g; CO^RIJELL UNIVERSITY -LIBRARY. '& 
 I \.- — — — - I 
 
 M This book %iibt to be taken ^ 
 
 M from the S-e^dinc Hoom. i 
 
 y^yi^ %AfLJirfki .rv^^Kir* lA(l-n_i r-i t-i-i > n ».■ AT /*SHM/^r- L ^^ 
 
 ^ WHEN dONE WITH, RETURN AT CJNCE TO 
 
 m 
 
 •0 / 
 
 SHELF. X4if 
 
 -N.. 
 
 
Pi I 
 
 l?77 
 
Cornell University Library 
 PA 251.P21 1877 
 
 A manual ot comparative philoloffl^^^^ 
 
 3 1924 021 600 378 
 
 OLIN LIBRARY -Clk^w.^.. ON 
 DATE DUE 
 
 iHMBTnt i: 'flx. itti 
 
 « 
 
 
 — IPMBUr 
 
 
 1 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 CAYLORD 
 
 
 m^^^i^^-'-t '■ ':^i'-.*;^*ea 
 
 l»ftlNTCDINU.S.A 
 
B Cornell University 
 
 M Library 
 
 The original of tliis book is in 
 tine Cornell University Library. 
 
 There are no known copyright restrictions in 
 the United States on the use of the text. 
 
 http://www.archive.org/details/cu31924021600378 
 
€Ia«ttJi0K |nss Bnm 
 
 A MANUAL 
 
 OF 
 
 COMPAEATIVE PHILOLOGY 
 
 PAPILLON 
 
Eontion 
 MACMILLAN AND CO. 
 
 PUBLISHERS TO THE UNIVERSITY OF 
 
 ®xfor&. 
 
^hx&nlian ^r^ss S«ms 
 
 A MANUAL 
 
 OF 
 
 COMPARATIVE PHILOLOGY 
 
 AB APPLIED TO THE ILLDBTBATION OF 
 
 GREEK AND LATIN INFLECTIONS 
 
 T;'LrPAPILLON, M.A. 
 
 FeUoto and Lecturer of New College, Oi^ord 
 formerly Scholar of Balliol, and Fellow of Merlon 
 
 SECOND EDITION, BEVISED AND CORRECTED 
 
 AT THE CLARENDON PRESS 
 
 M DOOO Lxxvn 
 
 (=" 
 
 [AU rights reserved'] 
 
PEEFACE. 
 
 This book contains the substance of lectures delivered at 
 Oxford in 1874 and 1875 to candidates for Honours in Classics 
 at the First Public Examination, upon the prescribed subject 
 of 'the Elements of Comparative Philology as applied to the 
 illustration of Greek and Latin Inflections,' and therefore does 
 not profess to deal with more than a very limited portion of 
 the wide field covered by the Science of Language. In the 
 course of my lectures I was often met by the difficulty of 
 recommending to pupils any English text-book that would give 
 them in a compact and accessible form the means of preparing 
 the subject for themselves ^; and I had hoped that the want thus 
 felt might have been supplied for our students by the Oxford 
 Professor of Comparative Philology. So long as there was any 
 prospect of help from that quarter, it would have been pre- 
 sumptuous in me to come forward. But Professor Max Miiller 
 was unable to spare time from more important labours; and 
 the preparation of this Manual was perforce entrusted to very 
 inferior hands. 
 
 As to the educational value of the limited area of philolo- 
 gical study which is here surveyed, I am aware that opinions 
 
 ' Since this wag written, Mr. Peile's ' Primer of Philology' (Maemillan) 
 has provided beginners with an admirable little introduction to the study 
 of Comparative Philology. 
 
vi Preface. 
 
 differ. It is said, for example, that a study of the forms of 
 Greek and Latin words is a mere effort of memory, testing 
 neither the reflective powers nor the application of deductive 
 method, and therefore not worth introducing as a separate 
 subject into the curriculum of study for Classical Examinations 
 at Oxford. It might, I think, be asserted with equal truth 
 that, inasmuch as some acquaintance with the history of Greek 
 and Latin Inflections as read in the light of Comparative Phi- 
 lology is essential to an iutelligent study of the grammar of 
 those languages, some knowledge of the elements of that science 
 might fairly be exacted from all candidates for classical honours. 
 I am not however concerned with a defence of the study of 
 Comparative Philology. Its bearings upon some of the most 
 interesting problems of ethnology, of history, and of religion 
 are too well known to require assertion ; and if a mere know- 
 ledge of the forms of two or more languages seems to carry 
 ihe student but a little way towards these higher regions of the 
 scienpe, it must be remembered that such elementary know- 
 ledge, small at first and gradually and carefully extended, is 
 the only sure foundation for more advanced research, with- 
 out which all enquiry into higher problems may lose itself in 
 a wilderness of conjecture. I believe that a minute study 
 and comparison of the forms of two such languages as classical 
 Greek and Latin, or of two or more among the languages of 
 modem Europe, with a due comprehension of the laws of 
 phonetic change that have operated to produce existing 
 divergences from common forms, is the best possible pre- 
 paration for an adequate grasp of any of the higher problems 
 into which the science of language enters. It has an interest 
 too in itself; the interest of tracing in different languages 
 the divergence, under regular processes of phonetic change, 
 of words and forms common to them all; the interest of 
 
- Preface. vii 
 
 detecting meaning and force in much that appears at first 
 sight arbitrary and unmeaning ; above all, the interest of 
 watching the life of a language and its perpetual growth and 
 change in the mouths of those who speak or have spoken it. 
 No apology, I think, is needed for any attempt, to lay in the 
 minds of boys or young men the foundation, however limited, 
 of such a study. 
 
 The arrangement which I have adopted is that which has 
 been found most convenient for teaching purposes. In its main 
 outlines it is the same as that adopted by Schleicher in his 
 ' Compendium der Vergleichende Grammatik,' and by teachers 
 in the schools of Germany, if I may judge from a useful little 
 summary entitled ' Sprachwissenschaftliche Einleitung in das 
 Griechische und Lateinisohe, fur obere Gymnasialclassen,' by 
 Professor Baur of Maulbronn^ It will be observed that the 
 names referred to below as of leading, authority are (unless 
 England has by this time established a claim to Professor 
 Max Mtiller) almost exclusively German ; and it -is not too 
 much to say that at present Comparative Philology cannot 
 be thoroughly studied without at least a moderate acquaint- 
 ance with the German language. But the best German 
 philological works (to say nothing of their size and cost) are 
 often, from the very exhaustiveness of their treatment, only 
 confusing to beginners, who require a smaller array of facts 
 more simply and clearly arranged. And valuable as are the 
 translations into English of such works as Bopp's ' Comparative 
 Grammar,' Curtius' 'Principles of Etymology,' or Schleicher's 
 'Compendium,' to the advanced student or teacher, they are 
 both in quantity and quality above the requirements of the 
 schoolboy or the undergraduate during the first period of his 
 
 » Now accessible to English readers in a translation by Messrs. C. Kegan 
 Paul and E. D. Stone (H. S. King and Co., 1876). 
 
viii Preface. 
 
 University life : to serve whom is the less amhitious, but I trust 
 not less useful aim, of the present work. 
 
 This (second) edition exhibits several modifications or altera- 
 tions of views expressed in the first edition, which are due 
 partly to my own further study, partly to the suggestions of 
 others. The account of the Greek alphabet, for example, has 
 been re-written, and I trust improved : the remarks upon the 
 physical conditions of the production of sounds (pp. 29, 30) have 
 been made clearer (I hope) than they were : and some altera- 
 tions have been made in the discussion of the 'three stages' 
 of language in chap. ii. In chap. viii. the explanation formerly 
 given of the so-called ' connecting vowel ' (o in \iyofiiv, i in 
 ferimus) has been abandoned, and the term 'thematic vowel' 
 adopted, as expressing . more nearly the result of the most 
 recent investigations ; a fuller, and in some respects diflPerent, 
 account is given of the terminations -a6a (3 sing.), -a-Be, -a6ov, etc. ; 
 and some details of verb-inflection are more fuUy if not always 
 differently treated. Of these latter changes many are due to 
 a study of the now completed work of Curtius, ' Das Verbum 
 der Griechisohen Sprache,' the omission of all reference to which 
 in my first edition may have appeared strange. I did not 
 refer to it because it was then unfinished (the second and 
 larger volume not having appeared), and I was unwilling to 
 seem too eager to assume the attitude so often characteristic 
 of English scholarship, viz. catching at and reproducing the 
 latest views of the latest German writer. The completion, 
 however, of Curtius' really great work makes such omission 
 now inexcusable ; and I gladly acknowledge obligations to it. 
 
