication of the independence of the size of the pod as far as concerns the number of seeds. From this it would follow that the same dimensions of the legume are in the main retained in the few-seeded and in the many-seeded pods of Vicia sepium. Further inquiry indeed shows that this is in a general sense the case, such differences THE ABORTION OF OVULES 347 as occur indicating only a slight increase in length of the pod as a result of a great increase in the number of seeds, and affording but scant basis on which to found an explanation of the differences in length between the pods of these two species. Thus the rule which we would apply to explain the differences between the two species largely fails when we apply it to the individuals of one of them ; and we are accordingly debarred from using this argument in explaining the fact that the long pod of Vicia sativa has many seeds and few failures and the short pod of V. sepium many failures and few seeds. Using legumes of the same set of plants of Vicia sepium and selecting a few at random, 1 found that a pod where all the twelve ovules had developed into mature seeds, a very rare event, had the same length as a pod where only five seeds had matured. So again a four-seeded pod and a nine-seeded pod had the same dimensions. In the aggregate, however, there is a tendency in the pod to increase its length as the seeds increase in number ; but it is quite insufficient to explain the difference in length between the pods of these two plants. Thus ten pods of Vicia sepium containing four or five seeds had an average length in the dry state of 30 millimetres, whilst five pods containing from seven to twelve seeds had an average length of 35 millimetres. Here the doubling of the number of seeds only resulted in the increase of the pod's length by one-sixth. There are two other points to notice in connection with the legumes of these two species of Vicia. In the first place, it will have been remarked that the aborted ovules, or more correctly speaking the seed-failures, are complemental to the matured seeds, the two going to make up the complete set of the original ovules. This is well brought out in the results tabulated below ; and on referring to the general table for ovular abortion in fruits given later in this chapter, it will be seen that this principle is characteristic of fruits of all kinds. 348 STUDIES IN SEEDS AND FRUITS In the next place, there is as a rule no " beading " to be observed in these fruits of Vicia, a character which is associated in moniliform pods with failure of the ovules. This is in part due to the fact that the failure of the growing ovule or young seed usually takes place in these pods at the extremities of the seed-cavity and not in the middle. But this only partly explains the absence of contractions, since much also depends on the structure of the pod itself. Table showing how the Aborted Ovules and the Small Imperfect Seeds of Vicia sepium combine with the Matured Seeds to FORM the Original Complement of the Ovules (12). (Twenty-four pods were examined. In a few cases the aborted ovules and imperfect seeds are differentiated, the results being given in the last two columns. ) Mature seeds. Aborted ovules and imperfect half-sized seeds. Total. Aborted ovules. Imperfect half- sized seeds. 2 • 10 12 2 10 12 8 II 6 2 8 II 3 5 7 10 9 13 9 13 9 13 7 II 7 II • 7 II 6 10 7 12 7 12 7 12 6 II 7 13 7 13 6 12 2 4 II 4 12 9 3 12 II II 12 12 Av. 117 All the ovules begin to enlarge and to turn green after the fertilisation of the ovary. Some of them fail before they attain a size of a millimetre and are termed aborted ovules. Others proceed further with their growth, but fail when about half-size (i'5 millimetre). The original size of the ovule is about 0*3 millimetre, their number averaging twelve. THE ABORTION OF OVULES 349 Table showing how the Ovules that fail to produce Seeds in THE CASE OF ViCIA SATIVA COMBINE WITH THE MATURED SeEDS TO FORM THE Original Complement (12) of the Ovules before Fertilisation of the Ovary. (Ten pods were examined. All the remarks on the size, behaviour, and original number of the ovules of Ftcia septum, as given at the foot of the preceding table, here apply. ) Mature seeds. Aborted ovules and imperfect half-sized seeds. Total. II II 10 11 » \l 9 12 II 10 12 ■ AV. I2*I To no fruits have I paid more attention to the relation The failure between the failure of the ovules and the form and size of the Albizzia fruit than to the legumes of Albizzia Lebbek. All my materials Lebbek. were obtained from one tree in Grenada ; but the observations were restricted on account of the season to the dry pods. These legumes well illustrate the commencement of the moniliform habit, but the contraction is rarely so marked as to merit that epithet. The first thing to notice is the relation between the number of seeds and the length of the fruit. It will be observed from the results tabulated below that there is a progressive increase in the pod's length from the single-seeded to the twelve-seeded fruit ; but, as also indicated in the pods of Vicia septum^ the increase in length is relatively small, since the twelve-seeded legume is only about twice the length of the single-seeded fruit. Thus we obtain here another indication of the independent growth of the fruit-case as regards the seeds. 350 STUDIES IN SEEDS AND FRUITS Table illustrating in the Case of the Dry Legumes of Albizzia Lebbek the Relation between the Number of Seeds and the Length of the Pod. Number of seeds in a pod. Number of pods examined. Average length of a pod. Range of length in inches. I 10 142 millimetres 5 '6 inches 4'9- 6*5 2 H 180 7'i „ 6-3- 8-s 3 13 193 7-6 „ 6-3- 87 4 H 218 8'6 ,, 7-2- 9-8 5 13 224 8-8 ,, 7-0-IO-3 6 10 236 , 9"3 >. 7-6-IO-5 7 8 251 9'9 .. 8-5-II-3 S 8 262 10-3 „ 9'o-ii'4 9 10 254 lo-o ,, 8-4-II-I lO II 290 "•4 „ IO"0-I2"5 II 8 282 ii'i ,, IO'I-I2-3 12 10 295 II-6 ,, IO-8-13-2 These data are included in a table in Chapter XV., where the relations between the length and weight of the pod and the number of seeds are compared. The slight difference in three of the results is due to an additional pod being employed in the above table. But it is around the early abortion of the ovules after the fertilisation of the ovary that our interest in the fruits of Albizzia Lebbek chiefly centres. With this feature in the history of these legumes are associated the contractions in the pod's width, which frequently give a fantastic shape to the fruit, as illustrated in the figures outlined in a later page. The legumes readily lend themselves to such an inquiry. The aborted ovules are conspicuous in the dry pod and present themselves as little black bodies 0*2 millimetre in size, attached to the end of the funicles, which are usually well preserved. The term aborted ovule is here applied, as in Ficia^ to ovules that fail soon after the fertilisation of the ovary. Ovules that did not abort in this early stage as a rule advanced sufficiently far in their growth to be designated seeds. It will be subsequently shown that the failure of seeds advanced in growth has little or no influence on the fruit. As the result of a large number of observations the following: infer- ences were formed. THE ABORTION OF OVULES 351 In the first place, the complemental value of the aborted ovules with regard to the seeds is to be noticed. I had no opportunity of examining the flowers of this tree ; but it is evident from what follows that, as in Vicia^ the ovules average about twelve in number. So we find that whilst the one- seeded pods have usually about eleven aborted ovules, the twelve-seeded pods have none. With the intermediate pods we accordingly find the same relation. Thus a four-seeded pod displays about eight ovules and an eight-seeded pod about four ovules. This complemental relation Is well displayed in the following table. Table showing the Complemental Relation in the Case of Twenty-two Legumes of Albizzia Lebbek between the Number OF Seeds and the Ovules that fail. (See previous remarks for further explanation.) Number of Aborted Full Number of Aborted Full seeds. ovules. complement. seeds. ovules. complement. II 12 5 8 13 II 12 5 7 12 II 12 6 7 13 10 12 7 4 II 9 II 8 4 12 8 10 S 4 12 10 13 9 3 12 9 12 10 2 12 9 13 12 12 s 12 12 12 8 12 13 '3 Pods with thirteen seeds are rare, probably not i per cent, in frequency. Those with two to four seeds are most numerous, the single-seeded pods being infrequent. Another important result was elicited by these observa- tions, namely, that the shape of the pod of Albizzia Lebbek is not affected by the failure of a seed when it is well advanced in growth. Thus if a seed fails in the body of the pod when it is only one-third or one-half of the mature size, no narrow- ing of the fruit occurs. The form of the pod thus gives an epitome of the history of the ovules, but not of the seed. (I am assuming here that, as in Vicia, Ulex^ etc., all the 35^ STUDIES IN SEEDS AND FRUITS ovules share at first in the results of the fertilisation of the ovary ; and it is to those that fail early that the epithet of " aborted " is applied.) It is the early failure of the ovule that alone determines the contraction of the legume, and it is the number of such contiguous ovules that deter- mines its amount. The external form of the legume of Albizzia Lebbek there- fore at once gives a clue as to the number of ovules that have aborted early and the number that have advanced to the seed- stage ; but it tells us nothing of whether the seed inside is fully formed or only half or a third of the normal size, the opening of the pod being required for that purpose. The form of the pod, therefore, depends on the fate of the ovules in an early stage of their history after the fertilisation of the ovary. After that is determined the ultimate form of the pod is fixed, and is not affected even if all the seeds fail before attaining half their normal size. With the early failure of the ovules are connected, as before remarked, the contractions of the pod, which may be so slight as merely to give a sinuous margin to the fruit, or so marked as to produce almost a beaded shape. The moniliform tendency is displayed when the failure takes place in the body of the fruit. When the ovules abort at the ends the pod acquires tapering extremities. Now, as respects the narrowing of the body of the fruit, I find that the extent of the contraction depends on the number of contiguous ovules that fail. Thus a typical many-seeded pod, t^^ millimetres broad in the seeded portion, has its width reduced to about 30 millimetres where a single ovule fails, to about 25 millimetres where two con- tiguous ovules abort, and to about 12 millimetres where three or four ovules fail. As a rule, when the space between the seeds is 12 to 15 millimetres broad there is no aborted ovule. When the inter-seminal space amounts to from 20 to 22 millimetres in width there is a single aborted ovule. When it is 30 to 'T^c^ millimetres there are two contiguous aborted ovules ; when 45 to 50 milUmetres, three contiguous ovules, and so on. THE ABORTION OF OVULES 353 The following outline-figures will serve to illustrate the fore- going remarks. Albizzia Lebbek C 9 inches. A 7z inches. B 6 inches. E Whinches. D Q^ inches. A KJ These figures illustrate the relation between the contractions of the legumes and the failure of ovules shortly after the fertilisation of the ovary, the contraction being greatest when contiguous ovules abort, and least when only one fails. All the figures are greatly reduced, but on the same scale. The seeds are indicated by large black spots, and the ovules by small black ones. In these diagrams it is assumed that the pod can be seen through. The seeds and ovules are, of course, arranged alternately on the two valves. Many of the features of the legumes of Albizzia Lebbek are exhibited by the irregularly moniliform pods of Cytisus Cytisus Laburnum^ though here the ovules do not fail early but proceed 23 354 STUDIES IN SEEDS AND FRUITS a little with their growth and abort as young seeds. When all the ten ovules form mature seeds, or advance considerably in growth, the pod is regular in form and displays no con- strictions ; but more often three or four of the seeds fail in a very early stage, and when contiguous produce a marked constriction in the mature fruit. The club-shaped single- seeded pods, where all the ovules fail except one near the distal end, are very remarkable. Reserving my treatment of beaded legumes for a later page, it may be here added that important data relating to this subject are supplied by the legumes of Ulex europaus and Entada polystachya. The examination of the Ulex pods brought out the fact that early abortion of the ovules never even provokes a tendency to moniliform contraction, and that the form of the pod, fixed soon after the fertilisation of the ovary, remains unaffected by the early failure of the ovules. This no doubt results from the structural characters of the pericarp. The legumes of Entada polystachya also gave their indica- tions. Here there is no early failure of the ovules after the fertilisation of the ovary ; but all advance considerably in growth, and when failure occurs it is the young seed about a fourth of the normal size that aborts. Here the fruit consists of a series of joints each containing a single seed and held together in the dry state by the " replum," a stout stem-like border that would of itself prevent any constriction of the pod. The usual absence not only of a seed but of any trace of an ovule in the two terminal joints leads one back to an earlier condition of things, when additional ovules existed that are now doubtless only represented by rudiments in the ovary, and disappear altogether in the fruit. The behaviour of legumes with one or two seeds is well represented by the fruits of Guilandina bonducella. The ovary here possesses two ovules, and it would seem that as a rule half of the pods mature one seed and the rest two seeds. In the first case the second seed usually attains a diameter of 4 or 5 millimetres before aborting, the mature size being about 25 THE ABORTION OF OVULES ^55 millimetres. We can thus perceive, as explained with regard to Albizzia pods, why no effect is produced on the form of the fruit by the failure of one of the seeds. Had the ovule failed in an early stage the effect might have been noticeable. But even then, as in the case of Ulex pods, it is probable that there would have been no constriction in the pod's width, a result due to the structure of the pod itself. Capsular fruits now claim our attention ; and it may be The failure said of them, as of the legumes, that the mature seeds, the capsulS*" young seeds that fail before they attain any size, and the ovules that abort early, make up the full complement of the ovules in the flower. This is well brought out in the average results given in the table preceding the summary of this chapter, and one need scarcely labour the point here. They all behave, for instance, like the fruits of Arenaria peploides. The primal complement of twelve ovules in this case can almost always be recognised in the fruit, whether the fruit only contains five or six seeds or as many as ten, the failures making up the balance. All the ovules of the unfertilised ovary can be thus accounted for. They are potentially seeds from the beginning. The occurrence of rudiments of ovules in the flower raises quite another question, which will be dealt with in a later page. As an example of the influence of the failure of ovules on Illustrated the form of capsules, I will first take that of Iris Pseudacorus. iJis^p*sluda- Here there are two rows of ovules in each of the three com- ^orus. partments of the fruit ; and they all begin to enlarge after the fertilisation of the ovary. If most of the ovules in each of the two rows mature as seeds, the seeds are closely packed and overlap each other and are irregularly wedge-shaped ; but if the ovules in only one of the rows become seeds, whilst those of the other row fail, then the seeds assume a disc-like form. Failure of some of the ovules is, however, a normal event, and much depends on their number and on their situation. Thus the average result given in the table at the end of the 356 STUDIES IN SEEDS AND FRUITS chapter for a fruit possessing 120 ovules is that 80 mature as seeds, 30 fail soon after the preliminary enlargement due to fertilisation, and 10 fail after they have attained a fourth or a third of the size of the normal seed. In the typical oblong fruit 9 or 10 ovules soon abort at the bottom of each Iris Pseudacorus Outline-figures of the capsules of Iris Pseudacorus showing the effect on the shape of the partial failure of the ovules and young seeds in different parts of the fruit, as described in the text. ^ . A = a typical regular fruit. B = a club-shaped or bulbous fruit. C = a fiddle-shaped fruit. D = a curved or arcuate fruit. compartment. The occasional early failure of nearly all the ovules in the lower third of the young fruit gives the mature capsule a bulbous or club-like shape. I found such fruits frequent in one locality. An oudine-figure of one of them is given above. If this became the rule, the bulbous form would become a specific constant. Should the early failure THE ABORTION OF OVULES 357 of the ovules and of the young seeds be principally restricted to the centre of each compartment, the fruit-walls collapse in the middle and we get a fiddle-shaped fruit, as shown in the figure. Then, again, if the failure of the ovules and young seeds takes place mainly in one loculus, whilst the seeds in the other two for the most part develop normally, the mature fruit then assumes a curved or arcuate shape, the concavity corresponding to the compartment in which the failures are situated. Some- times in arcuate fruits two loculi or compartments correspond to the concave side ; and then it will be found that seeds are lackins: in the centre of each of the two loculi concerned. If the third compartment had been similarly aflTected, a fiddle- shaped fruit would have been produced. However, the essential feature of the arcuate type of fruit is a full set of seeds on the convex side, associated with extensive abortion or failure of the ovules and very young seeds on the concave side. Such are some of the commoner types of deformation of these capsules ; but lesser irregularities arising from the same cause are numerous. Irregular forms of the capsule, such as are frequent with Iris Pseudacorus^ are not so common with Iris fcvtidissima. The Iris foetid- fruits of the last named are useful in illustrating the fact that ^"**"^* there is always a variation in the original number of ovules in each compartment of capsules of this type. The variation is not large, but it is sufficient to indicate a possible initial disturbing influence that would afi^ect the symmetry of the plurilocular capsule. There are here appended the results obtained from four flowers of Iris fatidissima for the three loculi. A : 18 + 18 + 20 ovules B : 20 4- 22 + 22 „ C : 17 + 17 + 21 „ D : 16 -h 16 + 18 „ Respecting the distribution of aborted ovules in the several compartments of a plurilocular capsule, I obtained the following 358 STUDIES IN SEEDS AND FRUITS data for Scilla nutans. Here each of the three loculi of the ovary contain from ten to twelve ovules, and each of the three compartments of the fruit on the average seven or eight seeds. As a rule early failure of the ovules and of the young seeds takes place with fair uniformity in the different compartments. Fruits displaying such variations in the number of seeds in each of the loculi as five, eight, seven, were frequent ; whilst those where the divergence was greater, such as three, eight, six, were infrequent. Out of a large number of fruits examined, none ever displayed a complete failure of the seeds in one compartment. I will now take the capsules of Primula veris, in which most of the abortions of ovules and of the failures of young seeds take place around the base of the placental column. A typical ovary contains on the average about ninety white ovules. Of these all at first respond to the stimulus of the pollen, but only about sixty mature as seeds. Of the remaining ovules about twenty-two abort early when still uncoloured, whilst the rest (eight) turn green and advance a little in their growth before they too fail. Thus in Primula capsules we have a colour- indication which enables us to distinguish between the early failures of the ovules and the later failures of the young seeds, the first white, the second green like the growing seeds. In Allium ursinum we have a three-celled ovary, each cell containing two ovules. Of the six ovules only three on the average mature as seeds, and often only two or even only one. In all cases the ovules that abort early and the seeds that soon fail combine with the mature seeds to make up the original complement of six ovules in the flower. The ultimate form of the fruit is, as might be expected, very variable, and irregularities almost appear to be the rule. Regular fruits have a seed in each cell ; but it is not uncommon to find two seeds in one cell, one seed in another, and none in the third, and then the shape is very irregular. However, all the ovules begin to enlarge after the fertilising process and at the same time all the cells enlarge as well ; but when the THE ABORTION OF OVULES 359 two ovules in a cell fail early, the growth of the cell does not proceed far. Here it is evident that the initial growth of the fruit is determined by the fertilising process, but its later growth depends on the seed. In Aquilegia each carpel contains on the average thirty ovules, Aquilegia. of which twenty-five mature as seeds, four abort early, and one fails after proceeding a little further in its growth. Most of the aborted ovules usually occur in a clump at the lower end of the follicle. This may be connected with the fact that in the flower the ovules at the base of each carpel are exposed by the gaping of the carpellary walls, and remain exposed during much of the ripening of the fruit. The tremendous waste of ovules, or rather of young seeds, Ravenala in the sub-drupaceous capsule of Ravenala madagascariensis has ^^^Ts." already been alluded to in Chapter XIII. in connection with the dehiscence of the fruits. Of the plants with many-ovuled flowers included in the table near the close of this chapter there is not one that comes near Ravenala in this respect, five- sixths of the ovules at a very moderate computation failing to mature as seeds. It would seem, however, that nearly all of them first reach an early stage of seed-growth. One cannot help thinking that this large sacrifice of seeds is due to the great pressure exercised by the growing seeds upon each other within the stony, and for a long time unyielding, walls of the endocarp. When dehiscence at length occurs, the fruit-cavity is seen to be full of aborted seeds, 2 to 5 millimetres in size, with only a few large matured seeds. Although the beading of legumes is well marked in some The beading genera, as in Erythrina and Sophora^ a slight tendency to the moniliform type is frequently displayed in other leguminous genera with long pods, such as Faba^ Genista^ Phaseolus, Poincianay etc. Though abnormal in such cases, it is due to the same influences that operate in the typical moniliform pod, namely, the early abortion of ovules and the early failure of seeds in the body of the fruit. In some other cases, as in those of the Laburnum and of Albizzia^ the tendency is pronounced ; 360 STUDIES IN SEEDS AND FRUITS but here the structural characters of the fruit act as final checks. In short broad pods hke those of Ulex there is no constriction as a result of the failure of the ovules. The same remark applies to those of f^icia ; but here the absence of any contrac- tion of the legume is due chiefly to the failures occurring usually at the ends of the fruit. When several ovules and young seeds fail at the ends of a legume, we have tapering extremities. Such failures do not necessarily lead to shorten- ing of a legume. An unusually short fruit is associated with an unusually small number of ovules in the flower. Length in a legume is one of the pre-disposing causes ; and in such a case even a dense ligneous texture, as in Poinciana regia^ may not be sufficient to suppress the tendency. Here the failure of seeds that have advanced in their growth seems to have but little influence ; but if a few contiguous ovules abort soon after fertilisation, or if a few contiguous seeds fail in a very early stage, a slight constriction of the pod is produced, the place of the seeds being occupied by ligneous material. Thus in a legume of Poinciana regia, where the space of four seeds was thus filled up, the width of the legume was reduced from 48 to 40 millimetres. Coming to the typical moniliform legumes, I will begin with those of Erythrina corallodendron^ which first led me to undertake this inquiry. Here there is a distinct relation between the beading of the pod and the failure of very young seeds ; but this is not the whole of the matter, since the narrowest con- strictions show no trace either of ovule or of seed. Here it may be that we have an indication of the influence of rudi- ments of ovules in the flower which disappear altogether in the fruit. In the accompanying diagram I have represented a typical seven-seeded dry pod about 115 millimetres long. At A, E, and F the constrictions are very narrow, 2 millimetres or less, and there is no trace of either seed or ovule. The other constrictions (B, C, D) are broader (2*5 millimetres to 3 milli- metres), and here aborted seeds i-i '5 milHmetres long occur, the length of the constricted portion being regulated by the number THE ABORTION OF OVULES 361 of contiguous seeds that have failed. These seeds when they aborted in the living fruit could not have been more than a sixth of the normal length, and when the failure took place the pod was probably about a third of its length when mature. The above results of my examination of the pods of Erythrina cor alio dendron would lead one to infer that the narrowest con- strictions, where not a trace of ovule or seed is to be recognised, are concerned with a much earlier stage of abortion, and this is also indicated by the terminal pointed beak. One may suppose that there were rudiments of ovules in the flower that were unable to respond to the fertilisation of the ovary, thus restrain- ing the response of the ovary in those places to the stimulus of fertilisation. In this connection we have to bear in mind that the fruit as well as the seed is a product of pollination. The flower in the case of the seven-seeded pod above described probably displayed thirteen ovules in its ovary, which I imagine would be typical for the species. Doubtless there were the rudiments of six or seven other ovules in the ovary, taking the beak of the fruit into consideration ; and the primal complement of ovules was therefore probably twenty, though the other seven ovules need never have passed beyond the primordial stage. In other words, to adopt Dr Goebel's standpoint with regard to rudimentary organs in plants, the assumption of a primal complement of ovules larger than that of existing flowers does not require us to assume that all of them " functioned " as mature ovules. Such were some of the speculations that passed through my mind when I examined the first moniliform pods ; and they have been exceedingly helpful to me ever since. The moniliform pods of Sophora tomentosa were examined Sophora in the moist condition. A typical legume contains six or seven °'"^" °^*' seeds separated from each other by narrow necks which often enclose shrivelled aborted seeds i to 3 millimetres long, the length of the neck depending on the number of aborted seeds, as indicated in the accompanying figure. We seem to be only concerned here with the failure of young seeds ; but still earlier 362 STUDIES IN SEEDS AND FRUITS failures are indicated when seeds, though actually in contact, are separated in a sense by a slight contraction of the pod, suggest- ing, as I would suppose, the existence of a rudimentary ovule in the flower which has disappeared in the fruit. The constriction of legumes is so generally associated with the early failure of Erythrina corallodendron SOPHORA TOMENTOSA Diagram of a seven-seeded dry pod, showing also six aborted seeds in the broader necks : drawn to a true scale. Diagram of a portion of a green pod, showing the effect of failure of the seeds on the form of the fruit. The pod has not begun to dry, and the seeds are soft and large : drawn to a true scale. ovules and young seeds that one would be scarcely justified in regarding a slight contraction as belonging to another order of things. Single-seeded fruits are spindle-shaped and contain the remains of at least six or eight seeds that failed in an early stage of their growth. When examining these pods I had no opportunity of inspecting the flowers ; but I should imagine that the average ovular complement would be twenty, though THE ABORTION OF OVULES 363 liable to considerable variation. The number of seeds in a pod varies from one to ten. The development of the seedless fruit has a most important The indica- bearing on the relation between the form of many fruits and seedkss ^ the failure of the ovule. The germination of the pollen-tube, ^™'*- writes Dr Jost in his Lectures on Plant Physiology (English edition, 1907, p. 370), has an exciting influence on the develop- ment of the fruit, so that fruits may be formed where de- generate ovules fail to become seeds. Again, Dr Pfeffer writes that the penetration of pollen-tubes may act as a stimulus to growth of fruits without any fertihsing influence being exercised {Physiology of Plants, ii. 173). We have here a means of explaining some of the curious forms of fruits. But much will depend on whether the ovules habitually fail in the same part of the young fruit, or whether their failure is occasional and not restricted to one situation. In the first case we have a persistent effect produced on the fruit's shape which requires a specific or a generic value, as in Anemone. In the second we have inconstant variation of the fruit's form, such as I have described in the instance of Iris Pseudacorus. It would, however, be rash with the scanty data at my disposal to push this view very far. Yet the ovules that fail in a Primula or an Iris capsule appear to be in quite a different category from the ovules that fail in the fruits of the Oak or of the Coco-palm. In the first case it would seem that the ovules were actually fertilised and afterwards aborted. In the second case it would appear that the ovules were incapable of being fertilised, since they persistently fail. Lord Avebury would regard such persistently functionless ovules as carrying The us back to the time when, in the ancestors of the plant, all the ovule, ovules developed into seeds {Seedlings, Internat. Sci. Ser., pp. 241 and 243). Professor Bower holds a similar view with reference to the abortive ovules in the beak of a fruit of Anemone nemorosa, regarding them as " the imperfect repre- sentatives of a plurality of ovules in the ancestry" {The Origin of a Land-Flora, 1908, p. 127). It should, however, be pointed 364 STUDIES IN SEEDS AND FRUITS out that this would .not follow if we accept the standpoint taken by Dr Goebel in his Organography of Plants (i. 61), that functionless organs in plants are not necessarily the vestiges of former completely developed ones, and that many more primordia are laid down than become functional. Many points remain to be determined before we can safely generalise in these matters. It is of importance, for instance, to ascertain by microscopical examination why, with the same complement of about twelve ovules in the flower, ten or eleven seeds are matured in Vicia sativa and only half that number in Vicia sepium. Then, again, it would be necessary to learn if the ovules that fail in the acorn, coco-nut, and similar fruits, have the same microscopical characters as the ovules that com- plete their development. Someindica- Nearly all the data included in the following table are from following iTiy own observations, with the chief exception of those relating ^^^^' to Convallaria^ which are taken from Lord Avebury's book on seedlings. Although with many-ovuled flowers there is great variation as to the number of ovules that mature as seeds, as many as 80 or 90 per cent, failing in Ravenala and as few as 17 per cent, in Aquilegia and Lychnis^ yet a rough average shapes itself for several of the capsules here dealt with. Thus with /m, Primula^ Scilla^ Stellaria^ and Arenaria about two-thirds of the ovules mature as seeds, and of the remainder the greater number (about 25 per cent, of the ovular complement) abort soon after fertilisation, whilst the residue advance a little in their growth and fail as young seeds. With leguminous plants the same rule prevails, though the data are insufficient for a numerical statement. Here also a large proportion of the ovules develop into seeds, but a considerable number fail, and of the failures most are concerned with the abortion of the ovule soon after fertilisation. In nearly every case the number of the ovules in the flower has been directly determined ; but in the cases of Ravenala and Opuntia it has been estimated from the total of mature seeds and of seed-failures. THE ABORTION OF OVULES 365 Table showing the Average Proportion t)F Failures of Seeds. (See below for explanation.) Failures B. Seeds of ovules and young seeds. A. Failures Failures Range of number of ovules Average number of ovules matured. of ovules. of young seeds. in a aJ cilla nutans as disclosed by my experiments on the LSbimenko fruits of the living: plant. When I experimented on these onthecon- . ° ^ - ^ ditions in the capsules m the summer or 1908, 1 was not aware that similar interior of experiments had been made by Lubimenko on the pods of p^c^s'"*"^"^ certain leguminous plants, Pisum sativum^ Colutea arborescens^ and Lathyrus latifolius, the results of which have since been published 374 STUDIES IN SEEDS AND FRUITS in a paper entitled " Etude pliysiologique sur le developpement des fruits et des grains" (^Comptes rendus, Ayigust 24, 1908). They are not concerned with the coloration of seeds, but they are interesting from the light they throw on the conditions in which it takes place. Selecting very young pods on the living plant with their sides still in apposition, he removed part of them by longitudinal sections, and found that they soon formed a new suture, closed themselves in, and developed normally. When, however, pods more advanced in growth and with their inner surfaces no longer in contact were experimented on, different results were obtained. By cutting out portions of the sides of the pods so as to bring the young seeds into direct communication with the outer air, it was ascertained that the growth of the seeds was arrested, the fruits falling off in about a week. The conclusions arrived at were that for the normal development of the seed a confined atmosphere (une atmosphere confinee) is needed, and that one of the functions of the pericarp is to maintain the internal air at a certain stable composition. The green parts of the pericarp, it is observed, decompose in light the carbonic acid arising from the respiration of the seeds and prevent its accumulation inside the fruit. Even in darkness the gas is kept within certain limits by its slow diffusion from the fruit. As regards the failure of young pods when the seeds are exposed to the air through windows cut through the pericarp, it may be surmised that capsules with their cells or locuH distinct from each other would be rather more suitable for such experiments. When an opening is made in the side of an ordinary legume the whole contents are exposed to the outside air, whilst in a capsule, by confining the experiment to only one of the cells, the rest of the fruit would be relatively unaffected. The author's My own experiments on the green capsules of living on the plants of Scilla nutans occupied the period between the middle S^?l1a ^^ °^ °^ J^"^ ^^^ ^^^ beginning of August, and were limited to one nutans. cell only in each fruit, the other two being left untouched. The cell was so incised that the seeds in it were freely exposed SEED-COLORATION 375 to the air. Though checked by plants in pots kept indoors, the results as described below were all obtained from plants growing in a hedgerow and exposed to the ordinary weather- conditions. In the early stage, when the young seeds are little more than bags of fluid, they are pearly white. As the seeds mature and their contents solidify, they become succes- sively dull white, yellowish, reddish brown, and finally shinmg black. The colouring stages and the maturation are completed in the green moist capsule before dehiscence. When the fruits were incised in the middle of June the seeds exposed were soft pearl-like bodies, which, if detached and allowed to dry, shrank to a mere skin. By the beginning of August the exposed seeds had reached the brown stage, the last but one of the stages of the colouring process ; but with the exception of the difl^erence in hue and their rather smaller size they were normal hard matured seeds. In the other two closed cells of each capsule experimented on, the seeds during this period acquired the typical black colour and were normal in their other characters. The experiment did not seem to aflfect the rate of the changes in the seeds of the other two uncut cells ; and when the capsules of other plants around were dehiscmg, these two cells were doing the same, displaying normal black seeds that contrasted in their hue with the brown seeds exposed in the incised cell. The upshot of these experiments is that the exposed seeds developed normally, with the exception of their failure to acquire the final black colour, for which en- closure in the fruit seems requisite. The foregoing experiment on the seeds of Sci/Ia nutans will serve to illustrat'e the complexity of the processes involved in seed-coloration. 1 will now proceed to discuss more at length the conditions under which seeds colour in leguminous pods, and in the first place I will take the blackening and black mottling of seeds. One of the chief points which we will endeavour to determine will be the connection between the coloration of the seed and the drying of the pod. Though the two processes are so frequently associated, it is quite 376 STUDIES IN SEEDS AND FRUITS The con- ditions in which the mottling of Vicia seeds takes place possible that the association is in a sense accidental, and that, as in the berry and capsule, seed-coloration may occur in the green moist legume. The seeds of Vicia sepium and Vicia sativa, which are usually mottled with black spots on a dark green or greyish- green ground, will first serve to illustrate this point. After examining a considerable number of their seeds, one might think that these plants produced in each case three kinds of seeds as far as coloration is concerned. There are first the greyish or greenish-grey seeds, then the seeds with the same green or grey ground-colours, but mottled with black spots, and then the seeds that are almost uniformly black, the mottled seeds being the most frequent and the most typical. A closer inspection soon makes it clear that the seemingly uniform black colour really arises from an intensification of the mottling, and that, in fact, all three kinds have the same ground- colour, the differences being due to the variation in the extent of the mottling. In one seed it is almost absent ; in the typical seed it is well developed ; and in another it is so dense that the mottled patches largely coalesce. In both these species of Vicia the soft seed of the green pod is green, with an embryo of a darker green. When the blackening and drying of the pod is well advanced these seeds gradually shrink, harden, and become duller or paler in hue, and then the black mottling appears, the shrinking of the cord ushering in the earlier changes. Long before the dehis- cence of the pod the typical characters of the resting seed have been formed, and the embryo exchanges the dark green hue of the unripe soft seed for the dull yellowish colour that is so characteristic of the resting state of seeds. With regard to the conditions under which the dark coloration of the seeds of these two species of Vicia occurs, the following remarks may be made. On the plant the drying and blackening of the pod precede the mottling of the seeds, which belongs to the latter stage of the shrinking and harden- ing process. But it appears to be essential for the develop- SEED-COLORATION 377 ment of the black mottling of the seeds that the pods should be left undisturbed on the plant. No mottling took place in any of my experiments on detached green pods with soft green seeds, whether the pod was allowed to dry in air or was placed in wet conditions, as in water or in wet moss. Experimenting on the soft green pre-resting seeds, 1 found that when totally submerged in water they failed to mottle and that when allowed to dry in air they did so very imperfectly. But when a soft green seed was placed on the surface of water so that a portion was exposed, in the course of a few days the exposed portion displayed mottling, whilst the under submerged part remained green. These results seem to signify that mottling occurs under conditions intermediate between those to which the air-drying seed and the submerged seed are subjected, such as would be presented in the confined conditions of the closed green pod. These indications of the Vicia seeds will become more significant when we come to consider those of other leguminous seeds. It is more difficult than usual in their case to dissociate the coloration of the seeds from the drying of the pod ; but the data go to show that the seeds of Vicia are capable of hardening and of acquiring their dark coloration in the moist green pod. In this case the drying of the pod might be regarded as interfering with the completion of the blackening process ; and in this way the dark mottling would present itself as the result of a check in the progressive blacken- ing of the seed. We would thus take it that Vicia seeds are mottled because they have failed to become uniformly black. More determinate results are offered in the blackening of the large, soft, white pre-resting seeds of two other leguminous plants, Mucuna urens and Dioclea reflexa ; and I will let each tell its own story. With Mucuna urens, where the immature seeds in the green pod are white, the soft unripe seeds harden and blacken as the pod dries and browns, completing all their changes in the closed pod, though the dark hue usually becomes paler The blacken- ing of the seeds of Mucuna urens. 