 I have still to admit, and to claim indulgence for, an imperfect 
 knowledge of the Sanskrit forms necessary for the illustration 
 of corresponding forms in Greek and Latin. But this edition 
 has, what the first had not, the advantage of revision (so far as 
 
Preface. vs. 
 
 the Sanskrit forms are concerned) by Professor Max Miiller, 
 wtose great authority will command a confidence that other- 
 wise was beyond my reach. To secure a uniform system of 
 transliteration from Sanskrit to Eoman character, I have 
 given below a Table of the Devan4garl letters with their 
 equivalents as employed in this book. Of the two alternative 
 modes of representing the ' palatal ' and ' cerebral ' mutes given 
 in Professor Max Muller's own table, I have at his advice adopted 
 that which represents them by the 'guttural' and 'dental' 
 characters respectively in a different type, e. g. k, kh. (guttural), 
 h, Jch (palatal) ; t, th (dental), t, th (cerebral) ; the advantage of 
 which method is pointed out on p. 33, note r. For the palatal 
 spirant % answering to English y (initial), and to consonantal 
 or semivowel i in Greek or Latin (Greek i, Latin i or j) I now 
 employ y, enclosing an i orj in brackets where Greek or Latin 
 words are in question, e. g. on pp. 200, Z05. It may no doubt 
 be fairly argued that j, now employed by many philologists 
 to denote V, is scientifically preferable to y, for the aid which it 
 gives to the immediate perception of etymological connections : 
 but I doubt if there is as yet sufficient familiarity, among those 
 for whom this book is intended, with the correct pronunciation 
 of Latin / (i semivowel) to justify me in abandoning the familiar 
 y as an expression of the ' y sound.' At any rate I hope that 
 by never using j to denote »T (i. e. the sound of j in judge), 
 I have avoided one source of confusion, and made it clear that 
 in English words only is j to have its English value. In Latin 
 words I retain j for the consonantal (semivowel) sound of i, v 
 for that of u : and though purists in Latin orthography will 
 perhaps object to any employment of the non-classical cha- 
 racters _;, V, the practical convenience of using distinct characters 
 for distinct sounds may be pleaded in excuse. In column V of 
 the table on p. 42 will be found stated the probable pronunciation 
 
X Preface. 
 
 of the letters of the E/Omau alphabet ,* and where this dif- 
 fers from the English pronunciation of the same letters (as in 
 the vowels a, e, i, U, the consonants c, g before e, i and the 
 semivowels j, v), the Eoman, not the English, pronunciation 
 is presumed in all Latin words cited. With this caution, I 
 hope that any confusion as to the relation between the letters 
 in question and the sounds represented by them, into which 
 English usage might lead us, may be avoided. 
 
 For corrections and hints towards the preparation of this 
 edition I have to thank Mr. D. B. Monro, Vice-Provost of 
 Oriel, and Mr. Henry Nettleship, Fellow and Tutor of Corpus 
 Christi College, Oxford. I have also profited by several an- 
 onymous reviews of the first edition, and particularly by a 
 courteous and suggestive criticism in the Academy of May 
 27, 1876, signed by Professor Wilkins of Owens College, Man- 
 chester. To one or two private correspondents (notably to 
 Mr. Q. R. Merry of the Edinburgh Academy) I am indebted 
 for valuable suggestions. To the Delegates of the Clarendon 
 Press I need only repeat my thanks for care taken and courtesy 
 shown in all arrangements for publication. 
 
 T. L. P. 
 
 New College, Oxfobd, 1877. 
 
ct. tl 
 
 o H 
 
 i 
 
 1 
 
 ^ 
 
 ^ 
 
 
 
 
 
 fl 
 
 d 
 
 o 
 
 ol 
 
 1 
 
 S 
 
 s 
 
 ■^ 
 
 1^ 
 
 •B 
 
 fl 
 
 1 
 
 tI 
 
 01 
 
 CJ 
 
 " 
 
 o 
 
 
 pf 
 
 f- 
 
 
 
 
 
 
 
 
 
 
 
 JS" 
 
 o 
 
 
 
 
 
 
 
 
 
 4 
 
 
 ■s 
 
 1 
 
 
 
 1 
 
 
 
 
 
 
 (i 
 
 
 ,. 
 
 • > 
 
 
 
 a 
 
 
 
 
 
 
 
 
 
 
 
 
 
 <tS 
 
 <« 
 
 '^. 
 
 <_,_ 
 
 <3 
 
 
 ^ 
 
 I—" 
 
 rf-" 
 
 J* 
 
 > 
 
 <S 
 
 ^ 
 
 ^ 
 
 fi? 
 
 •w 
 
 r 
 
 1^ 
 
 lb 
 
 t2 
 
 1 
 05 
 
 IS 
 
 ll-" 
 
 •a. 
 
 s 
 
 le 
 
 3 
 
 
 
 
 
 
 
 
 
 
 
 
 
 s 
 
 
 ^oo 
 
 -s 
 
 m 
 
 
 i§ 
 
 
 pf 
 
 n- 
 
 tf 
 
 
 S 
 
 
 
 
 
 
 § 
 
 A 
 
 t». 
 
 h 
 
 ^ 
 
 > 
 
 .^ 
 
 rw 
 
 »■ 
 
 IV 
 
 l€ 
 
 IB- 
 
 h3 
 
 
 
 
 
 
 I 
 
 •d 
 
 !« 
 
 fl- 
 
 c 
 
 a 
 
 fe 
 
 16 
 
 IJT 
 
 ^ 
 
 IT 
 
 tr 
 
 "^ - 
 
 1 
 
 
 
 
 -S § 
 
 X, 
 
 ■*! 
 
 :f! 
 
 J3 
 
 ^ 
 
 s^ 
 
 M 
 
 o> 
 
 'a 
 
 •a 
 
 ,^ 
 
 
 P- 
 
 W' 
 
 h» 
 
 fr 
 
 » 
 
 o3 
 
 
 
 
 
 
 1 
 
 6D 
 
 «» 
 
 TS 
 
 ■a 
 
 .o 
 
 (7 
 
 fB- 
 
 Iw 
 
 hr 
 
 IB- 
 
 TS . 
 
 
 
 
 
 
 $ s 
 
 
 ^ 
 
 
 ^ 
 
 ja 
 
 ■g-l 
 
 ^ 
 ^ 
 
 lef 
 
 \o 
 
 
 Pi 
 
 oj 
 
 
 
 
 
 
 a 
 
 M 
 
 -a 
 
 H^ 
 
 -»3 
 
 p< 
 
 ys 
 
 IP- 
 
 hJ 
 
 1= 
 
 P- 
 
 H 
 
 
 
 
 
 
 
 
 
 ^ 
 
 
 
 
 
 
 m 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 -^ 
 "§ 
 
 03 
 
 •S ft 
 
 09 
 
 -a 
 
 0) 
 
 4 
 
 '4 
 
 
 o 
 
 Ph 
 
 ij 3 
 
 u 
 
 J 
 
 •a ^- •- 
 
 fe « ^ 
 
 I (5 § 
 
 J is 0) 
 
 |2i iK :g 
 
 «3 <S .S 
 
 ^ jg .-a 
 
 o S a 
 
 H 
 
LIST OF AUTHORITIES REFERRED TO IN THE 
 PRESENT WORK. 
 
 BoPP's ' Comparative Grammar,' translated by Eastwick. 
 Max MtJLLER, ' Lectures on the Science of Language.' 
 
 ' Chips &om a German Workshop.' 
 SOHLEIOHEB, 'Compendium der Vergleiohende Grammatik.' 3rd edition. 
 1871. (One volume of a translation into English has appeared.) 
 CoESSEN, ' XJeber Aussprache, Vokalismus und Betonung der Lateinisohen 
 Sprache.' 2nd edition, 1868. {The authority for Latin philo- 
 logy, and the storehouse from which Peile, Eoby, Wordsworth, 
 and others mentioned below, have drawn much of their infor- 
 mation.) 
 •Kritisohe Beitrage,' and 'Kritiache Nachtrage zur Lateinischen 
 Formenlehre.' 
 Cdetius, 'Grundzflge der Griechischen Etymologie.' 2ud edition, 1866.. 
 (Translated into English by Prof. Wilkins and Mr. England, 
 of Owens College, Manchesten) 
 'Tempera und Modi,' an admirable treatise on the formation of 
 Tenses and Moods in Greek and Latin. Berlin, 1846. (Now 
 out of print.) 
 ' Das Verbum der Griechischen Sprache.' Vol. i. 1873 : vol. ii. 1876. 
 (The latest result of Curtius' studies, superseding much of 
 'Tempora und Modi.') 
 ' Greek Grammar.' (Published in English as ' The Student's Greek 
 
 Grammar.') 
 'Greek Grammar Explained,' or 'Elucidations;' a translation by 
 Mr. E. Abbott of ' Erlauterungen zu meiner Griechischen 
 Schulgrammatik,' published by Prof. CurtiiiB in 1863, as -a, 
 companion to hia 'Greek Grammar.' 
 
xiv Iiist of AuiAorities. 
 
 Leo Metek, 'Vergleichende Grammatik der Griechischen und Latein- 
 ischen Sprache." (EBpeoially valuable for its exhaustive col- 
 lection of examples, &oni which the student can form his own 
 induction as to particular formations.) 
 
 Pbile, 'Introduction to Greek and Latin Etymology,' 3rd edition. 
 (Maoimllan, 1875.) 
 A Primer of Philology. (Macmillan, 1877.) 
 
 KoBT, • Latin Grammar (Vol. I) on Sounds, Inflections and Word-Forma- 
 tion.' (Macmillan, 1871.) 
 
 Satoe (Eev. A. H., Fellow of Queen's College, Oxford), 'Principles of 
 Comparative Philology.' ist edition. (Trubner and Co., 1874.) 
 
 WoBDSWOETH (Eev. John, Brasenose College, Oxford), 'Fragments and 
 Specimens of Early Latin.' (Clarendon Press, 1874.) 
 
 Feeeab, ' Comparative Grammar of Sanskrit, Greek, and Latin,' Vol. I. 
 (The author's death unfortunately left the work half-finished 
 before he had treated of Verb-Inflections.) 
 
 Faeeae (Eev. Dr., Canon of Westminster), ' Chapters on Language,' and 
 ' Families of Speech.' 
 