378 STUDIES IN SEEDS AND FRUITS and of a blackish grey when the seeds assume the final resting state. The changes in the seed are preceded by the shrinking of the fleshy cord or funicle and by the blackening of the raphe and scar, the rest of the seed's surface remaining for a time a pure white. These seeds do not mottle ; but it is significant that though there is every appearance of a connection between the drying of the pod and the blackening of the seeds, the soft white seeds when detached blacken much more rapidly when wetted than when allowed to dry. Here, it seems, drying retards the process. It may be added that the subsequent blackening of the shrinking seed is apparently not connected with the previous blackening of the raphe and scar when the cord withers. The same thing happens with some white seeds (like those of Canavalia ensiformis) when the cord dries up, the black scar remaining a permanent feature of the white matured seed. From these indications of the seeds of Mucuna urens we turn to those supplied by the seeds of the kindred plant, Dioclea reflexa. These seeds are also white when immature in the green pod, and when the pod darkens, as it begins to dry up, the seeds commence to shrink, harden, and blacken. But there is this difference. The blackening process is usually incomplete, and mottled seeds result. They also follow the general rule that the seed acquires its fixed characters as a resting seed, whether in colour or in other respects, in the closed pod. As in Vicia sadva and Vicia septum^ three kinds of seeds as regards their colour can be distinguished, the black or brownish black, the reddish brown, and the mottled seed showing black patches on a reddish-brown ground. As in Vicia also, the mottled seeds are most typical of the plant, and the mottling may be regarded as the failure of the blackening process. But in Vicia the black seed represents the end of the colouring process, whilst in Dioclea it represents the beginning. When the blackening of a Dioclea seed fails altogether the seed is reddish brown ; but when the process is only partly checked there are formed black patches on a reddish-brown SEED-COLORATION 379 ground. This blackening process accompanies the early- shrinking and hardening of the soft, white pre-resting seed, and is most active after the seed has lost 20 per cent, of its weight ; but although often associated with the drying of the pod, there is good reason for holding that these changes in the colour, size, and consistence of the seed may occur, as in the berry, under very moist conditions. Whilst observing the habits of the plants of Dioclea reflexa The con- in their home in the forests of the Grand Etang in Grenada, I seed-color was able to notice the conditions under which the seed-colora- ascTrtSned tion took place. Whilst the reddish-brown and the mottled Jj ob^J2;_ seeds represented those that had undergone the greatest periment. amount of drying in the pod on the plant, the black seeds were invariably those obtained from pods lying on the ground in the most humid parts of the forest, where the seeds failed to dry properly, and, as remarked in another page, sometimes dispensed with the resting stage and germinated in the pod. Such black seeds, when not germinating, possessed coats insufficiently hardened and still somewhat flexible. Another indication that the blackening of the seeds of Bioclea reflexa is most complete under moist conditions was afforded by the circumstance that when the soft, white unripe seeds were detached and allowed to dry in a room they always began to colour and to mottle on the under surface, and were always much darker below than above. After three or four days the under surface was usually uniformly black, whilst the upper surface was only mottled. Such seeds when examined proved to be sensibly softer and considerably moister on the lower than on the upper side. Hence it was evident that a moist surface favoured the blackening process, whilst a rapidly drying surface retarded it. Very instructive is the behaviour of the mottled seeds of J^hemottling Phaseolus multiflorus (Scarlet-runner). That the source of the ofPhaseolus colouring is in the coats alone and is not connected with the J^"arlet-"^ coloration of the embryo is apparently indicated in the fact runner), that the embryo is pale green in the unripe or pre-restmg 38o STUDIES IN SEEDS AND FRUITS seeds of both varieties of the plant, the variety with mottled black and pink or red resting seeds and that with pure white resting seeds. The seeds of the last named are also white when very small and undeveloped ; but they need not occupy further attention here. The stages in the coloration of the variety with mottled seeds are as follows. In their immature state, when very small and soft, the seeds are pale green, with a pinkish or reddish tinge, the embryo being darker green. As the seeds increase in size they become uniformly pink or reddish, the embryo becoming paler in hue. When the seed approaches maturity as a pre-resting seed the pink colour of the coats deepens, and dark mottling begins when the soft full-grown seed commences to harden its coverings. (It may be noticed in passing that the soft, pink pre-resting seed parts with most of its colour when placed in water. This helps to explain the partial blanching which the half-sized pink seeds sometimes experience in the green pod, a change which has a bearing on the deprivation of colour in the white seeds of the other variety.) During these observations and experiments on the seeds of Phaseolus multiflorus some indications pointed to the probability that the black mottling and hardening of the coverings were independent of the drying of the pod, though usually associated with it. Other signs seemed to show that the mottling only became at all evident after the pod had begun to dry. How- ever, the point whether the preliminary drying of the pod was really necessary, or was merely an accidental association, was unexpectedly decided by an experiment begun with quite another object. Some fresh green pods containing only the large, soft, pink seeds, showing scarcely any black mottling, were kept in wet moss in a closed tin for nine days with the purpose of inducing the seeds to germinate without entering the shrinking stage. However, although no seeds were ger- minating, all of them, when the tin was opened, had hardened normally mottled coats. The pods themselves were com- SEED-COLORATION 3 8 1 mencing to rot, but the general conditions under which the seeds underwent these changes must have been those of extreme moistness. Without intending it, I had induced these seeds to colour, harden, and probably shrink a little as well, under conditions as moist as that of a berry. The bearing of the results of this experiment has been dealt with elsewhere. Here we will merely accept their indi- cation that the black mottling and induration of the coats of the seeds of Phaseolus multiflorus have but little to do with the drying of the pod with which they are so generally associated. The results for the seeds of the four leguminous genera TheW^!-^^^^ of plants above discussed are below tabulated. Though it is leguminous not easy at first sight to pick up the thread of the tangled data, gter^^^ed one inference seems to shape itself out of all the observations b^^^iedrying and experiments, namely, that the black mottling and dark coloration of the seeds is not determined by the drying of the pod. From its close association with these processes in the seed-coverings, the drying of the pod has certainly all the appearance of having much to do with them. But we have seen that in berries, as well as in many capsules, seed-coloration can have little to do with the drying of the fruit. This inference, though here drawn with regard to the black but may^^^ coloration of leguminous seeds, applies also to the brown conditions as coloration of seeds of this and other families, as will be brought "^^l^ ^t^^ out below, and probably to seed-coloration in general. In the berry, berry and closed capsule, the soft pre-resting seed colours, hardens, and shrinks under conditions exceptionally moist. But the same capacity of colouring and hardening their coats under moist conditions is exhibited by the seeds of legumes, though in nature disguised by its association with the drymg of the pod. We cannot also doubt that the early shrinking which accompanies the coloration and the hardening of leguminous seeds is at the same time an independent process, and is similarly not connected with the drying up of the fruit-case. 382 STUDIES IN SEEDS AND FRUITS Table illustrating the Stages in the Blackening and Black Mottling of Leguminous Seeds. Colour of the Coats. Colour of embryo. Pre-resting seed Resting seed of of green pod. dry pod. Pre-rest- Resting ing seed seed Whilst Whilst Prema- turely shrunken. of green of dry Full- shrmkmg shrinking Normally pod. pod. grown. detached from pod. normally in pod. shrunken. Mucuna urens White Blacken- Blacken- Blackish Dark grey White Dioclea reflexa White ing Blacken- ing ing Mottling Blackish Black mottling on reddish brown ground Black White ..• • ■ ( Green Dull green Mottling Dull green Green Yellow- Vicia sativa 1 or grey ; or grey ; mottling ish Vicia sepium j no mot- tling no mot- tling on a grey ground Phase olus multi- florus— {a) mottled Pink or Not mot- Deeper Not Black Pale White black seeds reddish tling pink; begins to mottle mottled mottling on pink or red ground yellow- ish green {6) white seeds White White White White Pale yellow- ish green White The brown colour of seeds On account of its predominance as the colour of seeds, brown deserves especial attention. As in the case of black, the indications by no means lend themselves to an easy interpre- tation. The brown coloration of resting seeds, which may vary in shade from a pale hue, as in Mahogany seeds, to a blackish brown, as in old seeds of Entada scandens^ seems to be usually developed in seeds that were white in the soft, moist, unripe condition. That the browning process, whether in the leguminous pod or in the capsule or in the berry, is usually com- pleted before the seeds are exposed to the air has been already established in this and in previous chapters. In all cases it is chestnut. SEED-COLORATION 383 associated with a shrinking and hardening of the soft, unripe, and pre-resting seed. But the brown coloration, with its associated processes, is to be noticed alike under moist conditions, as in baccate fruits, or under dry conditions, as in drying capsules and leguminous pods, a subject dealt with in preceding chapters. Although seeds acquire their brown colour in the closed not con- capsule and pod, their coloration, early shrinking, and harden- Si?drying of ing are so conspicuously associated with the early drying of the fruit, the fruit that it is not a matter for surprise that there should seem to be some causal connection between them. However, experiment showed that these changes in the seed may take place where no drying of the fruit has occurred. Perhaps the most conclusive piece of evidence in this direction is afforded by the aborted seeds of the Horse-chestnut {Msculus Hippocas- Horse- tanum). In full-grown moist capsules showing no signs either of dehiscence or of drying, and containing full-sized white moist seeds, it is not uncommon to find two or three aborted seeds only 2 to 4 millimetres across, but typically brown and hard-coated. The behaviour of the white moist seeds in a full-grown green capsule that has been placed in wet moss in a closed tin is also very significant. Though the capsule at the end of the experiment shows no signs of drying, since the air in the tin would be quite saturated with moisture, the seeds in a few days become normally brown, and experience a slight shrinking in size and a hardening of the coats. In spite, therefore, of their moist conditions, the Horse-chestnut seeds in this experiment experience the same changes that they exhibit in the drying capsule, thus behaving like the seeds of a berry. We can therefore no longer suppose that the drying of The regime the capsule is needed for the preliminary shrinking, hardening, fsreprodu^d and coloration of the seeds within. In the case of brown or p the colour- ing seeds of black seed-coloration the inference is the same. In the pod, the pod and as shown in the case of Phaseolus multiflorus, and in the capsule, as illustrated by Msculus Hippocastanum^ the regime involved in the coloration, early shrinking, and preliminary hardening the capsule. 384 STUDIES IN SEEDS AND FRUITS The brown- ing of unripe seeds in capsules and pods is associated with shrink- ing and hardening of the coats. of the seed and its coats is that which is displayed by the colouring, shrinking, and hardening seed in the moist berry. Whatever the nature of the connection may be, there is no doubt that the browning process of soft unripe seeds is associated in capsules and in leguminous pods with shrinking and hardening of the seed-coats. It has already been shown in the case of the soft, moist white seeds of the Horse-chestnut {Msculus Hippocastanum) that they suffer a loss of 1 7 per cent, of their weight during the browning that accompanies the preliminary drying in the closed capsule (see Chapter XI). In the instance of Entada scandens^ where the soft white seeds on removal from the green pod were allowed to dry on a table, the seeds did not attain their normal dark reddish-brown hue until they had lost about half of their weight. Passing through a preliminary yellowish stage, they lost in five or six days about 20 per cent, of their weight and became a light mahogany brown. After between two and three weeks, when they had lost about ^^ per cent., they assumed the typical colour of the resting seed, the drying process being prolonged until the loss amounted to 60 per cent. The white soft seeds of Entada polystachya turn brown in a similar way ; and here the two varieties, the large pale brown seed containing 10 per cent, of water, and the small dark brown seed with only 6 per cent., clearly show that the colour deepens as the seed becomes drier (see Chapter V). The white flabby seeds of the closed woody capsule of the Mahogany tree {Swietenia Mahogani) harden and turn light brown in four or five days after their removal. As a final example, the seeds of Iris Pseudacorus may be taken, which, when first exposed by the opening capsule in the early stage of browning, have already lost about 20 per cent, of their original weight as soft white seeds. It should, however, be observed that with fruits ripen- ing late in the season, as, for instance, in October instead of the latter part of August and in September, the seeds may be drier and farther advanced in the browning process before dehiscence occurs. Such fruits, however, are backward in maturing, and may even fail to dehisce altogether. SEED-COLORATION 385 A curious distinction between capsules and legumes unfolded itself as I collated my notes on fruits and seeds. As shown in the subjoined comparison of the results given in the following tables, the so-called immature or pre-resting seeds of capsules are usually white, whilst those of leguminous pods are usually green. The proportion of green seeds would have been still greater for the pods, if herbaceous leguminous plants had been as well represented as a strictly natural comparison would have required. The green immature seed is typical of the dehiscent legume, as is shown in the general table. Nearly all the pods with green seeds are thus characterised. On the other hand, the pods with white unripe seeds are generally indehiscent, as in Mucuna^ or break up into closed joints, as in Entada. Summary of the Tables given below, showing the Colours of Unripe or Pre-resting Seeds in Different Kinds of Fruits. Indication that unripe seeds are usually white in capsules and green in leguminous pods. Number of genera experimented on. Total. With green seeds. With white seeds. With seeds of other colours. Legumes Capsules Berries . 35 31 7 24 5 3 8 26 4 3 With the exception of Canavalia the genera behave uniformly. As frequently happens in such cases, a number of points have arisen during the development of this distinction, which it is now too late for me to elucidate. Though in the habit of noting the colour of immature seeds, I did not discover its significance as a distinction between types of fruits until I came to elaborate my notes on seed-coloration. 1 will not enlarge on this distinction here ; but some of the characters associated with it will be noticed when I deal a page or two later with white and green immature seeds and with the colour of the embryo. Below are appended the tables to which the summary of results above given refers. 25 386 STUDIES IN SEEDS AND FRUITS The Colours of Immature and Mature Seeds in Legumes, Capsules, and Berries. I. Legumes. Colours of seeds. Special characters Immature Mature of legume. (pre-resting). (resting). Abrus precatorius . Pink Scarlet Acacia (near arabica) Pale green Dark brown ,, Famesiana White, then pale ,, Indehiscent, green pulpy Adenanthera pavonina . Green, then yellow, then pink Scarlet Bauhinia (species of) Green Black Caesalpinia Sappan ,, Light brown ... ,, sepiaria (A) Green Mottled black (B) White and brown Cajanus indicus . Green Pale brown with slight black mottling Canavalia ensiformis White White ... ,, gladiata Pink Dull red ,, obtusifolia White Mottled brown Cassia bicapsularis Green Dark brown ,, fistula ,, Light brown Cytisus Laburnum Deep brown Dioclea refiexa White Mottled black and brown Indehiscent Entada polystachya „ scandens . " Brown Reddish or blackish brown 1 Breaks up into 1 closed joints Faba vulgaris (Broad Bean) . Yellowish white Brownish green Genista (species of) Green Chocolate brown 1 Guilandina bonducella . 1 Yellowish green, then olive Lead grey 1 Lathyrus pratensis Green Mottled black on light ground Leucsena glauca . Dark brown ... Lotus corniculatus '' xMottled black and brown ... Mucuna urens White Blackish Usually inde- hiscent Phaseolus multiflorus (A) Pink Mottled black on pink or red ground (B) . White White Pisum sativum (wrinkled) Green Green ' ,, (unwrinkled) . ,, Yellowish white ... Poinciana regia . Pale green Mottled black on Tardily dehis- light grey cent ground Sophora tomentosa Yellowish green Pale brown 1 Spartium (species of) Green Dark brown ... SEED-COLORATION I. Legumes — continued. 387 Colours of seeds. Special characters of legume. Immature (pre-resting). Mature (resting). Ulex europoeus Vicia Cracca ,, sativa .... ,, sepium. Vigna luteola Green, then yellow Green Chocolate brown Dark mottled Dark brown II. Capsular Fruits. Colours of seeds. Special characters Immature Mature of fruit. (pre-resting). (resting). yEsculus Hippocastanum White Dark brown (Horse-chestnut) Allium ursinum Blackish Aquilegia (species of) . Green Black Follicular Arenaria peploides White Dark brown Argemone mexicana ,, Black Bignonia (species of) ,, Light brown Siliquiform Canna indica ^^ Black Cardiospermum grandiflorum Green ,, Convolvulus Batatas White J, Datura Stramonium Digitalis purpurea Pale green Reddish brown Dodonsea viscosa . ,^ Black Gossypium barbadense . White ,, Hypericum AndrosKmum ,, Ipomcea pes-caprse ,, tuba ,, tuberosa . .. Brown ... " Black Iris foetidissima „ Scarlet ,, Pseudacorus ,, Light brown Lychnis diurna ,, Blackish ,, vespertina ,, Brown Papaver Rhceas )> Blackish ... Portulaca oleracea Black Primula veris Green Brown Ricinus communis White Mottled black on grey ground. Tri-coccous Scilla nutans ^ Black ... Sesuvium portulacastrum ,, ... Stellaria Holostea Red Swietenia Mahogani ,, Light brown Ligneous Thespesia populnea ,, Brown Baccate Veronica . . . . " " 388 STUDIES IN SEEDS AND FRUITS III. Baccate Fruits. The green hue of the unripe seed has but little influence on the colour of the resting seed with leg^uminous plants. Colours of seeds. Special characters of fruit. Immature (pre-resting). Mature (resting). Arum maculatum . Berberis (species of) Chrysophyllum Cainito (Star Apple) Momordica Charantia . Passiflora pectinata Pyrus Malus (Apple) Tamus communis . White Green White Green White Greenish yellow Reddish Brown Yellowish brown Purplish Brown Baccate capsule As seen in the previous table, green pre-resting seeds are frequent with Leguminosae. A few of the plants there named are albuminous, having large foliaceous embryos of almost the length and breadth of the seed, and enclosed between two slabs of albumen, namely, Bauhinia^ Cassia fistula, and Poinciana regia. Speaking of leguminous seeds in general, there seems usually to be little or no connection between the green coloration of the unripe or pre-resting seed and the ultimate hue of the resting seed. Thus with the seeds of Adenanthera pavonina, the successive stages of coloration in the pod are green, yellow, pink, and, finally, scarlet. It seems rarely to happen that the green hue of the unripe seed is retained in the resting seed. This, however, occurs with the wrinkled varieties of Peas, Pisum sativum. In smooth Peas the green is replaced by yellowish white. Then, again, the green colour of the immature seed may be exchanged in the resting seed for black, as in Bauhinia ; for different shades of grey, as in Guilandina honducella ; for light brown, as in Cassia fistula ; for dark brown, as in Leuc<£na glauca ; for chocolate brown, as in Ulex europaus ; and for a black mottling on a light ground, as in Vicia, Poinciana regia, C^salpinia sepiaria, and Cajanus indicus. SEED-COLORATION 389 Nor can we look for an explanation of this variation in The colour colour of the resting seed to the colour of the embryo in the bryoln the pre-resting or unripe seed. Thus exalbuminous seeds like ^n^'reYS those of Pisum sativum (unwrinkled variety), C^salpinia Sappan. seeds has no -rr- • • ^ • 7-7 1 • 1 1- 1 influence on Ulex europ^us^ Vicia sattva^ Lytisus Laburnum^ etc., which display the final great differences in coloration in the resting seed, all have ie°^minous^ green embryos as well as green coverings in the soft unripe s^®*^- condition ; and the same remark applies to the albuminous seeds of Bauhinia^ Cassia fistula^ and Poinciana regia. If we expect to find a clue in the colour of the leguminous embryo in the resting state, we shall be also disappointed. There is a uniformity of colouring in the embryo of the resting exalbuminous seed which stands in no sort of relation with the varied colouring of the coats of such seeds. As the green embryo enters the resting state, its assumption of a pale yellow or brownish-yellow hue indicates the destruction of the chlorophyll. That green embryos thus change their colour when the immature soft seed contracts and hardens in the shrinking stage seems to be a general rule. This is brought out by the table below in my discussion of green embryos of leguminous seeds ; but more examples might be given, where, although I have no note of the colour of the embryo in the immature seed, its green colour is indicated by its yellowish coloration in the resting seed, as, for instance, in Albizzia Lebbek. In making further reference to the green embryos so In unripe ^ r , ■ . -1 • J leguminous commonly round in unripe or pre-resting leguminous seeds, seeds green one must notice that when the embryo is green the seed-coat is g'^g'^'^^coats*^ generally green, but of a markedly paler shade. However, it "^"^J'j ^° is not difficult to find exceptions to the rule amongst exal- buminous seeds. Thus with Entada polystachya the coats of the soft unripe seed are white, but the enclosed embryo is bright green ; whilst with the Broad Bean {Faba vulgaris^ the embryo is bright green and the coats are white, tinged faintly with greenish yellow. Nevertheless, as indicated in the samples given in the table below, it is evidently the rule for 390 STUDIES IN SEEDS AND FRUITS Table showing the Colours of the Seed-coats and of the Embryo in Soft Immature, and Hard Matured or Resting Leguminous Seeds. Albu- minous Seed-coats, Embryo, or exalbu- minous indicated Immature (pre- resting). Matured Immature (pre- resting). Matured by A. and E. (resting). (resting). Acacia Farnesiana . E. Green Brown Green Pale yellow Bauhinia (species) . A. ,, Black ,, )> Csesalpinia Sappan . E. M Light brown Yellowish Cassia bicapsularis . A. Dark brown green Green ,, fistula . A. ,, Light brown ^, 1) Cytisus Laburnum . E. Dark brown Entada polystachya E. White Brown "^ '1 Faba vulgaris E. Yellowish Brownish '1 (Broad Bean) white green Genista (species) . E. Green Dark brown jj ,j Guilandina bondu- E. ,, Grey White ,, cella Lathyrus pratensis . E. " Mottled black on light brown ground Green " Leucsena glauca E. ^j Dark brown 1 , Phaseolus m u 1 1 i- E. Pink Mottled black Pale green White florus on pink ground Phaseolus m u 1 1 i- E. White White jj florus Pisum sativum E. Green Green Green Green (wrinkled) Pisum sativum E. ,j Yellowish ,, Pale yellow (unwrinkled) white Poinciana regia A. ,, Dark mottled ,, ,, Sophora tomentosa E. Yellowish green Light brown Pale green " Spartium (species) . E. Green Dark brown Green ,, Ulex europseus E. " Chocolate brown " " Vicia Cracca . E. ,, Dark mottled ,, ,, ,, sativa . E. ,, J, ,, ,, ,, sepium . E. »> " " " Supplementary Results for Seeds of other Orders. Thespesia populnea Gossypium barba- dense Dodonrea viscosa Convolvulus Batatas Ipomoea tuba . A, A, E. A. A, White Pale green White Brown Black Brown Whitish Pale greenish yellow Pale green Dark green Whitish Dirty white Pale yellow SEED-COLORATION 391 this family, both for albuminous and exalbuminous seeds, that in the pre-resting or so-called immature state green embryos and green seed-coats go together. In this connection I come now to refer especially to the The colour- behaviour of the embryos of albuminous seeds. The changes the enSyos in colour which the green embryos of full-si^ed unripe seeds °us greeds"" undergo when entering the resting stage, and subsequently when germination begins, are well illustrated for leguminous plants by Poinciana regia, Cassia fistula^ and Bauhinia. These seeds are similar in the general sense that the embryos, which have large foliaceous cotyledons, are nearly as long and as broad as the seed, and are placed between two slab-like masses of albumen. When these seeds are here characterised as unripe or immature, reference is made only to those seeds that have reached the maximum size in the soft condition in the green pod. In such seeds the embryos, being fully formed, are, as shown in Chapter XIX, quite able to dispense with the rest- ing stage altogether, and to proceed with their growth, or to germinate, as we term it. But on account of the cutting off of the fluid supplies and the drying of the pod, a resting period is imposed upon them. It is this break in the continuity of its life that is strikingly shown in the changes of the colora- tion of the embryo. In this respect my remarks will be mainly confined to the asillus- behaviour of the embryo of Poinciana regia, and I will make gggds of but a brief reference to those of the other two plants, since J^^JJ;'*"* they behave in a similar manner. In a green semi-ligneous pod of full size the seeds of Poinciana regia are pale green, and possess soft, flexible coverings. The albumen is white or colourless, and remains so during the subsequent stages. The embryo has dark green cotyledons and a white caulicle or stem ; whilst the pale green plumular bud, which is partly expanded and stands 3 millimetres high, is far more suggestive of a continuously growing plantlet than of an embryo that is shortly to be compelled by the stress of circumstances to enter the rest-period. When the pod begins to dry and to 392 STUDIES IN SEEDS AND FRUITS turn brown, the seeds respond ; and as they shrink become much paler, almost white, except in the centre, which becomes darker. The embryo is also paler. When the pod has become dry and the seeds have acquired their normal resting characters, the signs of active life are gone. The entire embryo has now a yellowish hue, and the plumular bud, that was beginning to expand before the influences resulting in the suspension of the life of the embryo prevailed, has now closed up again, its tiny leaves being closely appressed. But the impress of the rest-period remains after the seed has begun to germinate. The signs of wakening vitality in the embryo in the early stage of germination are only displayed in the lengthening of the axis or stem by the growth of the radicle. The cotyledons maintain their yellowish, lifeless hue ; whilst the plumule bleaches and becomes quite colourless, making little or no effort to unfold its leaves again, until the protruding radicle exceeds three-fifths of an inch in length. The bleaching of the green plumule of the pre-resting seed as the resting stage is reached probably occurs also with cucurbitaceous plants. It seems to happen with the Walnut (Juglans). Lord Avebury speaks of the plumule bearing five or six rudimentary leaves "often just tipped with green" (^Contribution to our Knowledge of Seedlings, i. 8). The lethargy displayed in the waking up of the dormant embryo of Poinciana regia came quite as a revelation to me. For some time after the radicle has struck into the soil the plantlet's existence is mainly hypocotylar and scarcely cotyle- donary or plumular. The plumular bud is very slow to unfold its tiny leaves, so long closed up during the rest- period ; but gradually it assumes a pale greenish -yellow colour. So also the cotyledons retain their yellowish, lifeless hue during all the germinating process. But notwithstanding, they considerably increase in size by absorbing the albumen ; and by the time they have extricated themselves from the seed-case there is but little of the reserve food left. They soon then acquire a more active green, and the plumule SEED-COLORATION 393 similarly responding to the young plant's needs, vigorous growth begins. The lesson of the embryo of Poinciana regia will be dealt with in Chapter XIX, where the general subject of the rest- period of seeds is discussed. Brief reference will now be made to the other two leguminous plants possessing the same seed-structure, namely. Cassia fistula yiwd Bauhinia. They Cassia ••If,- J , 1 • c -L. fistula and behave in a similar rashion as regards the coloration or the sauhinia. embryo, which in the immature seed is bright green in the first-named and pale green in the last-named plant ; but in both cases it assumes a pale yellowish hue when the seed enters upon the resting stage. The alteration in the colour of the embryo in seeds of Non-legu- the foregoing type, where the embryo with large foliaceous seeds, such cotyledons lies between two slabs of albumen, sometimes ^repitaifs and amounts to complete decoloration as the seed proceeds with Coiubrina ^ . asiatica. its development. Thus, in Hura crepitans (Euphorbiaceae) the embryo is coloured green only in the earlier stage of the growth of the seed. When the seed has attained its full size in the ripening fruit, and before the drying process has begun, the embryo is already white, and that state it retains. It would seem that the embryo in seeds of this type, whether it be white or yellowish in the resting seed, has usually functioning green cotyledons in the unripe or pre -resting seed. This seems to be true also of the seeds of Ricinus communis^ in which, although (as I found) the embryo of the resting seed is always colourless, the cotyledons may be green in the pre-resting stage (see Pfeffer's Physiology of Plants^ i. 596). It may be inferred again in the case of Coiubrina asiatica (Rhamnaceae), where resting seeds of the same structural type have pale yellow embryos. Since in the leguminous embryo the hypocotyl is relatively The colour insignificant, the cotyledons making up its mass, I have usually cotyl wfen spoken of green embryos rather than of embryos with green {elo^fare cotyledons. It would seem, however, that in this family when green. the cotyledons are green the caulicle or hypocotyl may be 394 STUDIES IN SEEDS AND FRUITS either white or green. But this point has only arisen during- the preparation of these pages ; and I append a few results of observations respecting it in the case of plants ready at my hand. Observations on the Colour of the Hypocotyl or Caulicle IN Embryos of Soft, Full-grown, Unripe Seeds when the. Cotyledons are Green. (With the exception of the two last in the first column all are leguminous.) Entire embryo green. Cotyledons green, hypocotyl white. Cytisus Laburnum. Acacia (species). Lathyrus pratensis. Leucsena glauca. Genista (species of). Cassia bicapsularis. Ulex europaeus. Pisum sativum. Sophora tomentosa. Spartium (species of). Euonymus. Acer Pseudo-platanus. Vicia sepium. , , sativa. ,, Cracca. Poinciana regia. Faba vulgaris. Note. — When the entire embryo is green, and the colour is dark, the hypocotyl is often paler than the cotyledons. With the exception of wrinkled Peas {Pisum sativum).^ green embryos have not often come under my notice in the typical resting seed. That they are not infrequent, however, is shown in the statement made in The Natural History of Plants (i. 622) of Kerner and Oliver that in Firs and Pines, Maples, and some Crucifera, in Loranthus^ Mistletoe, and the Japanese Sophora.^ the cotyledons are green whilst enclosed in the seed, their epidermis being provided with stomata. In this connection it may be noted that fruits of the Sycamore Maple {Acer Pseudo- platanus)., which I have been keeping for nearly two years, have embryos dark green in hue. It should be remarked that in the green fruit of this tree the green embryo is enclosed in whitish seed-coverings. Here also reference may be made to the dark green embryo of the seed of Montrichardia arborescens., an arborescent aroid that came under my observation SEED-COLORATION 395 in Tobago and Grenada. The seed is exalbuminous, and has a thin covering or skin, which, if I remember rightly, is brownish in colour. As I write I have by my side the exalbuminous seeds of Monstera pertusa^ a climbing aroid, gathered in Grenada over a year ago. In this case also the embryo is dark green. Further details on this subject will be found in Note 18 of the Appendix. Before leaving the subject of the colour of embryos, a Theconnec- few more remarks may be made on the relation in ex- albuminous albuminous seeds of the Leguminosae between white embryos relSng"seeds and the colour of the seed. If we can judge from the between V,7 white em- behaviour of the genera Canavalia and Phaseolus^ where bryosand white embryos seem characteristic, there is no connection theseed!'^° between the two. The four species of Canavalia with which I am acquainted, C. ensiformis^ C. gladiata, C. obtusifolia, and a Tobago species, of which the specific name is unknown to me, all have white embryos. In the first the seed is white, in the second dull red, in the third banded brown, and in the last pale brown. The same indication is afforded by four kinds of Phaseolus^ all of which have white kernels or embryos, namely, P. vulgaris (French Bean), with reddish-brown seeds ; P. multiflorus (Scarlet-runner), having two varieties, one with seeds showing black mottling on a reddish ground, the other with white seeds ; and a West Indian species with white seeds. From these data it would seem that, as already pointed out in the case of leguminous exalbuminous seeds where the embryo is green in the unripe and yellowish in the resting seed, there is no connection between the colour of the embryo and the coloration of the seed-coats. The hard red seeds of different species of tropical The red leguminous plants, such as Abrus precatorius^ Adenanthera leguminous pavonina^ different species of Erythrina^ etc., invite attention for pla"*^. many reasons. As presented to view in the opening pods, they must often attract the notice of birds ; but with the exception of Erythrina and Adenanthera I have not come upon many references to birds selecting them for food. In my 396 STUDIES IN SEEDS AND FRUITS book on the Solomon Islands (p. 293) allusion is made to my finding cracked seeds of Adenanthera pavonina in the gizzard of a Nicobar pigeon, and it is evident that they also serve as food for Indian parrots (Mr J. Scott in More Letters of Charles Darwin, ii. 349). Then, again, the Layards observed in New Caledonia that a small crow and different species of parrots fed on the seeds of Erythrina {Ibis, vi. 1882). But this is all that I know of the matter. Mr P. H. Gosse in his book on The Birds of Jamaica names a number of seeds that are eaten by them, but no mention is made of any of the hard red seeds above noticed. I will here confine my attention to the conditions under which such seeds acquire their red colour and to the changes they experience in this respect when absorbing water for germination, as illustrated by those of Abrus precatorius, Aden- anthera pavonina, 2indi Canavalia gladiata. All three have these features in common. They go through the shrinking and colouring processes in the closed pod ; in all of them the soft unripe seed is rose-pink in hue, the change from pink to red representing the last stage in the coloration ; and, lastly, the colouring matter is readily dissolved out in water, when the coats are pierced or the cuticle is not intact. In the case of Canavalia gladiata the soft, unripe, rose-pink seed belongs to the green pod. As the pod dries, the shrink- ing and hardening seed assumes a bright red hue, and by the time the pod is well dried and on the point of dehiscence the seed is fully contracted and dull red. Whilst drying, the pod does not discolour or darken, as seems to be the rule with legumes having dark-coloured seeds ; but it becomes gradually paler, and finally has a light brown, parchment-like appearance. When absorbing water and swelling for germination, the seed first resumes its original pink hue and then becomes a chestnut- brown. As shown below under Abrus precatorius, where the same thing occurs, this return of the germinating seed to the colour of immaturity is really due to the hydration of the coverings, the colouring matter being to some extent washed catonus. SEED-COLORATION 397 out, so that the coats have the sodden, wrinkled appearance of a washerwoman's fingers. With Abrus precatorius, as with Canavalia gladiata^ the Abruspre- characters of the resting seed are all acquired in the closed pod. Here again the pink colour of the soft immature seed in the green pod gives place to a scarlet red in the hard matured seed of a pod about to open ; and here also, when the seed swells for germination, it resumes the original pink hue of immaturity. With both these plants there is but little difference either in colour, size, or consistence between an immature seed taken from a green pod and a resting seed that has swollen for germination. That this change from red to pink in the germinating seed is due to the hydration of the coverings is indicated by the behaviour of the seed when allowed to soak in water after being filed. The colouring matter dissolves out and the red seed becomes almost blanched. In the case also of the seeds of Adenanthera pavonina all the colour-changes take place in the closed pod. As observed for me by Mrs H. B. Warde in Jamaica, the soft immature seeds in the green pod are first green and then yellow ; and as the pod begins to brown and dry the seeds turn pink, and when the shrinking and hardening processes approach comple- tion the permanent bright red hue is assumed. It is not necessary for the development of the last two stages in the coloration that the pod should remain connected with the plant. If the green pods are allowed to dry after picking, bright red seeds will be found in closed pods after a few weeks. Though I had not the opportunity of observing the actual stages in coloration on the tree, specimens of all of them were kept for me. When the seeds are absorbing water and swell- ing for germination, they do not, as with those of Abrus precatorius and Canavalia gladiata^ regain the pink hue of immaturity. The outer red skin is thrown oflF by the swelling of the under layer, and the seed assumes the yellowish colour of the second stage of coloration in the green pod, a stage earlier in immaturity. 398 STUDIES IN SEEDS AND FRUITS The longj duration of the red colour de- pends on the impermea- bihty of the seed. Red seeds do not stain water red. The red seeds of these three plants are more or less impermeable to liquid water. With Adenanthera pavonina all the seeds are typically impermeable, with Ahrus precatorius the great majority are, and with Canavalia gladiata the minority perhaps are impermeable. The impermeable seeds are of necessity non-hygroscopic, whilst the permeable seeds behave hygroscopically. A seed that responded in its changes of weight to the varying degrees of humidity of the atmo- sphere could scarcely be expected to retain its bright red hue for a long period. Where the impermeable cuticle remains intact the colour ought to withstand the test of centuries. This would probably be true of the seeds of Adenanthera pavonina. Except when kept in unusually dry conditions, I should not expect the seeds of Ahrus precatorius to preserve untarnished for many years their original scarlet hue, since the scar is their point of weak- ness. As an indication in this direction I may refer to 35 seeds of Ahrus precatorius^ now beside me, which I gathered from the plant in Fiji twelve years ago. Only 25, or 71 per cent., retain their original bright colour, the rest being brownish or even blackish in hue. This is a point in the history of red seeds that seems to be worthy of further investigation — I mean the permanence of the colour. Abundant data would be at hand in our own museums and in tropical countries. I may here refer to a curious fact that must be well known to students of vegetable chemistry. Although the red seeds of Ahrus precatorius^ Adenanthera pavonina^ and Canavalia gladiata stain water freely when the water is allowed to penetrate their coats, the solution has not the colour of the seeds. The seeds of Adenanthera impart a beautiful amber hue to the water, which deepens after the seeds have been removed and becomes like brown sherry. On the other hand, the seeds of Ahrus and Canavalia stain the water a dark green, which deepens after the seeds have been taken out, becoming steely or almost inky. SEED-COLORATION 399 The colour of the kernel or embryo in the case of these The colour three red seeds in the resting state varies somewhat. In kernels in Adenanthera pavonina it is yellowish, which indicates that the \^^ seeds of embryo was green in the unripe seed. In Canavalia gladiata osae. it is white. In Abrus precatorius it is tinged yellow outside, but is white on section. Before quitting the subject of red seeds I will briefly refer The stages to the stages of coloration of the orange or scarlet seeds of tLi^oaie'^*" Iris fcetidissima, as observed by me during two seasons in a p^^|.°f.