 Whitney (Professor of Sanskrit and Comparative Philology in Yale 
 College, U. S. A.), ' Life and Growth of Language.' (Pub- 
 lished in England by H. S. King and Co., 1875. 
 
CONTENTS. 
 
 CHAPTEE I. 
 
 PAOE 
 
 Intkoductoby 1-3 
 
 CHAPTER II. 
 
 CliASSrFIOATION OP LANGUAGES — 
 
 'Morphological' and 'Genealogical' classifications — Table of 
 the Indo-European family — Table of correspondence between 
 its members — ' Komanio' or ' Romance' languages ; their 
 origin — Table of correspondence between them — Features of 
 change from synthetic to analytic languages exhibited by 
 them 4-27 
 
 CHAPTEE III. 
 
 CLASSimCATION OF SOUNDS — 
 
 General principle of phonetic change to secure ease of articula- 
 tion — Comparative strength or difficulty of sounds according 
 to the physical conditions of their production — Consonants ; 
 their threefold classification — Vowels and Diphthongs. Re- 
 lation of sounds to letters — Sanskrit alphabet . . . 28-39 
 
 Appendix to Chaptbb III — 
 
 On the Greek and Roman Alphabets 40-48 
 
 CHAPTER IV. 
 Changes and Modifioations op Sounds — 
 
 General principles — ' Dynamic' change ; Reduplication, Vowel- 
 Intensification, and Nasalisation — 'Phonetic' change: Sub- 
 stitution, Loss, Assimilation, and Dissimilation of Vowels and 
 Consonants — Changes due to indistinct utterance — National 
 peculiarities of utterance — Grimm's Law .... 49-91 
 
xvi Contents. 
 
 CHAPTEE V. 
 
 Fi.6E 
 
 Formation of Words — 
 
 Badical and Formative Elements — Eoots, Stems, and Inflec- 
 tions — Processes of Word-Formation 92-101 
 
 Appendix to Chapteb V — 
 
 A. List of Nominal Suffixes — B. Derivative Verbs in Greek 
 
 and Latin 102-104 
 
 CHAPTEE VI. 
 
 NOUN-lNPLECTION — 
 
 Number of Cases — Gender and modes of Generic Distinction — 
 Declension — Suffixes of Case-Lnflections in detail — Compari- 
 son of Adjectives — Paradigms 105-141 
 
 CHAPTER VII. 
 
 Inflection op Pkonouns — 
 
 Pronouns witbout Gender : First and Second Personal, and 
 Keflexive — Paradigms — Pronouns witb Gender: Paradigm 
 of stem ta — Peculiarities of Latin Pronominal Declension . 142-157 
 
 CHAPTEE VIII. 
 
 Verb-Inpleotion — 
 
 Distinction between Noun and Verb — Augment, Thematic or 
 Connecting Vowel — Classification of Verbs — Person-Endings 
 — Mood-Signs — Tense-Stems and their formation — Infinitives 
 and Participles 158-236 
 
 APPENDIX I. 
 
 Specimens of Early Latin Inscriptions 237 
 
 APPENDIX II. 
 Formation of Adverbs in Greek, Latin, and Sanskrit — Table of 
 
 Prepositions . . , , . . . , , 243 
 
 General Index 251 
 
 Index of Sounds and Forms Explained 255 
 
CHAPTER I. 
 
 Inteoductoet. 
 
 The main object of the present work, as indicated by its title, Objectfof 
 is the philological explanation of the Inflections in Greek and ^^"^ ' 
 Latin. For the purposes, however, of such explanation it is 
 necessary to presume a certain acquaintance with the main re- 
 sults of the Science of Language or Comparative Philology, and 
 with the terms commonly in use among philologists ; and we 
 must begin with a clear understanding, (i) of the general classi- Preliininary 
 fication of Languages, and the place in the history of human 
 speech of those languages with which we are immediately 
 concerned, viz. Greek and Latin ; (2) of the classification of 
 Sounds and the letters by which they are represented, and of 
 the processes of change which sounds have undergone in human 
 speech ; (3) of the constituent elements of language — i. e. 
 'roots' or simplest forms, and formative elements, including 
 Inflections. Into questions of the origin of language, and the 
 connection between its simplest discoverable forms and the 
 ideas expressed by them, we need not enter. The balance 
 between the two extreme views of language as a conventional ' 
 
 ' The view that language ia 'conventional,' in the sense that people 
 meet or ever met together to deliberate on the meaning and changes of 
 words, is of course easily ridiculed. The term, however, may also be 
 applied to language in opposition to the idea that there is any necessary 
 connection between words or 'roots' and the ideas signified by them, or 
 that there are organic forces of growth in speech itself which, by some 
 mysterious natural process, without human agency, produce new material 
 
 B 
 
[chap. 
 
 Discussion 
 oftheNa- 
 tureand 
 Origin of 
 Language 
 unneces- 
 sary. 
 
 production — a view based upon the apparent meanlessness of its 
 formal elements — and language as an organic being, producing 
 those formal elements by virtue of a mysterious principle of 
 growth inherent in its nature, has been clearly drawn by Pro- 
 fessor Max Miiller in those Lectures on the Science of Lan- 
 guage^ ' to whose world-wide popularity (it has been well said) 
 Comparative Philology owes it present position and its present 
 charm :' and for this question, with others that relate to the 
 aim and methods of Comparative Philology, we cannot do better 
 than refer to a book which for every English student of that 
 science should be the avenue by which he approaches it. But 
 we may examine the relation of Latin and Greek words to each 
 other or to Sanskrit, or trace the history of varying forms in 
 any one language, without touching such questions as that of 
 the Onomatopoeic or Interjectional origin of human speech (the 
 ' Bow- Wow' and ' Pooh-Pooh' theories of Professor Max Miiller), 
 or deciding whether language arose from imitations of cries and 
 sounds (as it often does in the nursery) or from exclamations 
 expressive of pleasure or pain or other emotions of mind or 
 body. The furthest researches into the history of language 
 and the utmost possible analysis of written or spoken languages 
 bring us to certain primitive and elementary combinations of 
 
 and alter old. These ideas exercise a kind of fascination over some 
 students of language ; but it is reasonably maintained that spoken lan- 
 guage (as distinct from fhe facuMy of language implanted in man) is an 
 external medium of communication, learnt in childhood by mere imitation, 
 and not inherited as a race-characteristic, or independently produced by 
 each individual with his mental and bodily growth. The acquisition by 
 each individual of his own language in childhood is the gradual accumula- 
 tion, by imitation of those around him, of a stock of signs, which are so far 
 arbitrary and 'conventional,' in that each is bound to the idea signified 
 only by a tie of mental association, and not by any natural and necessary 
 connection. 'Language* is not a faculty or capacity, but a developed 
 result ; and the assumption that man is gifted at his birth not only with 
 the capacity, but also with its elaborated results, is n, theory, not of a 
 Divine, but of a ' miraculous ' origin of speech. 
 
 The question of the Nature and Origin of Language is treated in a 
 popular form by Prof, Whitney in his ' Life and Growth of Language ' (see 
 especially ch. xiv). He assigns a more important place to the ' imitative ' 
 or ' onomatopoeic ' principle than some philologists allow ; but on a ques- 
 tion of this nature much latitude of opinion is possible, and his remarks are 
 very suggestive and instructive. 
 
 ' See Lectures, Series I. Leot. v. on ' Comparative Grammar.' 
 
I.] Ini/fodMctory. 3 
 
 sounds which we call 'roots;' but we cannot arrive with any 
 certainty even at the ultimate ybrm of these roots. Th^ earliest 
 traceable condition of that 'Indo-European' speech, of which 
 (as we shall see) Sanskrit, Greek, and Latin, with the other 
 languages comprised under that term, are dialectical varieties, is 
 evidently far removed from the primeval time when language 
 first arose — as is clear, among other reasons, from its highly 
 developed vowel-system, and the employment of vowel change 
 rather than the earlier and simpler method of reduplication ' to 
 express modifications of ideas : and therefore speculations as to 
 the connection between its elements and the ideas expressed by 
 them can be of little use for the purposes of a science which (as 
 Professor Max Miiller points out) collects facts and accounts for 
 these facts as far as possible. It is at least probable that 
 language owes its origin to a combination of imitational and 
 interjectional sounds : for it is difficult to assign any other origin 
 to speech, and the phenomena of dawning speech in infants point 
 to this as one natural origin for conscious sound ; but the exist- 
 ing' materials of language, with which alone Philology has to do, 
 give us no data upon which to base any calculations as to the 
 exact mode in which such growth of speech began. Passing by, 
 therefore, such questions, we may proceed to the consideration 
 of the points already mentioned as introductory to the explana- 
 tion of Greek and Latin Inflections; viz. the classification of 
 languages, the classification of sounds, the changes and modifica- 
 tions of sounds, and the elements of word-formation. These will 
 occupy the next four chapters, and the discussion of Inflections 
 properly so called will follow in chap. vi. 
 
 ' On this point consult Peile'a ' Introduction to Greek and Latin 
 Etymology,' pp. 1 73 sqq. (3rd edition) ; and see below, ch. iv. 
 
CHAPTEE II. 
 
 ClvASSiriCATION OF LANGUAGES. 
 
 Twofold 
 classifica- 
 tion of lan- 
 guages. 
 
 a. Morpho- 
 logical, 
 
 Greek and Latin are 'inflectional' languages of the 'Indo- 
 European' family. These terms refer to a double classification, 
 viz. (a) morpJiological, according to the mode in which medifi- 
 cations of ideas are expressed by combination of primitive 
 elements; (6) genealogical, according to similarity of gram- 
 matical forms. 
 