^"s wood at Salcombe. All the stages took place in the moist closed capsule long before the fruit dehisced or showed any signs of drying up. Up to the middle of August, when the fruits were dark green and averaged from loo to i lo grains in weight, the seeds were white and soft, with jelly-like contents and no recognisable embryo. After this, as the capsules con- tinued to increase in size the seeds became pale yellow ; but their contents remained unchanged until the end of the month, when they began to solidify. September was principally occupied with the maturation of the seeds and with the growth of the fruit, the seeds assuming a deeper yellow hue. By the middle of October, when the capsules averaged i6o to 170 grains in weight, the seeds were solid, of full size, and of the typical orange or scarlet colour. There was not much altera- tion either in seed or fruit after this date, and in the beginning of November the most advanced capsules commenced to dehisce, and the remainder followed during the course of the month. It is worth noting that seeds may acquire much of the Thecolora- coloration which they possess as resting seeds whilst their ^hnst^Sie? contents are still fluid and before the embryo is formed, contents are _,, . , , , , 1 • n • 1 , , still fluid. This would seem to take place chiefly with monocotyledonous albuminous seeds of the type below exemplified. Thus, the white soft seeds of Allium ursinum, when they begin to ripen in the closed capsule, become dark red, though little more than bags of water. In the same way, the white soft seeds of the green capsules of Iris fatidissima colour yellow as the fruit 400 STUDIES IN SEEDS AND FRUITS ripens, though the embryo is not yet recognisable and the albumen is a mere mass of jelly. I should imagine also with Scilla nutans that seed-coloration precedes the formation of the embryo, the pearl-white immature seeds being merely sacs of fluid. Reference has already been made in the early part of this chapter to the red hue of the seed of Barringtonia speciosa when its contents are quite fluid. SUMMARY (i) The inquiry is mainly directed to the conditions of seed- coloration, to the How rather than to the Why (p. 368). (2) Questions relating to the specially adaptive nature of the colours of seeds are summarily dismissed on the ground that we are not justified in selecting one character that happens to be conspicuous to our senses, whilst ignoring the great number of other characters that can make no' such appeal to us (p. 368). (3) After referring to the wealth of seed-colour displayed in a typical native garden in Jamaica, the author cites cases of the develop- ment and disappearance of seed-colours before the fruit is ripe or before the seeds are exposed to view (p. 369). (4) After it has been shown that as a general rule seeds colour in the closed fruit, as illustrated in the case of berries, capsules, and legumes, inquiry is made as to the stage in the history of the fruit in which the coloration takes place, whether in the green, the ripe, or the drying stage, or in all three of them. Though it is established that coloration frequently takes place in the moist green and ripe capsule, the subject is acknowledged to be a very difficult one, especially as concerns the legume (p. 370). (5) In this connection reference is made first to the experiment of Lubimenko, in which the seeds of young leguminous pods were exposed to the outer air by removing portions of the pods, and to the conclusion drawn that for the normal development of the seed a con- fined atmosphere of stable composition is needed (p. 373). (6) The author then gives the results of his similar experiments in the case of green capsules of Scilla nutans on the plant, the upshot being that immature seeds exposed to the outer air by windows cut in the fruit-walls developed normally, with the exception of their failure to acquire the black colour of the resting seed. It thus became evident that the seeds acquire their shining black hue only in the confined atmosphere of the moist green capsule (p. 374). SEED-COLORATION 40 1 (7) The conditions under which seeds colour in leguminous pods and in capsules are discussed in detail, and the black and brown forms of coloration, including black mottling, are especially dealt with. After observing that the coloration, early shrinking, and hardening of the seed and its coats are so conspicuously associated with the early drying of the fruit, that the presumption in favour of there being a causal connection is very strong (more especially in the case of legumes), it is shown that this view is untenable. The results of experiment demonstrate that these changes in the seed can take place under conditions so humid that the drying of the fruit is precluded. In a word, the regime involved in the coloration, early shrinking, and hardening of the seed and its coats, both in the legume and in the capsule, is that which is displayed by the colouring, shrinking, and hardening seed in the moist berry. The seed colours normally in moist fruits and continues the process notwithstanding the drying of the fruit. In some cases, however, as in black motthng, the complete coloration of the seed is interfered with by the fruit's drying (p. 375). (8) The colours of unripe and mature seeds, that is, of pre-resting and resting seeds, are then compared in legumes, capsules, and berries, and the inference is drawn that unripe or pre-resting seeds 'are usually white in capsules and green in legumes (p. 385). (9) It is pointed out that the colour of the coats of the resting seed in leguminous plants has but little connection either with the colour of the coats of the pre-resting or unripe seed, or with the colour of the embryo in the pre-resting and restmg states ; but it is indicated that with unripe leguminous seeds green embryos and green coats usually go together (p. 389). (10) The changes in colour which the green embryos undergo when entering the resting stage and subsequently when germination begins are then discussed in the cases of seeds of Po'tnciana^ Cassia^ and Bauhinia. They all assume a pale yellow lifeless hue in the resting seed, and, as illustrated by the seeds of Poinciana regia^ display consider- able lethargy in the waking up of the dormant embryo during the germinating process. In some non-leguminous seeds, as with those of Hura crepitans^ the green embryo becomes decolorised and blanched when entering the resting state (p. 391). (11) The colour of the hypocotyl in leguminous embryos, as with green cotyledons, is shown to be sometimes white and sometimes green (p. 393)- (12) It is observed that whilst the embryos of leguminous resting seeds are usually pale yellow, green embryos are not uncommon in the resting seeds of other orders (p. 394). (13) With regard to white embryos in leguminous resting seeds, it is remarked that since in genera like Phaseolus and Canavalia^ where 26 402 STUDIES IN SEEDS AND FRUITS they occur, the coloration of the seed-coats varies greatly, there is no connection between the two characters (p. 395). (14) The red seeds of leguminous plants are then discussed, especially with reference to the conditions in which the red colour is produced and to the changes caused during germination. Those of Abrus precatorius^ Adenanthera pavon'ina^ and Canavalia gladiata are taken as examples (p. 395). (15) It is shown that the long duration of the red colour of the above seeds depends on the impermeability of their coats and on the non-hygroscopic behaviour of the seeds (p. 398). (16) The stages in the coloration of the seeds of Iris foettdissima are described (p. 399). (17) The coloration of immature seeds whilst their contents are still fluid and before the embryo is recognisable is remarked in the cases of some monocotyledons (p. 399). CHAPTER XVIII THE WEIGHT OF THE EMBRYO Any discussion of the proportional weight of the embryo in A difficult albuminous resting seeds must be surrounded by a host of ^" ^^^ ' difficulties. The first question to present itself is concerned with the utility of such a discussion, since, if we cannot bring the subject into some sort of relation with other matters affecting the seed, it would not be worth while following it up. But if this seems feasible, we are at once confronted with other difficult questions. Although not directly con- cerned with exalbuminous seeds, we cannot ignore their close connection with albuminous seeds. We must at the outset select some standpoint for viewing this relationship, and much depends on our answer to the query — Which is the older of the two ? I suppose that on biological grounds there can be no doubt The that the albuminous is the primal state ; but one has only to and the watch a germinating seed bravely endeavouring to strike into exalbumin- the soil, whilst its cotyledons still within the seed-case are appropriating the albumen, to decide that the albuminous state is the older condition. We here perceive the transition from an albuminous to an exalbuminous state in actual operation, the chief point of difference being that whilst the change, as generally understood, occurs in the seed before it enters the resting condition, here it takes place in the germinating seed after the resting-state stage is passed. Our plantlet is now doing what is often effected within the seed at an earlier stage 403 404 STUDIES IN SEEDS AND FRUITS before the resting state is imposed. All exalbuminous seeds have in a sense been once albuminous, and the distinction which we draw between exalbuminous and albuminous seeds mainly depends on whether the transition occurs before or after the rest-period. The after-ripening of seeds must be often concerned with the change from the albuminous to the exalbuminous condition. That which happens in the Jasmine, where the albumen is at first copious in the seed and dis- appears when the seed is ripe for germination, occurs with many other plants. It would be quite possible, for instance, in the case of the Ivy (Hederd)^ as described in Chapter XIX, to describe the seed as albuminous in its first stage and as exalbuminous when about to germinate. The interposition of the rest-period in the early portion of a plant's existence is responsible for many false distinctions and many incorrect comparisons. Other diffi- Another difficult point has been already indicated. In the cu t points, i-gstii^g albuminous seed, as is well known, the embryo may exist in all stages, from that in which it is imperfectly differ- entiated to that in which the plumular leaves are developed and we have a perfect plantlet within the seed. Between these two extremes all gradations occur. Then, again, we have often genera with exalbuminous seeds and genera with albuminous seeds in the same order. Thus with Sapotaceae, at first sight the seeds seem very similar in structure. Yet the proportion of albumen may vary in different genera, from that found in Achras^ where it amounts to about 84 per cent, of the kernel's weight, to its condition in Chrysophyllum^ where it may range between 20 and 40 per cent, in different species, whilst in Lucuma there is none at all. Of the 222 families of angio- sperms described in the System of Botany of Le Maout and Decaisne, about 18 per cent, possess both albuminous and exalbuminous seeds. Dr Goebel in his Organography of Plants (English edition, ii. 262) lays stress on the far-reaching nature of the changes in the form of the embryo arising from the different modes of THE WEIGHT OF THE EMBRYO 405 disposition of the food-reserve ; and one has only to refer to Lord Avebury's volume on seedlings in the International Scientific Series to become apprised of this fact. For example, we have its disposition in the cotyledons, as in many leguminous seeds ; its disposition in the hypocotyl, as in the Barringtonia; and with some Guttiferae, and its disposition outside the embryo altogether, as occurs with many plants. The embryo in different resting seeds varies so much in its stage of develop- ment and so much in its relation to the reserve supply of food that one hesitates to consider the matter of its relative weight at all. However, one has only to observe that whilst over 100,000 Juncus embryos are required to make up the weight of a single embryo of the Coco-nut palm, they are in relation to the kernel more than 200 times as heavy, in order to perceive that a study of the secondary relations involved in such measure- ments may lead to some interesting conclusions. Whilst the Coco-nut embryo is in an absolute sense one of the very largest and heaviest amongst embryos of its kind, it is in a relative sense, as compared with the kernel, one of the very smallest and lightest. The subject from this standpoint seems to lend itself for inquiry, and here again the principal instrument of investigation is the balance, weight as a general rule connoting size. The estimation of the weight of the embryo in very small Method of seeds, as in those of the Common Rush (Juncus), is easier than the weight it might at first seem to be. It soon became apparent to me g^^^o*^ that I could begin by ascertaining the proportional bulk and the proportional weight of the embryo as part of the kernel in a seed that could be easily examined, and that by a comparison of the two results a constant error might be found which could be applied in those numerous cases where, owing to the use of the balance being impracticable, one is compelled to rely on the proportional size of the embryo for the clue to its weight. As suiting my purpose the light brown seeds of dehiscing 4o6 STUDIES IN SEEDS AND FRUITS fruits of Iris Pseudacorus were chosen. A hundred embryos were found to make up the weight of a single kernel ; and since the kernel weighed 1-3 grain, this gave -013 grain as the weight of the embryo. It was then ascertained that in point of bulk about 85 embryos went to a kernel, which, interpreted as weight, gave the weight of the embryo as -015 grain, which is 16 per cent, greater than the actual weight determined by the balance. It was then assumed that weight calculated from relative size displayed an error of this amount, but, taking into consideration the fact that the method is at its best crude, and remembering that only approximate results could be looked for, I decided to ignore it altogether and to accept bulk as roughly indicative of weight. As it was evident that in very small seeds one would have to rely mainly on the bulk-data, the question arose as to the method of obtaining them. My choice lay between making my own seed-sections and utilising the materials offered by the illustrations in The System of Botany of Le Maout and Decaisne (English edition, 1873) '■> "^^^ ^^e last method was selected. From the sections of seeds there figured I could procure the requisite measurements, obtaining for myself, when needed, the data for the relative thickness of the seed. But in the first place I tested the method by comparing the results obtained by estimations from the figures in the above-named work with those supplied by actual measurements of the seed. They came fairly close together in the case of the seeds of Iris Pseudacorus^ 85 embryos going to make up a kernel by my own measurements of the seeds, and 75 according to the data offered by the illustration of an Iris seed of the same type in the general work. The seeds of the Elder {Sambucus nigra), weighing only -05 grain, were then employed. Actual measurements of a seed indicated that the embryo formed about a tenth of the bulk of the kernel, whilst the d^ta supplied by the figure of the same seed gave the proportion as one-twelfth. Still smaller seeds, those of Aquilegia weighing only "03 THE WEIGHT OF THE EMBRYO 407 grain, gave similar proportions for the two methods, the size of the embryo, whether obtained from the actual seed or from the illustration, being about ^-o^^ of that of the kernel. In this case the proportional weight of the seed-coats was taken as one-third of the total weight of the seed, which left -02 grain as the weight of the kernel. I was therefore dealing with an embryo which weighed not more than ^^ ^ part of a grain. From the case of the Aquilegia seed to that of the minute seed of a rush (Juncus) the step was not a very difficult one. Having found that about 5500 seeds of Juncus communis went to make a grain, I had to allow for the weight of the seed- coats, which, judging from the figures and from my own examination of the seed, was relatively much less than with a seed of Aquilegia^ so I placed it at about one-fifth of the seed's weight. The weight of the embryo was then calculated to be about one-tenth of that of the kernel, and from these data the following results were obtained : — i Entire seed -^-^qq grain or -00001 18 gramme. Juncus communis -j Kernel esVs 55 -0000094 •>■> i Embryo __i 68,750 •0000009 In this manner, therefore, the minutest of embryos are shown to be within reach of the balance. Whilst nearly 70,000 of the embryos of Juncus communis are required to make a grain, nearly 200 of them placed in line will make an inch, and nearly 8 will make a millimetre, the length of the embryo being one-third that of the seed, which, according to my measurement, is 0*4 millimetre long. With these preliminary remarks I will now give in tabular form the results of my observations on the weights of the embryos of albuminous resting seeds in the case of more than fifty plants. They are arranged in the order of the embryo's relative weight as a portion of the kernel. The indications supplied by such an arrangement are full of suggestiveness, and are especially discussed in the remarks that follow the table, as well as in the last paragraph of the chapter. 4o8 STUDIES IN SEEDS AND FRUITS The Weights of Embryos in Albuminous Resting Seeds. (Palms are indicated by P. When a figure of the seed has been utilised, as in the case of small seeds, L. is placed after the name, the System of Botany of Le Maout and Decaisne being usually employed. See text before and after table for further explanation. ) P. Cocos nucifera . P. Bactris . P. Cocos (schizophylla ?) Tamus communis, L. P. Acrocomia lasiospatha P, Licuala grandis . Mercurialis, L. P. Prestoea montana Anona Cherimolia P. Areca Catechu . P. Elseis guineensis Anona reticulata P. Hyophorbe Verschafftii P. Mauritia setigera P. Sabal umbraculifera P. Manicaria saccifera P. Oreodoxa oleracea P. Livistonia . P. Caryota Aquilegia, L. . Anona palustris . P. Oreodoxa regia . Dracaena Draco . Carex, L. . P. Cocos plumosa . Iris Pseudacorus Luzula, L. Hedera Helix . Veronica, L. Ricinus communis Papaver, L. Canna indica Jatropha Curcas Ravenala madagascariensis Sambucus nigra, L Rumex, L. Weight of the Embryo. Actual weight. Grains. Grammes SfOl 5a "5ff ■jkVtt ■STTT •129600 ■002160 ■012960 ■000015 •003240 ■000810 ■000013 ■001620 ■000648 012960 ■003240 •000648 ■001296 •097200 ■001296 ■064800 ■001080 ■004630 ■009260 ■000006 ■000926 ■001620 "003240 ■000026 •003240 •000810 •000017 •000926 ■000019 •006480 ■000006 ■006480 ■025920 •012960 ■000196 •000260 As a part of the kernel. 100 ■BTTTT TTD"!) rio rh irho Remarks. (S. entire seed C. coats ; K. kernel. Weights in grains. ) A ripe fruit of average weight with mature kernel. A dry fruit. ^5- S. ^27 ; C. ^02 ; K A dry fruit. S. -125 ; C. ^025 ; K. "i A dry fruit. S. 8^o ; C. 3*4; K. 4*6. A moist ripe fruit. A dry fruit. S. 4'o ; C. 1 'o ; K. 3^0. A fresh fruit. A fruit beginning to dry. A dry fruit. A fruit beginning to dry K. 02. K. 2^7. K. 5 -8. ; K. -04 S. ^03 ; C. S. 3^9 ; C. r2 A dry fruit . S. 6^o; C. o^2 S. •OS ; C. -oi an average. Fruit partly dry. Seeds partly dry ; S, 2*0 C. 07; K. r3. S. 03 ; C. •oi ; K. •02. S. •S ; C. '07 ; K. '73. (See Chap, XIX.) S. ■013; C. "003; K. '01 S. 2^7 ; C. '7 ; K. 2^o. S. ^0025 ; C. "0005 ; K. '0020. S. 2-6; C. ^6; K 2^o. S. 120; C. 4^4; K. 7-6, S. 5-5 : C. 2^5 ; K, 3*0. S. ■OS ; C. ■oz ; K. ^03. S. '054 ; C. ^014; K. "04. THE WEIGHT OF THE EMBRYO 409 The Weights of Embryos in Albuminous Resting Seeds — continued. Scrophularia, L. Berberis, L. . Anagallis, L. . Sparganium ramosum Saccoglottis amazonica Juncus, L. Hura crepitans Glaux marilima * . Viola tricolor, L. . Achras Sapota Cassia t . Ipomcea pes-caprae . , , tuberosa . Plantago, L. . Poinciana regia Colubrina asiatica . Chrysophyllum Cainito Exalbuminous seeds Weight of the Embryo. Actual weight. Asa part of the Grains. Grammes. kernel. ^iss •ooooio iV rU •000648 rV !2-bViT •000026 f\ Th •000648 tV r\ •045360 tV 7373-ir •00000 I rV li •II3400 i T.W ■000040 h 5^ •000108 i tV •045360 i I ■064800 h h. •032400 i 6 •388800 k i^U ■000216 h 2 •129600 i i ■010800 1 5 A •362880 i i Remarks. (S. entire seed C. coats ; K. kernel. Weights in grains.) S. "002 ; C. ^0005 ; K. ^0015. S. •h; C. -04 ; K. -I. S. '006 ; C. •ooz ; K. '004. S. •!! ; C. •oi ; K. i (fresh). S. 8^3; C. 1^3 ; K. 7^o. See p. 407. S. 20 ; C. 6 ; K. 14. S. ■oo; ; C. ^002 ; K. "005 S. 016 ; C. ^003 ; K. •013. S. 9^o; C. 4^8; K. 4-2. S. 8 ; C. 2 ; K. 6. S. 37 ; C. 17 ; K. 2^0. S. 22-5 ;C. 4-5 ; K. i8-o. S. -02 ; C. -oi ; K. "oi. S. 9^3; C. 4-3; K. 5-0. S. -60 ; C. -31 ; K. -29. S. 12; C. 5; K. 7. * Proportions of the embryo of Glaux maritima from figure in Das PJlanzenreich, iv. 237. t This represents a mean result for three species of Cassia {fistula, grandis, marginata). Here we have the weights of the embryos for about fifty- Remarks on three species and generic types of albuminous seeds, of which table, almost a third belong to palms. For reasons before implied, the relative weight of the embryo in proportion to the kernel is alone given. This is the only ratio that is generally applic- able, it being apparent that the inclusion of the seed-coats in the discussion would have but little significance in a series comprising the seeds of palms, whilst the proportional weight of the embryo with regard to the fruit would possess but little value, if only for the reason of there being many-seeded as well as single-seeded fruits. If required, the proportional weight of the embryo with reference to the entire seed (kernel and coats) can be 4IO STUDIES IN SEEDS AND FRUITS determined for any ordinary seed by making use of the data given in the last column of the table. Thus, in the instance of Canna indica, the embryo is given as ^ of the weight of the kernel ; but by employing the data given for the entire seed we obtain a proportional weight of gV- Then, again, if further particulars relating to the palm-embryos are needed, the weight of the kernel can be readily calculated from the results given in these columns, and by referring to Chapter XIV the weight of the entire fruit can in most cases be found. For instance, in the case of ripe coco-nuts, where the kernel makes up about one-fourth of the weight of the fruit, the total weight for the fruit illustrated in the table would be about 17,500 grains, and the relative weight of the embryo would be there- fore about s-tVo- Coco-nuts vary so much in weight both in their several stages and in their different varieties that the same result can be scarcely looked for. So again with the dry fruit of Licuala grandis^ since the embryo weighing -^-^ grain is equal to -5-^ of the weight of the kernel, we obtain 6*2 grains as the kernel's weight. \n a table in Chapter XIV the average weight of the entire fruit is stated to be 10 grains, from which the relative weight of the embryo may be placed at g-^^ of the fruit's weight. Another important point is that we are here dealing with the embryos of resting seeds. In most cases this is a seed that has dried spontaneously on the plant ; but with palms several other considerations arise, and it is often more than probable that a palm seed which has completed the normal drying process has lost its capacity for reproducing the plant. In some palm seeds it is likely, as in the case of that of the coco-nut, that during the ripening of the kernel the oil increases as the water diminishes. The embryos With nearly all the palm seeds experimented upon the ofpams. embryos belonged to dry fruits and were more or less shrunken. In a few instances, as with Cocos nucifera^ Mauritia setigera, and ^reca Catechu, the fruits were ripe and still moist, the kernel reaching maturity after the husk had commenced to THE WEIGHT OF THE EMBRYO 41 •dry. According as we take the ripe moist fruit or the fruit that has completed its drying process, the proportional weight of the embryo as concerns the kernel varies greatly. This is due to the fact that whilst the albumen of such a ripe fruit usually loses when dried about a third of its weight, the embryo loses about two-thirds. The effect of this, as shown in the results given below, is to halve the relative weight of the embryo as compared with the kernel in the spontaneously dried fruit. Thus with Cocos nucifera the embryo is •grV'e" °^ the weight of the kernel in the ripe fruit and 4X6-5 in the dry fruit many months old, though it is more than doubtful whether the shrunken embryo in the last case would retain its vitality. Table showing the Loss of Weight, when Drying spontaneously, OF THE Albumen and Embryo of the Ripe Seeds of Palms, AND ITS Effect on the Proportional Weight of the Embryo. Loss of weight. Proportional weight of the embryo as a part of the kernel. Albumen, Embryo. Ripe seed. Dry seed. Cocos nucifera Acrocomia lasiospatha . Mauritia setigera . Areca Catechu Oreodoxa oleracea Prestcea montana . Licuala grandis 33 per cent. 40 1? :; 66 per cent. 66 per cent. 66 „ 70 per cent. 60 5^ It is thus seen that the discussion of this subject as concerns the palm embryo presents serious difficulties. Every type of palm seed would probably have its own regime, and we are for ever thwarted by the interposition of the rest-period and its results. An important contrast is brought out by observing the behaviour of different embryos when placed in water. In the ripe coco-nut the embryo fills its cavity The embryo J . , . r . J • i. of Cocos and IS more or less in a state or saturation as regards its nucifera. water-contents, since on being placed in water its weight 412 STUDIES IN SEEDS AND FRUITS remains unchanged or is increased only 2 or 3 per cent. As the fruit proceeds with its drying the albumen and the embryo lose weight together during the first few months. The weight of the embryo, originally about 2 grains, is reduced to I '8 grain after three months and to i'5 after six months, the embryo still almost filling its cavity and showing only a slight collapse at the sides, whilst its weight is only increased by 14 or 15 per cent, when placed in water. There is but little marked shrinkage in the form of the embryo, a result that is certainly due to the increase in its oily constituents ; and we have already seen in Chapter XIV that as the kernel ripens the oil increases in amount. It is only with very old lifeless coco-nuts that we would expect to notice great shrink- age of the embryo. Thus the behaviour of the embryo of the coco-nut is evidently peculiar. When comparable with those of other palms in the early mature state of the fruit it is, as shown in the table above, full of water, losing two-thirds of its weight when drying spontaneously in the detached condition, and returning to its original weight when resting on water. But in the later stages of maturation, when the oil in the kernel increases in quantity and the fruit is drying, the embryo becomes more oily, and loses but little water when detached and allowed to dry, and only increases its weight slightly when placed on water. This peculiarity is also brought out in the next table. The embryos Now, the embryos of most of the other palms examined pahM^'^ behaved very differently, as shown in the results below tabulated. Whilst with the fresh ripe fruits the embryo usually fills its cavity and is more or less in a state of satur- ation, it shrinks considerably as the fruit dries, so that in the course of two or three months it usually presents itself as a more or less shrivelled object but partly filling the cavity. Such shrunken embryos, when placed in water, double their weight in a couple of hours and regain their original form. The shrinking of the embryo may even be evident in ripe THE WEIGHT OF THE EMBRYO 413 fruits that have been gathered only two or three weeks. This happens with the embryos of Mauritia and Cocos plumosa. Table showing the Condition of Palm-embryos in Fruits some time after gathering, and their behaviour when allowed to rest on the surface of water. Period since gathered. Condition of the embryo. Increase of weight when placed on water. Oreodoxa regia . 7 weeks Rather shrunk 40 per cent. ,, oleracea 8 months Much shrunk 130 .. Sabal umbraculifera 16 ,, Shrunk 160 ,, Acrocomia lasiospatha I^ n Somewhat shrunk 100 ,, Bactris 2 .) A little shrunk Cocos nucifera 6 „ Slightly shrunken 15 per cent. ,, plumosa . 2 weeks Much shrunk 100 „ ,, schizophylla 2 years Very much shrunk 200 ,, Areca Catechu 5 weeks Greatly shrunken 200 ,, Licuala grandis . 18 months Much shrunk 100 ,, Prestoea montana . 4 M Greatly shrunk 250 ,, Hyophorbe . 2 .. Shrunk Elreis guineensis . 2 ,, Rather shrunk 90 per cent. Livistonia . 2 years Much shrunk Caryota 2 months Rather shrunk Although in four cases the behaviour of the shrunken embryo in water was not tested, it is evident that the same effect would be produced. The gain of the shrunken embryo in water represents the original loss in the drying process, the embryo acquiring its full outlines and approximately its original size and weight in the moist ripe fruit. It should be allowed to rest on the surface of the water for a couple of hours, when it usually ceases to gain weight. The albumen of the seed containing the shrunken embryo takes up much less water. Thus, to take the behaviour of a Sahal seed sixteen months old, whilst the albumen, when placed in water, added only 33 per cent, to its weight, the embryo increased its weight by 160 per cent. This is consistent with the principle before stated that when a ripe palm fruit dries spontaneously the albumen loses about one-third of its weight and the embryo about two-thirds. That palm seeds retain their vitality but a short time would seem to be the rule. Mr Hart, late Superintendent 414 STUDIES IN SEEDS AND FRUITS of the Botanic Gardens at Trinidad, tells me that the limit for Acrocomia, Oreodoxa, Sabal, Thrinax^ etc., when the fruits are protected from the sun and rain, would be from three to six months, whilst for Mauritia^ he says, the limit would be only a week or two. Unless the embryo increases its oil during its loss of water in the drying process its longevity would seem to be but slight. The reciprocal relation between oil and water in the embryo is a matter of importance for certain palms. Thus I would assume that the oily embryo of El^eis guineensis would possess a greater staying capacity than the watery embryo of Areca Catechu^ though in both cases marked shrinkage might take place in the case of the embryo removed from the fresh ripe fruit. Let the fruits remain on the palm and the difference in the behaviour would assert itself ; but if both fruits are allowed to dry in the detached condition, their embryos will probably be similarly shrunken. One would only look for the contrast in the condition of the embryos in the case of fruits that have dried on the tree. Speaking generally, I would consider that palm seeds where the albumen is ruminate, as with Areca^ Caryota^ and Thrinax^ would preserve their vitality for a much shorter period than where it is homogeneous, as in the majority of palms, the more rapid drying of the ruminate albumen being promoted by its peculiar structure. One may mention in passing a fact which, though familiar to the students of palm fruits, may be new to some of my The water of readers. The water filling the cavity of the fruit is not e coco-nu . p^^^^j^j- ^q Cocos nucifera (Coco-nut), but occurs with other palms of the same tribe of the family, at least in the immature condition of the fruit, and in fruits so small that 300 or 400 of them are needed to make up the weight of a single coco-nut. Thus in Bactris the hard black shell or endocarp of the ripe fruit is soft and white in the immature fruit. The cavity within the shell of the young fruit is lined by jelly-like albumen and is quite full of water. As maturation proceeds the albumen solidifies and increases so as ultimately to fill THE WEIGHT OF THE EMBRYO 415 the cavity, leaving only traces of the original central hollow. The narrow fissure-like central cavity found in the albumen of moist ripe fruits of other genera of the same tribe of palms, such as Acrocomia and ELeis^ gives an indication of a similar history of the young fruit. I may remark that even the large cavity of the coco-nut may be sometimes nearly obliterated. The kernel is so thick in a variety growing in the Moluccas that there is scarcely any central space (Tropical Agriculturist for 1833). It would be possible to greatly lengthen this chapter, but I have approached the subject mainly as a preliminary to the discussion of the rest-period in the next chapter. Measurements of weight may have but little meaning in themselves, but they acquire a significance when we arrange them in order, as in the general table before given. The indications of the separate seeds are full of suggestiveness. Here we have a series of resting seeds, beginning with those where the embryo constitutes only ^o^oo °^ ^^^ weight of the kernel (the other 1999 parts belonging to the albumen) and ending with seeds where the embryo has appropriated all the reserve-food, storing it either in its cotyledons or in other parts of its substance. The series speaks eloquently of the true relation between an albuminous and an exalbuminous seed. But it is as concerns the rest-period of seeds that Therest- the story of these figures has its most important lesson for g^eds. ° us, nature having imposed it on the young plant at all stages of its early development. It is around the mystery of the resting seed that the discussion in the following chapter will chiefly centre. SUMMARY (i) Any discussion of the weight and size of the embryos of rest- ing albuminous seeds must be beset with difficulties. Foremost comes that concerned with the relation between the albuminous and exalbuminous state, and it is assumed that the first is the primal condition. Then there are the disturbing facts that not infrequently 41 6 STUDIES IN SEEDS AND FRUITS the same order possesses plants with both albuminous and exalbuminous seeds, and that the rest-period has been imposed on the embryo in very different stages of its development (p. 403). (2) The method of measuring the weight of minute embryos is then discussed, and it is shown that even those of Juncus communis^ of which nearly 70,000 go to a grain, are not beyond the reach of the balance (p. 405). (3) The author then gives his results for the weight of embryos in the case of more than fifty kinds of plants, of which nearly a third belong to palms. The only weight-ratio of the embryo that is given is its proportional weight as a part of the kernel, though in the case of ordinary seeds data for estimating other ratios are added. It is con- sidered that the embryo-kernel ratio is the only one that is generally applicable ; and in the table the results are arranged in order, beginning with those seeds where the embryo forms a very small proportion of the albumen and terminating with the exalbuminous seed, where the albumen has all been appropriated by the embryo (p. 408). (4) The embryos of palms are specially dealt with, and though the subject has many difficulties, some broad results follow from the author's observations. In the first case it is shown that as^ a rule the embryo shrinks in the drying fruit about twice as much as the albumen, and that in consequence its proportional weight with regard to the kernel is much less in the dry than in the moist fruit (p. 410), (5) Then stress is laid on the fact that this great shrinkage of the embryo in the drying fruit usually occurs in the first few weeks or months, and that for this reason the seeds of palms could scarcely be expected to retain their vitality for a long period. Facts are given which show that although the period may be as little as two or three weeks, it does not generally exceed six months. On account of rapid drying being favoured by their peculiar structure, it is con- sidered that ruminate seeds of palms would possess the least staying power (p. 412). (6) Allusion is made in passing to the fact that the coco-nut is not peculiar in possessing in the unripe condition a large cavity filled with water, the fruits of other genera of the same tribe of palms being thus characterised, even in cases where they are only an inch in size when full-grown (p. 414). (7) The author concludes with the remark that he has approached the subject of the weight and size of embryos mainly as preliminary to the discussion of the rest-period in the next chapter. CHAPTER XIX THE REST-PERIOD OF SEEDS If everything comes from the egg, it is certain all lines of biological investigation lead back to it, and thus it is that in the resting seed our interests converge towards that point in a plant's life. Many botanists of eminence have dealt with this matter, and the salient facts must be familiar to my readers. There are, however, certain aspects of the subject which have come more particularly under my notice ; and ever since I studied the vivipary of Mangroves, like Rhizophora and Bruguiera, in the Pacific, twelve to fifteen years ago, the matter of the rest- period of seeds has been frequently before my mind. The rest-period represents a break in the continuity of the A break in young plant's existence. As Goebel well puts it, the seed here of a*^young^ ^ submits to an interruption in its development ; and so it is P'^"^'^ '*^^- that the real mystery of the seed lies not so much in the resumption of active vitality implied in germination as in the suspension of its vitality. That singular phase in a seed's life, the entering into the rest-period when it is quite able to proceed continuously with its growth and to go on to germination, is the true mystery. The potential vivipary of plants I hope to establish later, meaning thereby the inherent ability of the embryo to proceed with its growth. But in the first place it is necessary to acquire a correct The degree notion of the prevalence of the resting habit in seeds, since we vaieiTce of are apt to invest this character of seeds with a universality that h'^t-^.^fn^"^ it does not actually possess and with a persistence that it can seeds. 417 27 41 8 STUDIES IN SEEDS AND FRUITS scarcely claim. Since much of the substance of the two follow- ing pages will be found in the text-books, it will be sufficient to refer to the works consulted at the end of the chapter. Goebel points out that whilst with Ferns, Lycopods, and Equisetums there is no rest-period, in Seed Plants with few exceptions the embryo experiences sooner or later an interrup- tion of its development which is resumed in germination. In reality, though the pronounced exceptions are few, there are many in degree, as is indicated by the transient nature of the rest-period in a large number of plants and by the prevalence of after-ripening, a term appUed to the growth of the embryo in the resting seed before and after detachment from the parent plant. As regards the shortness of the period, let us take our own forest trees. The seeds of the Oak, Beech, Elm, Poplar, Horse-chestnut, Maple, Fir, etc., have, as is well known, a very transient germinative capacity, as a rule only preserving this power until the next spring, and even then usually requiring to be planted soon after gathering. Probably not a few of them, when their seeds germinate during a mild autumn, supplement the short rest-period within the seed by a period of repose outside the seed, remaining stationary during the winter under the protection of the fallen leaves, and domg little more than protrude their radicle an inch or two. This capacity in the case of acorns is alluded to later on in this chapter. Many other seeds are known to behave similarly. Thus, those of Oxalis and Salix fail usually after a few weeks or months. Professor Ewart strikes the true note when he remarks that in very many cases seeds are very intolerant of even ordinary air-drying. According to De CandoUe, the seeds of most Rubiaceae, Myrtaceae, and Lauraceae lose their germinative capacity soon after detachment from the mother plant. The seeds of the Palmaceae also often retain this capacity but a few months. In reply to a letter, Mr Hart, late Superintendent of the Botanic Gardens in Trinidad, tells me that those of Oreodoxa, Sabal, Thrinax, Acrocomia, Attalea, THE REST-PERIOD OF SEEDS 419 etc., will, if kept in a protected position, maintain their vitality for from three to six months. If, however, there is an excess of atmospheric moisture on the one hand, or drought on the other, the time is much shortened. Some seeds of palms, he adds, like those of Mauritia^ will not keep more than a week or two. Tropical seeds in general, according to Mr Hart, are as a rule possessed of a very fugitive vitality. To keep seeds in stock is, he says, an absurdity, and the only practical rule is to clean partially and sow at once for the best success. It is thus evident that the rest-period must be often a transient feature with the seeds of many plants. Not only is this the case, but there are many plants, as already indicated, where the seeds experience after-ripening, the immature embryo Theafter- of the resting seed continuing to grow up to the time of ger- se?(S"^°^ mination. Such seeds, as Ewart observes, have apparently no rest-period. Amongst examples given by Goebel and others are those of Anemone^ Corydalis^ Crinum, Ranunculus Ficaria^ Eranthis hiemalis^ Gnetum gnemon, Utricularia^ etc. ; but the same may be inferred of numbers of other plants where the embryo is immature or but slightly differentiated, such as Cuscuta, Orobanche, Monotropa^ Balanophorece^ etc., named by Kerner, and Stylidium^ Gagea, and Erythronium^ suggested by Ewart. But in this matter we can make a much wider cast with our net. After-ripening must often be counted upon by the agriculturist and the gardener. They know that certain seeds cannot be forced. It is the experience of the gardener, says Kerner, that many seeds have to mellow or ripen before germination ; and he reminds us that in many cases seeds germinate in the spring under apparently much less favourable conditions of temperature and moisture than they enjoyed in the late summer and autumn of the previous year, when they were first detached from the plant. There is another familiar feature in connection with resting The varying seeds which has been already implied, namely, the great varia- dSpment tion in the degree of development attained by the embryo fn thrrTs't-^° when entering the resting stage. In some the embryo is in ingseed. 420 STUDIES IN SEEDS AND FRUITS different stages of incompleteness. In others it is ready to germinate. No one has stated the matter more clearly than Goebel, and to no one are we indebted for a more authoritative discussion of the subject. It is indeed quite a commonplace feature in seed-life, and I hesitate to add the results of my own observations on a matter long recognised. The data supplied by Goebel in his Organography of Plants (ii. 248-254) supply a most important lead for the investigator. Differences of this kind we are wont to associate with a genus or a family, but here we learn that, as in Anemone and Utricularia^ they may be found within the limits of a species and even in the same individual. We are told that we can only conjecture about the causes of this behaviour ; but the problem restricted to such narrow limits presents an inviting field for the investigator. As will be subsequently pointed out in other plants, the differences in the degree of development of the embryo in these cases is probably associated with the displacement of the period of fruit-maturation as regards the seed. Thus, I would suppose that when the embryo in the resting seed is able to produce its cotyledons and even its first leaves, the fruit is much later in maturing than when the embryo is merely an " unsegmented acotyledonous body." We should then have to inquire into the causes of the postponement or acceleration of the matura- tion of the fruit with reference to the seed and its embryo. But before discussing this view of the matter I will adduce further evidence in support of the contention that embryos, whatever may be their stage of development in the resting seed, are inherently able to continue the growth suspended through the intervention of the rest-period. This is of course implied for many plants in the after-ripening of their seeds and in the occasional premature germination of seeds on the plant during exceptionally humid weather ; but I desire now to show that all seeds, or rather their embryos, possess an inherent capacity of dispensing with the rest-period. That the embryo of an albuminous seed is able to continue its growth THE REST-PERIOD OF SEEDS 421 after removal from the seed was long ago established by Van Tieghem in the cases of Maize and Mirabilis (Nobbe, pp. 310, 31 1). Here, however, the resting seed was concerned ; and it is more to the point to establish it for the embryo removed from the seed before drying and shrinkage begin. I found that embryos of Iris Pseudacorus^ removed from Detached , . , , r 1 • • 1 • 1 • embryos of their bed or albumen m the moist, uncontracted pre-resting ins Pseuda- seed and then placed in water, increased their length in a few '^°'^^" days from 4 to 7 millimetres and displayed the plumular nob. The progressive growth of the embryo as the fruit grows and matures is in this plant very evident. Whilst the seed maintained much the same dimensions (7 millimetres), I obtained the following results for the growth of the embryo in different stages of the fruit's development : — Immature fruit . . . embryo 1*5 mm. long Ripe fruit before dehiscence . „ 2—2*5 •>■> Fruit beginning to dehisce . . „ 3-4 „ With the object of inhibiting the rest-period and inducing Experiments j-^ ,^ -i.