 {a) The 'morphological' classification idistinguishes three 
 ' stages' of growth in language^ : — 
 
 1. 'Kadical' or 'Isolating,' in which the simplest elements of 
 speech or 'roots'^ are employed as words, without modification 
 of their own sounds or combination with each other : mere 
 juxtaposition of isolated roots expressing modification of ideas. 
 Chinese and its kindred dialects are examples of this stage. 
 
 2. 'Agglutinative' or 'Terminational;' in which roots are 
 joined together to form words. In such compounds one root 
 loses its independent form, and appears as a sound expressive of 
 meaning tacked on (as prefix or suffix) to the other, which 
 
 ' For details which are beyond the provinee of the present work, see 
 Max Miiller's Lectures, Series I. Leot. viii (on 'Morphological Classifi- 
 cation '). 
 
 ' It is necessary here to anticipate the distinction (explained below in 
 oh. v) between the ' radical ' elements in words or ' roots,' i. e. the simplest 
 part of each word which expresses its general idea or meam'ng, and the 
 formative elements by which this general idea, common to many words, is 
 defined and modified. For purposes of instruction, oh. v. may, if it seems 
 convenient, be taken before ch. ii. 
 
Classification of Languages: 5 
 
 remains as the primitive element or ' root' of the word. The 
 Finnish and Tataric languages, and the dialects of the aborigines 
 of the Pacific Islands (grouped by Professor Max Muller and 
 others under the name ' Turanian') appear to be in this stage. 
 
 3. ' Inflectional ; ' in which both roots (i. e. the ' root' proper 
 and the suffix) are modified according to regular processes for 
 the expression of meaning. The Semitic and Indo-European 
 families of speech fall under this head. 
 
 This classification may be illustrated by constructing an illustration 
 imaginary history, passing through these three stages, of some stages of 
 Greek or Latin word, e. g. e'/w {iho). The ultimate forms- or cai growth! 
 ' roots' to which philological analysis has reduced the two 
 syllables of which this word is composed, are i (idea of ' going') 
 and ma (ist personal pronoun). We should have these roots 
 combined in the Eadical stage by simple juxtaposition \i ma] ; 
 in the Agglutinative, by suffixing olie to the other, and modi- 
 fying the root thus suffixed [i-ma, i-mi\ ; in the Inflectional, by 
 modification of both roots [ai-mi, ci/ii]. 
 
 The student must bear in mind that this is a classification, 
 
 act of languages, but of varieties of linguistic development ; and 
 
 that though, when w© speak of Chinese as 'Isolating,' Hun- T^e tiu-ee 
 
 . , . . 'stages not 
 
 garian as ' Agglutinative, Greek and Latin as ' Inflectional,' we always mu- 
 
 ., „ ' , , . . . , , tuallyexclu. 
 
 give a correct idea of the general characteristics of those lan-sive; 
 guages, and their structural contrast to each other, we do -not 
 imply that there is in any one of these languages no trace of 
 the structural characteristics of the others. The higher ' agglu- 
 tinative' dialects (e.g. Finnish and Hungarian) are almost inflec- 
 tional, and in some respects analogous to the Indo-European 
 languages ; while in inflectional languages many forms are 
 found in which the 'root' maintains its own form and dis- 
 tinction from the suffixes, no less completely than in an agglu- 
 tinative language. Thus to take the Greek word just employed 
 as an illustration ; beside el/u, a typical inflectional form, we have 
 i-fiev, i'-Tf, in which the root proper remains unmodified. Again, 
 beside el/il (sum), we have the dialectical variety ia-jA (Aeol.), in 
 which the root as of the Sanskrit asmi remains in a much less 
 jnodified form — ^the « being only a phonetic variety of a (see 
 
6 Classification of Languages. [chap, 
 
 p. 36) — and which in this respect of root modification marks 
 a transition stage, and that not very far upon the road, from 
 Agglutination to Inflection. And as in English the line 
 between mere juxtaposition and word composition is often 
 hard to draw, cp. e. g. ' house floor' and ' housetop,' ' godlike' 
 and 'godly,' so it is often in languages of low development a 
 matter of doubt where isolation ends and agglutination begins ; 
 and even Chinese, the purest example of the first, is by some 
 regarded as being in its colloquial forms and in some of its 
 dialects a language of compounded words, 
 nor always On the other hand, it is misleading to speak of these ' stages 
 
 historical for , , . i ^ ■ . . » 
 
 each Ian- of development as markmg a definite histoncal progress of 
 individual languages from a lower to a higher state of civili- 
 sation. Philology offers no proof that all inflectional languages 
 have previously passed through an agglutinative stage of 
 development, or that isolating languages must pass on with 
 increasing civilisation to an agglutinative and finally to an 
 inflectional stage. The facts indeed of language tend to show 
 that ' an isolating or agglutinative stage does not imply civilisa- 
 tion or the reverse, and that no amount of culture, no amount 
 of years, and no amount of foreign intercourse, has been able to 
 change the radical character of a language '.' "What has once 
 become, remains : Chinese, which at a very early stage became 
 what it is, a language of the ' isolating' type, remained in 
 that condition, and, like Chinese civilisation, stopped once and 
 for all in its growth. It remained, so to speak, a precocious 
 child ; while of inflectional languages it may be said that they 
 grew to manhood before they stopped — the agglutinative stage 
 representing an intermediate period of advance from childhood 
 towards manhood. As languages, sufficiently developed for the 
 purposes of national speech, those of each type remain, and will 
 remain, what they have once become — isolating like Chinese, 
 agglutinative like Tataric, or inflectional like Latin. This fact, 
 however, need not preclude the supposition that in the earliest 
 growth of language — a growth which lies far beyond the 
 
 ' Sayce, ' Principles of Comp. PMlology,' oh. v. p. 137. 
 
II.] Classification of Languages. 7 
 
 remotest period to which the evidence of, or inference from, 
 the facts of written or spoken languages carries us back — the ele- 
 ments out of which it grew must have exhibited the progressive 
 stages of which we speak. ' Agglutination,' for instance, pre- 
 supposes two things Vhich could be ' glued' together : that is, 
 it presupposes an earlier stage of ' isolation.' ' Inflection' pre- 
 supposes something which was not ' inflected,' but was capable 
 of being so ; that is, it presupposes a state of either isolation or 
 agglutination, or of both in succession. These ' stages' then are 
 (as has been already pointed out) varieties of linguistic develop- 
 ment, the developed results of which, in written or spoken 
 languages, exhibit as a rule the characteristics of one or other 
 variety, not of two or of all in succession. 
 
 With this limitation it seems perfectly reasonable to reconcile 
 this theory of three stages of development with the facts of 
 language, which ofier no evidence of actual progress by indi- 
 vidual languages from one stage to another. They are types 
 or varieties of development; not historical stages of growth. 
 Accoijdingly, though spoken Chinese and some of the higher 
 agglutinative dialects are said to approach more or less to 
 ' inflection, their general characteristic remains what it always 
 has been. And in Indo-European languages, though analysis 
 of the verb-inflections (above, p. 5) seems to point to an earlier 
 agglutinative stage and a still earlier period of simple isolated 
 roots, we cannot as a matter of fact point to a period when the 
 two parts e. g. of ei/ie were ever of full and equal power as inde- 
 pendent words in the Greek language ; and though we may trace 
 formative suffixes back to roots with general meanings [e. g. -ta/r 
 in ipater, mater, etc., and the terminations in -rap, -tor of nouns 
 of agency, to the root = to ' cross' or ' get through' (with a thing) 
 which appears in trans and t'hrovigK\, we do not thereby prove 
 the previous existence, as historical fact, of an isolating or ag- 
 glutinative stage in what are, and so far as our evidence goes 
 always have been, inflectional languages. Comparison and 
 analysis of the Indo-European languages enable us to form 
 some conception of the state of civilisation attained by tj^ose 
 who spoke the primitive mother tongue upon the table 'lands 
 
8 Classification of Languages. [chap. 
 
 of Central Asia — a civilisation probably much below the con- 
 temporary civilisation of China ; yet the language of China re- 
 mained in the isolating stage, while that of our Indo-European 
 ancestors must (if we regard development through all three 
 stages as necessary) have already passed from the stage of 
 isolation through that of agglutination to the inflectional stage,, 
 in which it is presented to us by the earliest ascertainable data 
 of Philology. These data show us the case and person endings, 
 for example, in the condition of grammatical forms — inflections 
 or modifying suflSxes ; and there is no trace, as far as meaning 
 goes, of their having ever been separate or agglutinated particles. 
 While Chinese has remained in the most primitive stage, fos- 
 silised, so to speak, like the whole Chinese civilisation, the 
 Indo-European languages, so far as we can trace them, have 
 always been in the most advanced stage : but for enquiry into 
 the causes of this difference, which is an enquiry into the mental 
 characteristics of the different races of mankind, the facts of 
 language offer no foundation : and it is of these facts alone 
 (as has already been said) that Comparative Philology takes 
 cognisance. 
 
 fc.Genealo; (6) Genealogical Classification : — 
 
 fioation. This classification gives us three main 'families' or groups of 
 
 languages, according to similarity (a) in vocabulary, (6) in gram- 
 matical forms : viz. 
 
 1. Semitic, including Hebrew, Arabic, and their kindred 
 dialects. 
 
 2. Indo-Ev/rojpean (otherwise called ' Indo-Germanic' or 
 ' Aryan'), including the following subordinate classes or groups 
 — Indie, Iranic, Hellenic, Italic, Keltic, Slavonic, and Teu- 
 tonic. 
 
 The languages comprised under both these heads are ' inflec- 
 tional.' 
 
 3. 2'MTOmflM' (otherwise called 'Ural-Altaic,' 'Mongolian,' 
 
 ' The name ' Turaman! familiar from its use by Prof. Max MtUler, is 
 retained as a designation of the * class ' of languages, for which some prefer 
 
II.] Classification of Languages. 9 
 
 ' Tataric,' or ' Scythian'), including the languages of the nomadic 
 races of Asia and Europe (Laplanders, Hungarians, Samoyeds, 
 Turks, Mongols, Tartars, etc.), and the dialects of Siam, Malay, 
 and the Polynesian Islands. These languages are all ' agglu- 
 tinative,' and though it is impossible to trace in them anything 
 like the same family likeness of vocabulary or grammatical 
 forms, as in the Indo-European or Semitic groups — the absence 
 of such family likeness being one distinguishing feature of the 
 dialects of a nomad population, among whom no nucleus of a 
 political, social, or literary character has ever been formed, 
 which could give fixity to and create definite standards for 
 language — we may accept, as sufficient warrant for the appli- 
 cation to them of the term ' class ' or ' group,' the statement 
 that ' some of the Turanian numerals and pronouns, and many 
 Turanian roots, point to a single original source ; and the 
 common words and common roots which have been discovered 
 in the most distant branches of the Turanian stock, warrant the 
 admission of a real, though very distant, relationship of all 
 Turanian speech^.' 
 
 It is, however, with the ' Indo-European ' languages that we ' Indo-Euro- 
 are now concerned. The fact implied by this name is now a guages, 
 familiar truth established by Comparative Philology, viz. that a the term, 
 comparison of the languages of all the civilised races of Europe, 
 and two at least of those of Asia (Indian and Persian), proves 
 beyond doubt that these languages are branches of a common 
 stock, and with reasonable probability that there was once a 
 time when the ancestors of Germans and Slaves, of Greeks, 
 Italians, and Kelts, of Persians and Hindiis, were settled in a 
 rudimentary stage of civilisation upon the plains of Central 
 Asia, and spoke the same language, subject to such dialectical 
 variety as must always arise in a primitive state of society, with 
 imperfect communication between the scattered members of a 
 single nation or even of a single tribe. This time is, however, 
 pre-historic ; and though, upon the evidence of language, phUo- 
 
 one or other of the names here quoted. See Sayce, p. 21, note; Whitney, 
 'Life and Growth of Language,' pp. 231, 232. 
 1 Max MuUer's Lectures, 1. viii. 
 
lo Classification of Languages. [chap. 
 
 legists sketch out for us the state of civilisation and manners of 
 this primitive people, and even claim to restore the 'Ur-sprache ' 
 or primitive language ^ anterior to any dialectical variation, we 
 cannot regard these as historical facts, the onlj fetch before us 
 being the phenomena exhibited by different kindi-ed languages, 
 by Sanskrit, Greek, Latin, etc., which we group under the term 
 ' Indo-European.' 
 
 This term seems to be that of widest meaning, and most 
 obviously inclusive of all the languages in question. The name 
 Indo-Germanic, employed by many German gtholars, is hardly 
 comprehensive enough of the European branch of the family ; 
 while the names Sanskriiic, Japhetic, and Mediterranecm (sug- 
 gested by Ewald), are each open to the objection of counte- 
 nancing misleading notions ^. There remains the term Aryan, 
 popularised in this country by Professor Max Miiller's Lectures, 
 and employed by many philologists as a designation of the Indo- 
 European family; but by some in the more restricted sense of 
 Indo-Iranian, i. e. to denote the Asiatic sub-division of the 
 Indo-European family. This latter usage, however, is likely to 
 create confusion with its much more frequent use in the wider 
 signification of Indo-European *. 
 
 ^ Wherever in the following pages reference is made to the primitive 
 ' Indo - European ' form or ' type ' of words in kindred languages, such 
 type must not be conceived of as necessarily having, or having had, real 
 existence, hut as an imaginary form showing in combination the elements 
 which have been diflferently retained in different languages. 
 
 " Sanskritic might suggest the idea that all Indo-European languages 
 are derived &om Sanskrit ; Japhetic, from the Hebrew point of view of the 
 three ancestors of the human race, would include tribes in Northern Europe 
 and Asia who speak Turanian languages ; while Mediterranean refers only 
 to one phase in the history of Indo-European nations, and the central 
 position once occupied by, but now no longer belonging to, the people who 
 spoke these languages. 
 
 ' The term Aryan ha9 the advantage over Indo-JEfuropean of being short 
 and (as a word of foreign origin) of lending itself more easily to any 
 technical definition that may be assigned to it : and as a mere ticket or 
 label of classification, there is no doubt much to be said for its use. 
 I should not therefore presume to discard it altogether ; but I still think 
 that the fact implied on the face of the term Indo-European (a term 
 sanctioned by the high authority of Bopp) is a good reason for on the 
 whole preferring this latter term. The existence too of another and more 
 limited use of the term Aryan (as = Asiatic or Indo-Iranian) is somewhat 
 against its acceptance as the technical term for the whole family of Ian- 
 
II.] Classification of Languages. ii 
 
 The languages comprising the Indo-European (or Aryan') Subdivisions 
 lamily may be arranged thus m three mam divisions : — European 
 
 family. 
 
 A. Asiatic Division : — 
 I. Indie. 
 
 a. Sanskrit, the ancient literary language of the Vedas, 
 or sacred hooks of the Hindus : Prakrit (including 
 Pali, the sacred language of the Buddhists in Ceylon) 
 being the provincial dialects of the mass of the com- 
 munity '. 
 
 6. Modem Indian dialects, Hindi, Bengali, Mahratti, etc. 
 
 II. Iranic. 
 
 a. Zend (or Old Bactrian), the language of the Zend- 
 Avesta or sacred books of the religion of Zoroaster. 
 
 b. Old Persian, of the ' cuneiform ' inscriptions. 
 
 c. Modern Persian. 
 
 d. Armenian. 
 
 B. South-West European Division ; — 
 
 III. Hellenic. 
 
 a. Ancient Greek. 
 h. Modern Greek. 
 
 guages, however great the preponderance of authority for the wider use. 
 In deference, however, to this authority, it should be retained as a collateral 
 term with Indo-European. For the origin and uses of the term arya 
 I need only refer to Prof. Max MUUer's Lectures, I. vi. pp. 224-236, 
 1st edition. [I am glad to find, what at the time this note was first 
 written I did not know, that I have the support of Mr. Peile in preferring 
 ' Indo-European ' (' Introduction,' p. 34, 3rd edition).] 
 
 ' The word ' Sanskrit ' (Samskrita, ^^TT = confectm, comtrmtus) 
 nleans what is rendered 'fit' or 'perfect,' i.e. for sacred purposes; hence 
 'purified,' 'sacred,' 'Prakrit' (Prftkrita, BT^if ='derived' or 'secondary') is 
 the term applied to the spoken dialects which gradually rose out of Sanskrit 
 as from a source or type (prakriti), by the natural process of change and 
 corruption which the pure Sanskrit underwent in its adaptation to the 
 exigencies of a spoken dialect. The various modifications of Prakrit are 
 the links which connect Sanskrit with the modem dialects of Hindustan, 
 It should be noted that there is a large body of Prakrit (or non-Vedic) 
 literature included under the general term Sanskrit; the older Sanskrit 
 literature being generally specified as ' Vedic' 
 
13 Classification of languages. 
 
 IV. Italic. 
 
 a. Latin, Oscan, Umbrian, the three dialects of ancient 
 
 Italy. 
 6. The modern 'Romance' (or 'Romanic') languages; 
 
 viz. Italian, Spanish, Portuguese, Proven9al, French, 
 
 Wallachian, Romansch. 
 
 V. Kdtic. 
 
 a. Cymric or Armorican, including Cornish, "Welsh, and 
 
 the dialect of Brittany. 
 h. Gadhelic, including Gaelic, Erse, and Manx. 
 
 C. North-West European Division : — 
 
 VI. Slavonic. 
 
 a. Lettic — Old Prussian and Modern Lithuanian. 
 h. Slavonic Proper — Bulgarian, Russian, Polish, Bohe- 
 mian. 
 
 VII. Teutonic. 
 
 a. High German (spoken in the upper countries of the 
 Rhine, Main, and Neckar), including Old High 
 German, Middle High German, and Modern Ger- 
 man. 
 
 6. Low German (in the Lowlands of North Germany, 
 Holland, and Belgium), including (i) Gothic, Anglo- 
 Saxon, and English ; (2) Old Saxon and Frisian, and 
 Dutch. 
 
 c. Scandinavian — Old Norse and Modern Icelandic, 
 Swedish and Danish. 
 