-Tij inhibiting pre-resting seeds to proceed at once with germination, 1 placed the rest- at different times a number of seeds of Iris Pseudacorus^ Vicia P®"° * septum^ Arenaria peploides^ and Quercus Robur under favouring conditions in the moist, uncontracted state and obtained success- ful results. Thus, after keeping some of the freshly gathered, ripe, non-dehiscing fruits of the Iris in wet moss under warm conditions (6o°-7o° F.) between ten and fourteen days, I found that the drying of the fruits and seeds had been prevented and that some of the seeds were germinating. The full-grown, soft, uncontracted seeds taken from the green legumes of Vicia sepium behaved in the same way under the same conditions of experiment. In four or five days they commenced to germinate, and in five days the seedlings were half an inch long. The white, soft seeds from the green capsules of Arenaria peploides responded to my experiments precisely in the same fashion. So also with Quercus Robur^ it is not difficult to procure the rapid germination of ripe acorns in September and October. 422 STUDIES IN SEEDS AND FRUITS If fresh green acorns of full size and still vitally connected with the cupule are placed in wet moss under warm conditions, some of them will be found germinating within a week. The germinative capacity of so-called unripe seeds does not seem to have been fully appreciated by foresters, gardeners, and horticulturists, the advantages to be derived from dispens- ing with the rest-period being obvious. As subsequently shown, nature offers us some valuable suggestions in this direction in the case of the Oak. Many plants must afford similar indications. Thus Pfeffer points out (ii. 205) that the seeds of Senecio vulgaris and Stellaria media can germinate as soon as they are ripe. Take again the soft, scantily protected seeds of Pithecolobium filicifolium, the Bastard Tamarind of Jamaica. They germinate a few days after falling from the tree, or else lose their vitality altogether. The causes of the phenomena displayed in the resting of seeds must be sought far back in the plant's life-history, not in the seed alone, but in the seed as it depends on the fruit, and in the fruit as it depends on the parent plant, and in the parent plant as it responds to its conditions of existence. Therefore, in dealing with the causation of the rest-period, we should proceed in this order of investigation : the seed, the fruit, the mother plant, and, lastly, the conditions. Yet it is at first requisite to distinguish between the general causes that determine the suspension of growth and the special influences that determine the stage of development of the embryo at which the rest-period is imposed. Any discussion of the general causes must necessarily begin with an inquiry into the influence of the fruit, and be then extended to the influence of the parent, and then back to the conditions. Being un- prepared to venture into such a wide field of investigation, I will confine my remarks to the influence of the fruit, and that only in an illustrative fashion. The influence The biological disconnection of the seed indicated by the shrivelling of the funicle is proximately determined by the limit of the fruit's vitality. The fruit dries, the funicle shrivels up. THE REST-PERIOD OF SEEDS 423 and the rest-period begins. In other words, the fruit dies and the seed lives, or rather it retains the potentialities of life. In Chapter XIII I have dealt with the different behaviours of the legume and the capsule as regards dehiscence, the first I dehiscing after drying is nearly or quite complete, the second before drying commences or in its earliest stage. As the result of these changes, the seeds shrink and harden rapidly when exposed in the drying, dehiscing capsule, and less rapidly, but not less effectually, in the drying but still closed pod. If it were not for the drying of the fruit there would be no reason why the soft seeds in the moist living fruit should not proceed continuously with their development and dispense with the rest-period altogether. In theory this should be brought about by preventing the drying of the fruit. In practice actual experiment has shown that this can be arranged, as I have done with different fruits of Vicia, Arenaria, Querciis, and Iris, by placing the fruit in moist, warm conditions when in the full-grown living state. Under such circumstances the capsular valves separate, whilst the legume decays rather than dehisces, and the seeds will be found germinating in a week or two without having experienced any pronounced check in the development of the embryo. But these are not the conditions usually presented in nature. The capsule dehisces and dries, whilst the legume dries and dehisces ; and the further development of the embryo is arrested by the resulting shrinking and hardening of the seed. On the assumption that the continuous growth of the embryo-plant is the primal normal condition, there is an obvious lack of co-operation or co-ordination here, since experiment is able to arrange for the working together of the conditions so as to ensure the uninterrupted growth of the embryo. There is a lack of co-ordination in the capsule, because the seeds are exposed in the moist fruit before the embryos can lead an independent existence. There is a lack of co-operation in the legume, because the pod begins to dry before the seeds can sprout. 424 STUDIES IN SEEDS AND FRUITS The rest- The rest-period therefore represents the failure of co- fromthr" ^ operation between the parent, the fruit, and the seed. Over Snadonof ^°^^ ^^^^ ^^^ ^^^'^^ hangs the fate of ultimate detachment the fruit and from the parent, and according as there is concurrence or not the seed. \ . ' • i c r i we get a viviparous or a resting seed. Successrul co-operation ensures not only that before the fruit begins to dry the seeds are ready to germinate, but also that the germinating seed should quickly find suitable conditions for further growth, either by the timely fall of the fruit or by the liberation of the germinating seed. But even here in the great majority of cases germination on the plant takes place in a dying or decaying fruit. Nature is seen at her best in the co-ordination of the growth forces of the fruit and the seed in those plants, like the Mangroves {Rhi'zophora and Bruguierd)^ where the fruit still lives and the seed still grows, until at length the seedling drops from the mother plant. This is the truest form of vivipary. But to return to the question of the lack of co-operation between the seed and the fruit in the legume and capsule, there is not much significance in the mere statement that dehiscence occurs early in the capsule and late in the legume. But there is a good deal of meaning when, viewing the possibilities of vivipary, we state that dehiscence takes place too early in the capsule and too late in the legume. If vivipary took place in the capsule or in a legume, it would be under the moist conditions illustrated in my experiments, where soft, uncontracted pre-resting seeds were induced to germinate without entering the resting stage. But in the capsule nature defeats such an end by bringing about the early dehiscence of the living fruit and the rapid drying of the exposed seeds. In the legume nature would usually render such an event impossible by bringing about the failure in the living but still closed pod of the connection between the parent and the fruit. The pod dries, the funicle shrivels, the seed shrinks and enters the resting stage, and last of all the fruit dehisces. Regarded from the possibility of vivipary, this therefore is the significance THE REST-PERIOD OF SEEDS 425 of the early dehiscence of many capsules and the late dehiscence of the legume. For the continuous growth of the embryo, which we assume to be nature's primal condition, the opening of the fruit is wrongly timed in both cases, too early in the one, too late in the other. Let us take a conspicuous example of the ushering in of The case of • 1-1 CI iri- 1 Poinciana the rest-period m the case or the seeds ot a ligneous legume regia. like that of Poinciana regia. In the chapter on seed-coloration, I have already described the remarkable changes which the embryo of these seeds undergo when entering the resting stage and when resuming active life and growth in the germinating process. The embryo of the large, soft pre- resting seed in the green and living pod is a plantlet full of vitality with green cotyledons and green, partially unfolded plumular leaves. When the fruit begins to dry the green hue of the embryo disappears. The plumule folds up its tiny leaves, and the rest-period commences. The closing in and the cutting off from the external world of the soft green respiring seeds of Poinciana regia seem almost tragical, and one marvels at the fine adjustment which allows these unresistant seeds to hold their own when every- thing is hardening into tough wood around them. Let one of the seeds fail in its early stage and the ligneous tissue invades its area and occupies its place. What concerns us here is that when the embryo loses its green hue, as the tissues harden around the seed, it ceases to respire, and with its vitality completely suspended it becomes buried in a hard woody fruit that dehisces but tardily. In one's ignorance one almost doubts the wisdom of such a suspension of active life in the warm genial climate which this tree enjoys, since the embryo has already advanced considerably in its growth and is well able to proceed with its development. Coming to the special influences that determine the stage The special of development of the embryo when the rest-period begins, determining one may observe that the great variation displayed by the ^jj^ ^^^ °^ embryo in this respect ought to be associated with correspond- period. 426 STUDIES IN SEEDS AND FRUITS ing variations in the period of the fruits' maturation relatively to the seed. If the fruit matures early and dries quickly the embryo will not have reached the same stage of development when the rest-period begins as when the fruit matures and dries late. Where the fruit reaches the limit of its growth far in advance of the seed, we might expect the embryo to be small and but partially differentiated. Where the fruit is not so advanced in growth, the embryo would be much more developed when the suspension of vitality sets in. Where matters are reversed and the embryo grows quicker than the fruit-case, as in Avicennia, the plant is viviparous. Here, however, germination is associated with the rupture and death of the fruit-envelopes. The truest form of vivipary, as already observed, is seen in Rhizophora, where the fruit still lives and the seed still grows, the young plant remaining for a long time attached to the parent. It lies with the future inquirer to ascertain how the mother plant through the fruit determines the stage at which the rest-period is to be imposed on the embryo in the seed. Over both seed and fruit, as previously remarked, hangs the fate of ultimate detachment from the parent ; but this fate may be avoided if the two co-operate so that when the fruit is ripe the seeds have already begun to germinate. The seed depends on the fruit and the fruit on the parent plant ; and since the parent has its part to play in determining the relation of growth between the seed and its fruit, it follows that it has the first word to say in shifting the plane of the rest-period. I may perhaps be allowed to suggest to some investigator that he should inquire into (a) The relation between the stage of development acquired by the embryo in the resting seed and the time of maturation of the fruit ; (b) The relation between the early and late maturation of the fruit (relatively to the seed) and the conditions influencing the mother plant. THE REST-PERIOD OF SEEDS 427 The problem, however, is an extremely complex one. A method of approaching it is indicated in Chapter XIV, where it is shown that two types of fruits can be differentiated when we deal with the proportions of parts for the successive stages of the fruit, as tabulated on p. 303. But this ignores the transition from the albuminous to the exalbuminous state of seeds, which involves a factor of paramount importance, and one that carries us back to a very early state of the seed's development. The whole subject will acquire a very com- plicated character when we introduce this consideration into the discussion. A chance observation in May 1908 led me to suspect that The winter the embryo of the seed of the Ivy {Hedera Helix) grew con- fhe embryo tinuously through the winter and that germination occurred °^*^^n^ without any rest-period in the spring. My suspicion was to plant, some extent confirmed when I found early in June that many of the embryos had nearly doubled their length since the last observation, and that some seeds were germinating within the fruit on the plant. However, the actual growth of the embryo in the winter had yet to be established. To this end my sister, Mrs H. Mortimer, made periodical collections of the berries at Redland, Bristol, during the winter 1908-9 ; and on my return to England from the West Indies in the spring I made use of these materials, the principal data obtained from them being incorporated in the table subjoined. The behaviour of the embryo in the spring was observed by me during the four years, 1 908-11. It will be seen from the results tabulated on p. 428 and The growth from the accompanying figures that the berries increased embryo of gradually in size from the beginning of November 1908 to the Mardf "^*° latter part of January 1 909, the green colour giving place to a blackish hue and the maximum growth corresponding to the complete blackening of the fruit. The increase during this period of the solid constituents of the growing fruit as the water-percentage diminished is especially noteworthy. Up to January also the seeds grew with the fruit, their increase in 428 STUDIES IN SEEDS AND FRUITS Observations on the Growth during the Winter and Spring of THE Fruits, Seeds, and Embryo of the Ivy (Hedera Helix). (Weights in grains and lengths in millimetres.) Moist fruit. Air-dried seed. Embryo of moist seed. Loss of weight Pro- of the Remarks. Length portion Colour. Weight. Weight. after soaking. Length. of bulk of moist seed. Colour. dried fruits. 1908. Nov. 9 Green i-69grs. •06 gr. 4-5 mm. 7 mm. 1% White 79% „ 18 ,, 2-86 „ *I2 ). S"o » i'5 .. I '5% ,, 79% Dec. 3 ,, 4*02 ,, ■20 ,, S"5 .. 1-6 ,, i-a ,, 76% ., 17 Blacken- ing 4*96 ,. •24 ., 6-0 „ 2*0 ,, ^■S7o " Jan. 9 Black 5 '44 ,, •34 >, 6-0 „ 2'2 ,, 3-o% „ 65% ., 24 ,, 5"5o .. "35 „ 7'° » 27 » 4'o% ,, 63% Feb. 21 ,, 4-10 ,, •36 „ 67 „ 2-5 ,< ra ,, 60% Mar. 19 " 4-02 ,, •34 „ 6-3 » 2-3 .. 3-5% " 6i% Some seeds 1909. germinat- Apr. 20 4-5 M 10"/; Green (57%) ing on the May 6 ... 6-0 „ 20% " plant on May 6. 1910. Mar. 11 2-0 ,, White On May 22 Apr. 2 2-5 .. ,, some seeds „ 8 3"o >' germinat- May 3 40 „ ,, ing on the ,, 22 S'o ., Green plant. Explattation. — The observations of November 1908 to March 1909 were made from collections obtained from the same plant by Mrs Mortimer at Redland, Bristol. Each sample, consisting of from forty to sixty berries, was weighed at once, and subsequently weighed and examined by me when the drying was complete some months after. The other observations were made by me from fruits gathered at Salcombe in South Devon. In the case of the Bristol fruits the data for the embryos were obtained after soaking the seeds in water for a day. The water-contents indicated by the loss of weight of the berries when air-dried are taken from the table at the end of Chapter XII. The entry for April 20, 1909, relates to an observation made on fruits in the same stage in the following year. The diminution in the weight of the fruit in February and March is due to most of the larger fruits having fallen. THE REST-PERIOD OF SEEDS 429 Figures illustrating the Growth on the Plant of the Seed AND Embryo of the Ivy (Hedera Helix) from November 1908 TO June 1909. Seed, 4 '5 mm. Embryo, 07 ,, December 3. Seed, 5 '5 mm. Embryo, 16 ,, January 9 Seed, 6 mm. Embryo, 2*2 mm. January 24. Seed, 7 mm. Embryo, 275 mm. February 21 Seed, 67 mm. Embryo, 2-5 ,, March 19. Seed, 63 mm. Embryo, 2*3 ,, May 16. April 20. Seed, 6 '3 mm. Embryo, 4 mm. Seed, 6 mm. Embryo, 6 ,, Early in June. Germinating seeds. All enlarged ; in the case of the seeds twice the natural length, and in the case of the separate embryos four times, but the germinating seeds are drawn about one and a half times the normal length. With the exception of the germinating seeds, therefore, all the seeds are drawn to one scale, and since the separate embryos are also drawn to one scale, it follows that the changes in size are respectively in true proportion. These figures are intended to illustrate the results given in the preceding table. 430 STUDIES IN SEEDS AND FRUITS size being principally due to the growth of the endosperm. The small embryos also added to their bulk, but at a rate not much faster than the seed. They displayed but little evidence of having grown at the expense of the food-reserve, and remained white, with their cotyledons appressed. From the end of January up to the middle of March there was a sus- pension of the growth of the fruit, seed, and embryo ; and here the observations for that season ended, the embryo remaining in the same colourless, inert condition. The winter was severe, and I think it very probable that under milder conditions there would not have been this check. The latter part of the history of the growth of the embryo in the seed on the plant is supplied by my own observations during four successive springs. Since their general results agree, I have only given in the table those for 1909 and 19 10. It is there indicated that after March the embryo grows with fair rapidity. Taking all the data of the four years, the average growth in the spring would be as follows : — By the end of March or the beginning of April the embryo would be about 3 millimetres long, or just half the length of the kernel, but remaining white and showing no enlargement of the cotyledons. During the latter part of April and the early part of May most of the berries fall to the ground, their detachment being hastened by wind and rain. The seeds of those that remain usually display embryos 4 or 5 millimetres long, increasing perceptibly at the expense of the albumen and with enlarged green cotyledons. As May advances the embryos attain the length of the kernel (6 to 7 millimetres), some of them becoming even longer, so that they are compelled to accommo- date themselves to the kernel's length by bending, as shown in the figure for May 16. A few seeds will be found germinating within the fruit, which has already begun to shrivel and soon drops off. Whilst the berry is attached to the plant the radicle pierces the seed-coats, but not the pericarp, the hypocotyl becoming bent over the seed inside the fruit. In these germinating seeds the albumen has largely disappeared. THE REST-PERIOD OF SEEDS 431 In different years, when the season is unusually mild, all the stages above referred to the month of May will be found in April. The seeds will not generally be found all germinat- ing in the same berry, but all will show embryos advanced in growth. Nature only offers a small number of fruits to illustrate the germination of the seeds on the plant, since the final stage is usually anticipated by the early shrivelling and detachment of the berry when the embryo is 3 or 4 millimetres long, the result of late frosts, wind, and rain. However, of the surviving berries the proportion with germinating seeds in May will vary from 10 to 50 per cent. By the beginning of June all the fruits have fallen. I have here been describing the behaviour of Ivy berries in the mild climate of South Devon. The growth of the embryo in the spring is a good deal influenced by the situation of the plant. Thus in sunny places sheltered from the cold winds it will be much in advance of that found in plants growing in bleak, exposed localities. The nature of the growth of the embryo of the Ivy during Two modes the winter is well brought out in the table. The seed grows SJe^embrJo"^ with the berry and the embryo grows with the seed, the thg^glJ^on" increase in its proportional bulk being but slight. There is the plant, little or no growth at the expense of the endosperm, the embryo remaining white and the cotyledons retaining their small dimensions. But with spring in progress the berry and the seed no longer add to their size. The embryo, now 4 or 5 millimetres long, grows independently. Its cotyledons enlarge and its whole surface becomes green, this independent growth being associated with a gradual diminution of the food-reserve, so that when the seed is found germinating in the berry on the plant, the albumen has mostly disappeared. The two kinds of growth of the embryo, first during the winter with the seed and the fruit, and then in the spring at the expense of the albumen, are the conspicuous features in the vivipary of the seeds of the Ivy. We can thus distinguish two stages in the after-ripening of these seeds. 