 Comparative Of these sub-divisions the Asiatic (A) contains most that is 
 and order of ancient in sounds and fabric of language, and fewest strongly- 
 frompritni- developed individual forms. The South- West European (B) stands 
 next in this respect ; while the North-West European group (C), 
 shows most individuality of development, and fewest remains of 
 a common stock. The diffusion of this common stock under all 
 these different forms over the Continent of Europe is generally 
 recognised as the result of successive migrations westward from 
 the original home of the Indo-European or Aryan nations in 
 
oi * fca J3 
 
 ^Hi-9 
 
 ^ 
 
 tjj 
 
 •H 
 
 Ci-tr 
 
 A 
 
 ^ 
 
 ^ 
 
 sr 
 
 |> 
 
 Ch 
 
 ,^ 
 
 
 g 
 
 f 
 
 CD 
 
 r. 
 
 a 
 
 •-' 
 
 It "73 
 
 1 
 
 
 
 
 1 
 
 tei 
 
 H 
 
 ■^ 
 
 g* 
 
 < 
 
 1 
 
 o 
 
 CD " 
 
 
 5' 
 
 cu 
 
 P.^ 
 
 
 
 
 
 
 
 
 
 
 
 
 
 ^ 
 
 1 
 
 a 
 
 
 b 
 
 3 
 
 
 
 
 R' 
 
 ff 
 
 
 r 
 
 
 H 
 
 a 
 
 
 ■rt 
 
 
 <D 
 
 r 
 
 1 
 
 
 1 
 
 
 
 (71 
 
 
 
 
 
 
 
 
 
 n 
 
 ri-O 
 
 !^ 
 
 
 o 
 
 
 s, 
 
 3. 
 
 
 ^ 
 
 
 ^ 
 
 s. 
 
 
 J 
 
 
 CL. 
 
 f 
 
 
 1 
 
 
 |- 
 
 
 
 n 
 
 
 
 
 
 B 
 
 
 
 & 
 
 D 
 
 (r? 
 
 IS 
 
 
 -i9 S 
 
 K» tip- 
 
 aSi^-" ft 
 
 £?> S n r\ 
 
 9 
 
 ot; 
 
 w 
 
 c^S Pa » 
 
 So? 
 S' a B 
 
 (5. 
 
 o O c S 
 CD ta S.'?S 
 
 
 ht)02 
 
 f- 
 
 
 OCSJ 
 
 Q 
 
 OS 
 P £'• o ." g" 
 
 ^ 
 & 
 
 S 
 
 P a.B H2. K, 
 
 
 to g'lS 
 
 a-S'2. 
 B p-o 
 
 O 
 
 0-5- ■ « 
 
 9 
 p- 
 
 00{>0 
 
 
 s 
 
 ?g=?s^g0 ^ 
 
 fed 
 a 
 
 o 
 
 !» 
 
 EC O m 
 
 p p s 
 
 a 
 
 
14 Classification of Languages. [chap. 
 
 Central Asia ' : and this being bo, the phenomena just noted lead 
 us to infer that the first to separate and lose connection with 
 the parent stock were the ancestors of the nations comprised 
 under group (C) ; that these were followed by the ancestors of 
 those under group (B) ; the ' Aryan ' proper comprised under 
 group (A) alone remaining East of the Ural Mountains. This 
 inference from the phenomena of language is borne out by the 
 geographical position of the different branches of the race. If 
 we take a map, we shall find that, as a general rule, the more 
 eastward the position of an Indo-European people, the more 
 traces of what is old and common to other languages of the same 
 family are retained in its language ; while the further north-west 
 and west they have gone, the less of what is old and the more new 
 formations does their language retain. Nor does the western- 
 most position of certain languages in the group which stands 
 second in order of separation (e. g. the French, Spanish, and 
 Keltic), interfere with the truth of this general statement ; such 
 position being due to special historical causes, e. g. the spread of 
 the Roman Empire to the Atlantic Ocean, and (in the case of 
 Keltic) the gradual pressure of the Teutonic nations, driving the 
 Kelts further and further westward. These Kelts, whom we 
 meet with as the conquerors of B«me under Brennus (b.c. 390), 
 and 100 years later as the invaders of Macedonia and Greece, 
 and of whom Herodotus speaks as dwelling in the extreme 
 west of Europe '■*, apparently in Spain, must have spread into 
 
 ■ It mnat, however, be remembered that the evidence of successive 
 order of separation, furnished by the closer relationship of particular 
 dialects, is at the best vague, and the conclusions drawn from them 
 indefinite and uncertain, so far as anything like the establishment of a 
 historical order of separation is concerned. If it can be shown that Latin 
 is most closely connected with Greek, it can, on the other hand, be 
 shown that in many respects Greek is most closely connected with San- 
 skrit : and probably all that it is really safe to aflSrm is that the various 
 dialects of the Indo-European family after a long continued community 
 separated gradually, until under different circumstances they established 
 thbir respective national independence. 
 
 ' Hdt. ii. 33, iv. 49. He speaks of them as Ifm tSiv 'HpaitXrjtaiv oi'qXlwv, 
 and (after a tribe called KiJi/ijtoi) ^ffxaroi vphs ^Xfov ivaiiiaiv rSiv iv rg 
 Eipiiwin, His language is that of a man living on the shore of the Medi- 
 terranean, to whom all knowledge of these western countries came from 
 people who bad saUed through the Straits of Gibraltar, outHde the so-called 
 
II.] Classification of Languages. 15 
 
 Switzerland and Tyrol; and, after occupying Gaul, Belgium, arid 
 Britain, were driven by pressure of the Teutons to the extreme 
 north and west of Gaul and the British Islands, where their 
 language has survived to our own day, though gradually dis- 
 appearing (like Cornish) under the influences of increased com- 
 munication with the mass of the English-speaking population. 
 Some philologists, indeed, take a different view^ and maintain 
 that, looking to the present distance from the original home of 
 the respective Indo-European nations, the Kelts must have been 
 the f/rst, and the Slavonians the last to move westward ; and 
 that the Slavonians, finding the rest of Europe occupied, were 
 forced to make their new home in its northern and eastern 
 regions. This may be so ; but in the absence of history, lan- 
 guage (which has been called 'fossilised history') is our best 
 guide ; and language seems to postulate a longer separation 
 from the primitive stock in the case of the Teutonic and 
 Slavonic groups than for any of the others. 
 
 The following diagram (adapted from Schleicher's 'Com- 
 pendium') will illustrate the successive migration and bifur- 
 cations of the Indo-European family — the separations being 
 indicated by lines striking downwards, and the degree of 
 separation or proximity by greater or less deflection from the 
 uppermost Une^. The vertical order of the column to the 
 right hand corresponds to the horizontal order of the previous 
 Table. 
 
 ' Pillars of Hercules.' For other examples of Herodotus' relative use of 
 terms ' iiom a Mediterranean point of view,' compare i. 6, ivrbs "AXvos 
 troT&iiov : 1. 74 (of the Halys), pfoiv avai, and an instructive note to the 
 latter passage in Woods' edition (' Catena Classicorum ' series). 
 
 ' Schleicher's diagram is possibly open to modification, in respect of the 
 position assigned by him to Keltic. He believes in a ' Graeco-Italo-Keltic ' 
 period, marked by the division of the a sound into a, e, o, and (after the 
 separation of the Greeks) in an ' Italo-Keltic ' period marked by loss of 
 aspirates, retention of spirants, and loss of the old middle voice; while 
 finally, after separating from the Italians, the Kelts lost the ablative and 
 reduplicated perfect. Other philologists, however, connect Keltic with the 
 North European languages, tracing a general analogy (e.g. in the number of 
 diphthongal sounds, being four in Keltic and Teutonic, but six in Hellenic 
 and Italia dialects) between Slavonic, Teutonic, and Keltic. The argu- 
 ments on both sides are briefly but clearly stated by Mr. Peile ('Introd.' 
 pp. 27, 28, 3rd edition), who inclines upon the whole to Schleicher's view. 
 
i6 
 
 Classification of LaTignages, 
 
 [chap. 
 
 SSp of San- ^^ looking down the right-hand column, we find that the 
 |^"*'<^™ek. three first languages of which any considerable literature re- 
 
 ^ 'a. 
 
 o 
 
 1^ 
 
 o <! 
 
 i 
 
 Q 
 
 ^ » 
 
 d 
 
 H 
 
 O 
 tsi 
 
 Hi 
 bd 
 
 f> 
 
 Q 
 
 W 
 
 OS 
 
 mains are Sanskrit, Greek, and Latin, all 'dead' languages, 
 embodied in their respective literatures, not subject to the 
 
■11. J XJlasdfication of Languages. 1 7 
 
 constant changes of spoken languages, and therefore retaining 
 a more complete inflectional system than any other languages 
 known to us, though even in th^m the inflectional system is 
 already in a state of decay. These three languages thus form 
 the basis of philological enquiry for the Indo-European branch 
 of human speech ; and for the illustration of the inflectional 
 system of any one of them, the three together furnish all or 
 nearly all the available data. It is important to understand 
 clearly their relation'ship, as parallel branches of a common 
 stock, in the same generation of the genealogical tree. They 
 are sisters, or at furthest cousins ', and are not in any sense 
 derived from each other. Latin is not derived from Gfreek, 
 neither is Greek or Latin derived from Sanskrit, in the sense 
 in which the. liiodem Komance languages are 'derived,' i.e. 
 descended from, classical Latin. This relationship may be 
 proved by internal evidence from any of the languages given 
 in our Table. Thus Greek aor^p, Latin Stella ( = ster-vda), 
 German stern, English star, must be fuller and therefore older 
 forms than Sanskrit tkick, in which the initial s has been lost ', 
 Or again, if we take the grammatical forms, e.g. of ei/t" [Sanskrit 
 asmi, Latin svmi], we shall find that in i sing. Latin sum, 
 retains the s of the root as ifs) which Attic Greek «/i4 has 
 lost ; in 2 sing. Aeolic Greek eV-o-l is more perfect than Sanskrit 
 asi ; in i plur. Greek ccf-iuu retains the full root to- {as) as 
 .compared with Sanskrit smas, while the Latin termination -mus 
 (Sanskrit -mas) is older than Attic Greek -pew, the older dia- 
 lects preserving -/xf s : in 2nd plur. Latin estis is the most complete 
 fofm, Greek eVre the next, Sanskrit stha the most mutilated, 
 •having lost both initial and final letter ; in 3 plur. Latin sunt 
 (Sanskrit santi) is fuller than the oldest dialectical form in 
 .Greek [eWi Doric and Aeolic], from which the root es has entirely 
 .disappeared — much more so than the still more weakened form 
 eiVi. The Teutonic languages retain a correspondingly stronger 
 form than Greek, in German sind. French scmt, Italian sono, 
 Spanish son, are modifications of suni, A similar comparison 
 
 ' In the Veda, is found a fuller form, star, or stri. 
 
 
i8 
 
 Claswfication of Languages. 
 