432 STUDIES IN SEEDS AND FRUITS There are therefore two singular features that must be closely linked together in the life-history of the Ivy — the autumnal flowering and the ripening of the seeds on the plant during the winter, followed by germination in the spring, without the intervention of a rest-period of more than a few weeks' duration. Much depends on the causal connection between them. If the plant is viviparous because it flowers in the autumn, then the vivipary appears to be adaptive ; but if it flowers late because it is viviparous, then the autumnal flowering would be an adaptation. It may be that the cause of the late flowering is to be found in the absence of any proper rest-period for the seed. For if the plant flowered in the spring, it would mature its fruit at the close of the summer, and the seedlings would be cut off by the winter's cold. The retention of the viviparous habit would lead to extinction, unless flowering occurred in the autumn. It is likely that a diff^erence in the mode of ripening of the seeds may explain why Hedera Helix (as stated by Kerner) grows in Central Europe without any protection from the winter's cold, whilst the Ivy of Southern Europe [Hedera poetarum\ which is very similar in characters, can only survive the winters of Central Europe under a protecting roof. I come now to my observations on the normal tendency to vivipary displayed by the seeds of acorns {Quercus Robur). This is not only exhibited in the occasional germination of these fruits on the tree, but in the actual stages of growth of the seed within its shell before maturity is reached. The steady growth of the seed on the tree long after the pericarp or shell has begun to dry has been discussed at length in Chapter XIV. It was then said that the tendency of a seed to continue its growth on the plant after the pericarp or fruit-case has commenced to dry and lose weight, finds its final expression in the germination of the seed on the plant, or, in other words, in vivipary. Such was the tendency displayed by the oaks near Salcombe, in South Devon, during the successive autumns of 1908 to 191 1 ; and doubtless it is characteristic of this tree in other localities. THE REST-PERIOD OF SEEDS 433 It can be easily demonstrated that ripe moist acorns are able to proceed at once with germination if placed under conditions inhibiting the drying of the fruit. Thus on September 17, 1908, I collected some ripe acorns and placed them at once in damp moss in a warm cupboard. They were still biologically connected with the cupules, and their shells, though beginning to brown, were still thick and moist. Within eight days I found some of them germinating normally, and one of them when planted grew healthily under protection during the winter. (Whilst preparing this chapter (September 191 1) I repeated this experiment with green ripe acorns show- ing no signs of drying, and possessing, as in the first case, entire shells. In five days half of them were splitting their shells, and several of these were protruding the radicle.) Every autumn I noticed a small but variable number of ripe acorns showing signs of germination on the trees in the splitting of their shells and in the slight protrusion of the radicle. The growing seed had burst the fruit-case, and in many cases it was evident that the seed was larger than its shell. This was recorded at the end of September 1908, in the first half of October 1909, in the middle of October 19 10, and in the second week of September 191 1. A number of the split nuts placed at once in wet moss in two different autumns were found in four or five days well advanced in the germinating process. When the acorn begins to split at its sides it is full-sized, moist, and green, and is still vitally con- nected with the cupule. Usually the protrusion of the radicle is not great on the tree ; but I can recall a case where its growth was considerable, and where the inner surfaces of the cotyledons were turning green whilst the fruit was still attached. This attempt at germination on the tree soon brings about The fate of the fall of the nut. The shell browns rapidly as it dries, and that^germin- the fruit is soon vitally disconnected from the cupule. Generally Jlj^^" *^^ the fallen acorn dies ; but it must frequently happen in moist mild weather that it continues the growth commenced on the 434 STUDIES IN SEEDS AND FRUITS tree ; and if it is subsequently protected by the fallen leaves, there is no reason why it should not survive an ordinary winter's weather and be ready for active growth in the spring. Here is an experience that bears directly on this point. Some acorns which had germinated a few days after being picked from the tree in October were left covered up in the basin of wet moss in my greenhouse and forgotten. Early in March I was surprised to find that several of the acorns were still alive, with radicles protruding about an inch, the moss being still damp. The winter growth had been very slight, the germinating seeds having undergone a period of almost complete repose. Healthy seedlings were raised from them. During the months of January and February there had been no artificial heat in the greenhouse, and the contents were at times frost-bound, the lowest reading of the thermometer inside the house having been 23° F. The proportion of acorns exhibiting the early stage of The pro- germination on the tree in the autumn varied in different years. portion of ^j^^g -j^ ^^g season of 1909 I placed it at 2 per cent. In 3.corris tn&t _ i. j * begin to 19 1 1, after the abnormally long and dry summer, extendmg germmateon ^^^^ '^^^ ^^^^ ^^ ^^^ ^^^^^ ^^^ proportion in the fourth week of September was as high as 10 per cent. The exceptionally dry season had not affected the foliage of the trees, whilst the fruits were much larger and much more abundant than m the previous years. Whilst in 1908 the average weight of an acorn was from 60 to 70 grains, and in 19 10 from 50 to 60, in 191 1 it was about 100 grains. The fruits ripened nearly a month earlier than usual, being mature in the middle of September instead of early in October. I had a remarkable experience with ripe acorns during Thegermin- three successive Octobers. After being gathered from the detached tree in the green moist stage, before any loss of weight by acorns drying had occurred, I placed them on each occasion in a d?^nr^' dry saucer in a room which was rather damp, as there was ^°^"'- no artificial heat. The nuts were quite entire and showed no signs of splitting their shells. On the first occasion THE REST-PERIOD OF SEEDS 435 eleven out of fifty-five nuts were found in the typical germinating state about seventeen days after. The fruits had germinated whilst actually drying ; and I estimated their ■ loss of weight when germination began at from 10 to 20 per cent., the total loss of weight of a mature fruit when completely air-dried being from 50 to 60 per cent. The same thing happened under the same circumstances in the two following Octobers. In an experiment carried out whilst writing this chapter, I placed thirteen ripe acorns in a dry cup. In nine days one was protruding its radicle and two were splitting their shells, the aggregate loss of weight since the seeds were gathered in the moist entire condition from the tree being just 10 per cent. We are not concerned here with after-ripening, such as frequently occurs with detached albuminous seeds, since in the mature acorn on the tree the embryo is fully formed and develops a plumule which may be turning green. The interest lies in the fact that germination took place in the case of a mature exalbuminous seed in a drying fruit. It is evident that with the acorn the ripe moist seed is not necessarily prevented from proceeding continuously with its growth by drying in air for a week or two. But the drying process must be slow. In most cases the loss of water is too rapid and the tendency to proceed at once to germinate is suppressed. Evidently the detached ripe acorn, provided that the drying is checked, can make good use of the water it holds, which is more than is actually needed for the continued growth and germination of its seed. Anything that impedes the air-drying process of the freshly detached acorn will assist the seed in its endeavour to dispense with the rest-period altogether. In this way we may explain Uloth's observation of acorns germinating in ice. The explanation given in Nobbe's book Uloth's (p. 104) is, that the requisite water would be supplied by the of acorns melting of the surrounding ice through the natural warmth germinating o to to in ice. .of the seed. However, we learn from my observations above Vivipary in Artocarpus incisa. Crinum. 436 STUDIES IN SEEDS AND FRUITS given that the detached fresh acorn does not need any water from outside in order to proceed at once with the germinating process. In fact, we have seen that such acorns will germinate without a rest-period after losing a good proportion of their weight by drying. Any check to the drying process of the fresh detached nut would directly aid the seed in proceeding continuously with its growth and in dispensing with the usual period of repose. This check would be found in the inclusion of acorns in ice. Drying would thus be inhibited, and the acorn would have sufficient free water within its own substance for the uninterrupted growth of the seed. For important particulars relating to the stages of growth of the acorn the reader is referred to two tables given on PP- 303. 311- Germination within the fruit on the tree may take place in the case of the seeded variety of the Bread Fruit {Artocarpus incisa). This came under my notice in Grenada and Tobago. Mr Anstead, the Superintendent of the Botanic Gardens in Grenada, assured me that this habit was well known in the island. In the case of a fruit that had just fallen I found a third of the seeds (about sixty in all) germinating or " showing eyes," as the coloured people call it. In another fruit that had fallen the day before three-fourths of the seeds were germinating. It was evident that germination began on the tree in the ripe fruit ready to fall. With the ordinary seedless variety suckers burst through the ground all around the tree. They are absent altogether with the seeded kind. I made some observations on the viviparous habit of detached Crinum seeds. This habit is well known, so that I will refer the reader to Note 29 of the Appendix for my remarks on the subject. Here, in the case of seeds packed away in my collections, germination took place in seeds that had lost more than half of their weight, the embryo increasing its size ten- or fifteen-fold within the seed during the drying process. THE REST-PERIOD OF SEEDS 437 Some of the works quoted in this chapter are — Goebel's Organography of Plants ; Kerner's Natural History of Plants ; Ewart's "Longevity of Seeds" {Proc. Roy. Soc. Victoria^ 1908) ; Ewart in Trans. Biol. Soc. Liverpool^ 1896 ; Nobbe's Handbuch der Samenkunde ; Schroder in Untersuch aus dem Botan. Inst, zu Tubingen ^^Vind ii., 1886. SUMMARY (i) The rest-period represents a break in the continuity of the young plant's life, and is the effect of external conditions acting through the parent and the fruit on the seed. The author's main object in this chapter is to establish the inherent capacity of all embryos to proceed uninterruptedly with their growth, whatever their stage of development when the resting state is imposed. (2) In dealing with this subject he discusses the prevalence of the resting state in seeds, and the well-icnovv^n circumstance just implied that the rest-period has been imposed on seeds in all stages of develop- ment of the embryo. (3) In the first case, he shows that this state of repose is often very transient with a large number of plants, and that when we reflect that many plants experience " after-ripening " in their seeds, when the embryo continues to grow after the seed has entered the resting state, the rest-period is deprived of some of its prominence as a feature in plant-life (p. 418). (4) As concerning the second point, he directs attention to the great importance of Goebel's observations on Anemone and Utricularia^ from which we learn that all the stages, from that of the unsegmented acotyledonous embryo to that of the embryo producing its cotyledons and even its first leaves, may be found not only in the resting seeds of different individuals of the same species, but even in the same individual. It is suggested that the problem thus confined within such narrow limits presents an inviting field for further inquiry (p. 420). (5) The author points out that the inherent ability of the embryo to continue its growth without the interruption of the rest-period is itself implied in its existence in such varied stages of development in the resting seed. The familiar after-ripening of seeds, above noticed, and the well-known occasional germination on the plant of seeds accustomed to submit to a normal rest-period, are also facts indicative of this inherent capacity (p, 420). (6) Experimental proof is adduced to show that in the case of any plant taken at random, such as Arenaria^ Iris^ Ficia^ Quercus, etc., it is 43^ STUDIES IN SEEDS AND FRUITS possible, by placing under suitable conditions the soft uncontracted seed of the moist living fruit, to induce it to proceed continuously with its growth and to germinate without any resting stage (p. 421). (7) Coming to the question of causation, it is urged that we must distinguish between the general causes of the rest-period and the special influences that determine the stage of development of the embryo at which the period of suspended growth is imposed (p. 422). (8) In references to the general causes it is shown that although in some ways certain climatic influences may be recognised, as those concerned with excessive humidity, yet the subject is really far more complicated than such an explanation would suggest. Indeed, the causes must be sought far back in the plant's life history, not in the seed alone, but in the seed as it depends on the fruit, and in the fruit in its dependence on the mother plant, and in the mother plant in its responses to its conditions of existence (p. 422). (9) Being unprepared to undertake such a profound inquiry, the author here limits himself to the influence of the fruit, and that only in an illustrative way. He shows that the suspension of the active growth of the seed presents itself as the result of failure in the co-ordination or co-operation of the growth of the seed and the fruit. Only in the truly viviparous plant, as in Rhi-zophora^ is there complete co-ordination. Over both seed and fruit hangs the fate of ultimate detachment from the parent j but this fate may be avoided if the two co-operate, so that when the fruit is ripe the seed has already begun to germinate (p. 423). (10) Taking the capsule and the legume, it is remarked that there is a lack of co-ordination in the first because the seeds are exposed in the moist fruit before the embryos can lead an independent existence ; and there is a lack of co-operation in the legume because the pod begins to dry before the seeds can sprout. There is not much significance in the mere statement that dehiscence takes place early in the capsule and late in the legume. But there is a good deal of meaning when, regarding the possibilities of vivipary, we state that dehiscence occurs too early in the capsule and too late in the legume. Nature has wrongly timed the opening of the fruit in both cases (p. 423). (11) As showing the way in which the rest-period may be imposed, and how the life of a young plant well able to proceed with its growth may be abruptly suspended, the case of the seeds in the woody legume of Poinciana regia is taken (p. 425). (12) With regard to the special influences that determine the stage of development of the embryo at which the rest-period is imposed, the author looks for them in the different stages of growth of the fruit and the seed. Since, however, the rest-period is directly determined by the limit of the fruit's growth, and since the limit of the fruit's growth is determined by the mother plant, it follows that the mother plant has THE REST-PERIOD OF SEEDS 439 the first word to say in shifting the plane of the rest-period. But re- stricting his remarks to the special influences of the fruit, the author points out that the more the fruit's growth is in advance of that of the embryo, the earlier should be the onset of the rest-period, and that only when the two are co-ordinated does true vivipary occur (p. 425). (13) Then follow the results of the writer's observations on the viviparous tendency displayed by the seeds of the Ivy [Hedera Helix), and of the Oak [Quercus Robur). In the case of the Ivy it was ascertained that the ripening of the seed on the plant during the winter is followed by germination in the spring without the inter- vention of a rest-period of any long duration. The ripening in the winter months is characterised by the associated growth of berry, seed, and embryo, the embryo growing as the endosperm increases. The ripening in the spring is confined only to the embryo, which grows at the expense of the endosperm. The result is germination on the plant in the case of the seeds of berries that remain long attached to the parent (p. 427). (14) In the case of the Oak it is established not only that the freshly detached moist acorns can be readily induced to pass on to o-ermination, but that there is a decided tendency for the ripe acorn on the tree to dispense with the rest-period. This viviparous habit has been already regarded as the final expression of the tendency of a seed to continue its growth on the plant after the fruit-case has commenced to dry, a capacity that was established for the seed of the acorn in Chapter XIV. As a result the seed not infrequently becomes too large for the fruit-case and splits the shell, the radicle protruding when the" acorn remains some time longer on the tree. The proportion of acorns beginning to germinate on the tree varied between 2 and 10 per cent. (p. 432). (15) The chapter is concluded with some notes on the germination of the seeds of Artocarpus incisa (seeded variety of the Bread-fruit tree) in the fruit on the tree, and on the continuous growth of the embryo of Crinum during the drying of the seed (p. 436). CHAPTER XX THE COSMIC ADAPTATION OF THE SEED Intro- The physics of seeds ought to be a subject of deep interest, if ^^ °^^' only from the circumstance that whilst the seeds, generally speaking, can live or retain their vitality in any climate, the parent plant is as a rule rigidly restricted in this respect. The fact that the seed is less specialised for terrestrial conditions The seed is than the parent plant is one of the first suggestions that nature ised and less offers to US when we approach the consideration of seeds from than^the"^^ the cosmic standpoint. It is one of the purposes of this plant. chapter to extend this distinction by showing that seeds might live on a planet where conditions destructive for the parent plant prevail. Where the discussion appears disconnected and inconsistent, the defect is usually due to the circumstance that I have here strung together notes and ideas jotted down generally during botanical rambles in the last four years. To endeavour to adjust some of them would be to displace others, so I have preferred to let them stand, feeling assured that in the opening up of new ground of this sort the reader will be to my faults a little blind. It was the behaviour of the seed of Guilandina bonducella in the oven and in the balance that first led me into these specula- tions. The spectacle of a plant-embryo living its own life in its hermetically sealed case and irresponsive to outside con- ditions seemed to offer a near approach to an unconditioned existence on this planet. Though crude and only partly true, this notion proved to be very suggestive ; and I came to see 440 THE COSMIC ADAPTATION OF THE SEED 441 that the transition from the seed to the full-grown plant is a transition from the less conditioned to the more conditioned state of plant-life. The seed often appears to be largely independent of its conditions on the earth's surface ; but this could only be true in a relative sense as compared with the plant. The plant is conditioned only for terrestrial existence, the seed for existence in the cosmos. This it is that makes the seed so often seem to be out of touch with terrestrial conditions. Whilst there is so much about the seed and its fate that appears to be haphazard and to be determined by accidents when regarded from the standpoint of our planet, it may be, as before observed, that regarded from the broader standpoint of the cosmos such a lack of harmony does not exist. The seed in the universe or cosmos may be like a great traveller on the earth adapted to all climes and acquainted with all peoples. He is cosmopolitan in his habits, and as such seems fitted for all conditions. Yet if we were to ask the peoples of the different countries amongst whom he had lived, we should find that they judged him merely from the restricted standpoint of their experience, and that only in proportion as he acquired proficiency in their special way of living would he be regarded as a profitable member of their society. His general fitness would not be appreciated by a North American Indian if he could not follow a trail, or by a Pacific islander if, when stranded on a coco-nut islet, he could not climb the tree. Thouo-h in a general sense fitted to live everywhere, in a special sense he would be suited for nowhere. Yet the judgment of the savage would be the pity born of ignorance. So it is with ourselves and the seed. We only notice its itspotenti- want of special fitness for its terrestrial life. Whether it will senrus'with reach a suitable soil or whether it will ever germinate at all fJ^^^^