 [chap. 
 
 Evidences 
 of relation- 
 ship be- 
 tween lan- 
 Buages. 
 
 of grammatical forms is employed* by Professor Max Miiller 
 (Lectures, Series I. Lect. V), to refute the theory that French, 
 Italian, and Spanish are derived from, the Proyen9al language, 
 which is, according to that theory, the only true ' daughter ' of 
 Latin. French som/mes, Ues, sont, besides Provenfal sem, etz, 
 son, are justly pointed to as fatal to such a theory. 
 
 If may be well to subjoin a few specimens of that cor- 
 respondence between the Indo-European languages which is at 
 once the evidence of their relationship and the basis of their 
 classification into families. The evidence is twofold, (a) in 
 vocabulary, (6) in grammatical structure. 
 
 (a) In vocabulary, it is easy to find in any two of these 
 languages numbers of common words, and to prove them all 
 related by proving each one related with each of the rest in 
 succession — Latin with Greek, Greek with Sanskrit, Slavonic 
 with German, and bo on. But for words identical in all or 
 nearly all branches of the family, we are restricted to such 
 special classes of words as numercds and pronouns. These 
 appear to have been less varied by multiplication and substi- 
 tution of synonymous terms than any other class of words — 
 except, perhaps, the terms indicating degrees of near relation- 
 ship, father, mother, daughter, brother, etc. ; and hence all the 
 Indo-European nations, however widely separated, and however 
 different in manners and civilisation, count with the same words 
 and use the same pronouns in individual address ; allowance, 
 of course, being made for the changes brought about by the 
 phonetic laws of individual languages. 
 
 (6) Stronger still is the evidence of correspondence in gram- 
 matical structure, as shown in a common system of word- 
 formation, declension, and conjugation. This portion of lan- 
 guage is that which, in the case of intermixture of languages, 
 by the adoption into one lauguage of terms belonging to another, 
 most resists any trace of intermixture. A foreign word admitted 
 to citizenship in another language is declined or conjugated on 
 the system of the language which has adopted it ', and the study 
 of language offers no trace of a mixed grammatical apparatus in 
 the same language. This being so, uniformity of grammatical 
 
It.} Ctassijication of Zanguages. 19 
 
 structure in a number of distinct languages must be one of the 
 strongest proofs of their substantial unity. 
 
 The table on p. 20 exhibits specimens of the correspondence, 
 in vocabulary and inflection, of the Indo-European languages : — 
 
 The most familiar illustration of a ' class ' of languages, and 
 on the whole the most instructive attainable example of dialectic 
 growth, is to be found in the modern ' Eomance ' or ' Eomanic ' 
 languages, so called as being all descended from the 'Lingua 
 Bomana,' spoken in the different provinces of the Eoman 
 Empire. In these we have not only a body of highly culti- 
 vated languages, each with its subsidiary dialects, and evidently 
 sprung from a common stock ; but we have also, what we have 
 not in the case of the great Indo-European group, the mother 
 language, the ' Ur-sprache,' from which they have aU sprung ; 
 and we can trace historically, with tolerable accuracy, the pro- 
 cesses of change and divarication which have produced them. 
 They all rose about the same period of the Middle Ages, out 
 of the condition of local jpatois, the result of illiterate provincial 
 corruptions of the Latin of ordinary popular pronunciation, 
 which even in classical times had differed in many respects 
 from the literary dialect of Rome, and had degenerated still 
 farther and faster when the decline of literature took away 
 the only check upon arbitrary pronunciation and erroneous 
 grammar. In the provinces upon which the Homan con- 
 querors imposed the use of the Eoman language', that lan- 
 guage was subject in its use to all the innovations produced 
 by ignorance, caprice, or the purely physical causes which dis- 
 pose the vocal organs of different nations to different sounds. 
 When therefore the various nationalities of modern Europe 
 
 ' In Britain, though a Roman province for 400 years, the Eoman 
 civilisation was too partial (being confined to the towns) to leave its 
 impress in the use of the Koman language, which in Graul and Spain 
 survived the conquest of those countries by Teutonic invaders. 'What 
 strikes us at once in the new England," says Mr. Green, ' is that it was the 
 one purely German nation that rose upon the wreck of Borne. In other 
 lands, in Spain, or Gaul, or Italy, though they were equally conquered by 
 German peoples, religion, social life, administrative order, still remained 
 Roman. In Britain alone Rome died into a vague tradition of the past.' 
 ('History of the English People,' ch. i. sect, ii.) 
 
 ca 
 
be 
 
 o 
 
 >!■. 
 
 
 
 I 
 
 O 
 
 o n 
 
 ^ a 
 
 o > 
 
 a I. 
 
 •5 
 
 aeS 
 
 
 -I 
 
 Ills I 
 
 ■ §11 j f I 
 
 11.! 
 
 ■See 
 
 t3 S S 
 
 1 
 
 3 
 
 
 t-H.S 
 
 ■o 
 a 
 
 o, 
 
 ■i 
 
 j«: M QD o 0.^ 
 
 -.:.Sa 
 
 v.t^'O > p>' N M n n.*3 
 
 £ S S 
 
 nil 
 
 ffi MIS >:S Gc oa n P E9 
 
 ■g 
 
 
 w 
 
 arj 
 
 a B 
 
 SS 
 
 ee.i 
 
 
 
 Jll? 
 
 as 35 
 
 J -s 
 
 i-3'i|s--^ll§^ 
 
 oa4a-«a(C<a 3 GO A N 
 
 ll-t=:i 
 
 
 £. 
 
 
 ag 
 
 III 
 
 ^ 03 ^ oQ cn CO 
 
 i_j ^ 43 QJ 03 fl) 
 
 a a ion's i'S a I 
 
 95-5 ■§ 
 
 'S ^ 
 
 ^'rtgts. 
 
 Ill 
 
 
 o ® 
 
 11 
 
 ■ 43 o ® © 
 
 ill |-|s till 1= 11 I I I 
 
 S*S Q."5 *-.(/ 
 
 -s.a.8 
 
 -3-t C.|:<g.| 
 
 I 
 
 ill 
 
 
 § a 
 
 .a.. 
 
 =1= 
 
 ! 9..S g-5 
 
 ?a 
 
 II aSl4 
 
 rHiNOS'^ioeOt^OOOiO 
 
 ua 
 
 .. u S 
 
 C "U ¥ . ^ 
 
 e as, _s 
 
 ^ 
 
 
 S S 3 »«.S 
 
 S3 O. 
 
Classification of Languages. 31 
 
 began to take shape andcoliesibn out of the chaos of the Middle 
 Ages, the dialects of the Latin-speaking peoples, in what had 
 been Roman provinces, had sufficiently diverged from each other 
 to form the starting-point of so many distinct languages, each 
 with its own national peculiarities, but with strongly marked 
 traces of their common origin — in vocabulary, in inflection, and 
 in grammatical structure — running through them all. The 
 following brief summary given by Professor Whitney ('Life 
 and Growth of Language,' pp. 183, 4), enumerates the Eomanie 
 group with sufficient precision : — ' Pragments of French are the 
 oldest, coming from the tenth century; its literature begins one 
 or two centuries later ; the earliest Italian, Spanish, Portuguese, 
 are from the twelfth, or hardly earlier. These four are the 
 conspicuous modern members of the group. But there was 
 also, in the eleventh to the fourteenth centuries, a rich lite-' 
 rature of the chief dialect of Southern France, the Proven9al, 
 which, except for a recent sporadic effort or two, has ever since 
 been unused as a cultivated tongue. There exists, too, in the 
 northern provinces of Turkey, in Wallachia and Moldavia, a 
 broad region of less cultivated Romanic speech, witnessing to 
 the spread of Roman supremacy eastward : it is destitute, 
 of a literature. Moreover certain dialects of southern (S. and 
 S. E.) Switzerland, are enough unlike Italian to be ordinarily 
 ranked as an independent tongue, under the name of Rhaeto,'- 
 Romanic or Rumansh.' This last is the dialect spoken in the 
 Grisons, and met with by English travellers in the Engadin. 
 
 Full materials for the philological study of the Romanic 
 languages are available in the ' Grammatik der JRomanischen 
 Sprachen ' of Professor Diez ', than which (says Professor Max 
 Miiller) ' nothing can be a better preparation for the study of 
 the comparative grammar of the -ancient Aryan languages.' 
 Subjoined is a table (compiled from this work) of comparative 
 forms in the six languages examined by Diez (viz. those above- 
 mentioned with the exception of Rumansch) ; these languages 
 
 ' A French translation of Diez' Grammar is better arranged than the 
 original, and contains additional matter supplied by Prof. Diez himself and 
 incorporated by the translators. 
 
22 Classification of Languages. [chap. 
 
 being placed in the order of (upon the whole) nearest resem- 
 blance to the Latin original. The evidence of relationship 
 needs no further comment ; and it must be borne in mind 
 that the relationship to each other of the various branches of 
 the Indo-European family is precisely analogous to that of these 
 Homanic languages to each other ; and that the evidence for it 
 in the two cases is different not in kind, but only in degree, in 
 proportion as the divarication of dialects has in the one case been 
 wider and longer, and historical observation of their phenomena 
 less attainable. 
 
 The transmutation of Latin into its derivative languages, 
 illustrated by the table on pp. 24, 25, exhibits certain features 
 of change from ' analytic ' to ' synthetic ' languages, the germ of 
 which can be traced even in classical Latin. We shall have 
 occasion, in treating of changes and modifications of sounds 
 (chap, iv), to notice the course of changes in the form of words, 
 which are constantly at work in every spoken language, and the 
 laws of phonetic change following the universally observed ten- 
 dency to secure ease in articulation at the expense often of 
 clearness. The formative suffixes of nouns and verbs — In- 
 flections — ^were the parts which were first affected. Originally 
 distinct elements with a meaning of their own, and a clear etymo- 
 logical connection with important elements of the language, 
 especially the pronouns, they have become in the language of 
 Cicero and Virgil * so far worn away as to appear in many cases 
 mere unmeaning terminations in the written language, and in 
 all probability still less recognisable in popular pronunciation of 
 everyday life. This rubbing away of the distinctive suffixes by 
 which grammatical relations were expressed, naturally led to 
 difficulties in retaining the right discrimination, e. g. of case 
 and tense, and to substitution of other expedients for expressing 
 such distinctions. 
 
 ' In my former edition I adopted the new-fasHoned spelling ' Vergil :' 
 and it,may appear a retrograde step to return to 'Virgil,' I am disposed, 
 however, to agree with Dr. Kennedy (Commentary on Virgil, Introd. 
 p. xxxviii), that while 'Virgilius' in Latin is indefensible, and ' Vergilius' 
 alone correct, it is unnecessary in oar Anglicised version of the poet's 
 name to abandon the fikmiliar ' Virgil.' 
 
II.] Classification of Languages. ■ 23 
 
 (i.) Case inflections gave way to prepositions. This process 
 is gradually at work in all languages, from the earliest stage, 
 which is the richest in inflections. As the anah/tical tendency 
 of language (getting rid otinflections and substituting separate 
 words for each part of a conception) advances, prepositions are 
 more and more developed to give precision to the obliterated 
 forms and extended meanings of case-endings. In Greek and 
 Latin (as the form of many of them indicates) they were origi- 
 nally adverbs, serving to define more clearly the meaning which 
 belongs to the case-ending by itself : and as the requirements of 
 language become more complicated, they become more and more 
 necessary to distinctness and accuracy in language. Hence they 
 are often used in prose where they would be omitted in poetry^; 
 and it is always the case which determines the meaning of the 
 preposition, not vice versa. 
 
 In modern Greek, and in the derivatives of Latin, prepo- 
 sitions have almost superseded cases; and the growing ten- 
 dency to use them appears in the New Testament, where they 
 are far more numerous than in classical Greek''; and in e. g. the 
 practice of the Emperor Augustus ', who made use of them in 
 order to speak as clearly as possible. "We are told that he pre- 
 ferred to say ' impendere in aliquam rem,' 'includere in carmine' 
 (instead of ' alicui rei,' ' carmine '). The tendency is foimd often 
 enough in earlier times, e. g. ' ad carnificem dare ' (Ter.), ' Ful- 
 gorem reverentur ah auro ' (Virg.). 
 
 (ii.) Prom the difficulty of retaining distinctions of tense 
 comes the use of active auxiliary verbs. In the passive voice 
 sum was always so employed ; and traces of a similar use e. g. 
 
 ' e.g. ab, ad, with ablat. or accus. of motion; or ah with 'ablativus 
 agentis.' 
 
 ' e.g. iaSiovaiv anb tuv xpixioiv, ' eat of the crumbs,' where classical Greek 
 would employ the 'partitive genitive' alone. See Farrar's Greek Syntax, 
 pp. 86, 87. 
 
 ' ' Praecipuam curam duxit sensum animi quam apertissime exprimere ; 
 quod quo faoilius exprimeret, aut nee ubi leotorem vel auditorem turbaret 
 et moraretur, neo prepositiones verbis addere, neque conjunctiones iterare 
 dubitavit, quae detractae afferunt aUquid obscuritatis etsi gratiam augent.' 
 Suetonius, ' Vita Octaviani,' Ixxxvi. 
 
 N.B. The last words of this quotation recognise the fact that analytic 
 languages gain in accuracy what they lose in conciseness. 
 
24 
 
 Classification of Languages, 
 
 [CIIAP. 
 
 TABLE OF CORRESPONDENCE BETWEEN 
 
 Words and 
 
 Forms 
 Compared, 
 
 Latin 
 Original, 
 
 I. Italian. 
 
 II. Spanish. 
 
 Article. 
 
 Nom. S. 
 Gen. S. 
 Nom. PI. 
 
 ille, -a (used in 
 late provincial 
 Lat. as Article) 
 
 il>, lo (masc), la (fern.) 
 del, dello, della 
 i. gli, le 
 
 el, la, lo (neut.) 
 del, dela, delo 
 los, las 
 
 Declen- 
 ' sion. 
 
 Nom. S. 
 Nom. PI. 
 
 r corona, -x\ 
 1 annu-s, -i > 
 (flos, flores 3 
 
 coron-a, anno, flor-e 
 coron-e, anni, flor-i 
 
 corona, afi-o, flor 
 coron-as, afi-os, flor-es 
 
 Numerals. 
 
 1 
 
 2 
 
 3 
 
 1000 
 
 unua, -a 
 duo 
 tres 
 mille 
 
 uno, una 
 
 due 
 
 tre 
 
 mille, pi. mil* 
 
 una, una 
 dos, duas 
 tres 
 mU 
 
 Personal 
 Pronouns. 
 
 Nom. S. 
 Nom. PI. 
 
 Gen. PI. 
 
 ego, tu, ille 
 nos. vos, illi 
 (nostrum 1 
 (illorum ) 
 
 io; tU! egli, ella 
 noi; voijeglino,elleno 
 
 denoi, voi; loro 
 
 y6; tu; el, ella,elIo 
 nos;vos;ellos, ellas 
 
 de 
 
 Verbs. 
 
 /Sing. 1 
 1 .. 2 
 « .. 3 
 »\ PI. 1 
 
 I ■' 2 
 \ .. 3 
 Imperfect 
 Prete-(1S. 
 rite t 2S 
 Plup.Subj. 
 Gerund 
 Infinitive 
 
 canto 
 
 canta-s 
 
 canta-t 
 
 canta-mus 
 
 canta-tis 
 
 cantant 
 
 canta-bam 
 
 canta-vi 
 
 cantasti 
 
 cantassem 
 
 cantando 
 
 cantare 
 
 cant-o, vend-o 
 cant-i, vend-i 
 cant-a, vend-e 
 cant-iamo, vend-iamo 
 cant-ate, • vend-ete 
 cdnt-ano, v6nd-ono 
 cantava, vend6va 
 cantai, vend6i 
 cantasti, vendesti 
 cantaasi, vendessi 
 cantando, vendendo 
 cantare, v6ndere 
 
 canto, vendo 
 canta-8, 'vend-es < 
 canta, vend-e 
 cantamos, vend-emos 
 cantais, vendeis 
 cantan, venden 
 cantaba, vendia 
 cants, vendi 
 cant-aste, vendiste 
 cant-ase, vendiese 
 cantando, vendiend6 
 cant-ar, vender 
 
 Auxiliary 
 Verbs, 
 
 used in ac- 
 tive voice 
 
 uaedin pas- 
 sive voice 
 
 habeo 
 habemus 
 sum 
 sumus 
 
 ho cantato 
 abbiamo cantato 
 sono stato 
 Siamo state 
 
 he cantado 
 hemos cantado 
 soy cantado 
 somos cantados 
 
 1 il, lo are the two syllables of ills. 
 
n.] 
 
 Classification of Languages. 
 
 ^s 
 
 THE 'EOMANIC LANGUAGES DESCENDED FROM LATIN. 
 
 III. Portuguese. 
 
 IV. Provenfal. 
 
 v. French. 
 
 VI. Wallachian. 
 
 (masc), a (fern.) 
 do, da 
 OS, as 
 
 lo, la 
 del, dela 
 li, las 
 
 le, la (OldF.li) 
 du, de la ( „ del) 
 les, les 
 
 -1, -le (suffixed) 
 a*— lui „ 
 
 1 
 
 coro-a, anno, flor 
 coro-as,anno-s,flore-s 
 
 coron-a,ans, flor-s 
 coron-as, an, flor 
 
 courcinne, an, fleur 
 couronne-s, an-s, fleur-s 
 
 coron-e, an, floare 
 corone, an-i, flor-i 
 
 hum, huma 
 dous, duas 
 
 mil 
 
 uns, una 
 dui, duaa 
 trei, tres 
 mil, pi. mila 
 
 -un, une (Old P. uns) 
 deux ( „ dous) 
 trois ( „ treis/ 
 miUe < „ mU) 
 
 un, una (o) 
 
 doi (doo, doao) 
 
 trei 
 
 mie, pi. mil 
 
 eu; tu; elle, ella 
 nosivos; elles, ellas 
 
 de ; deUes,dellas 
 
 ieu; tu; elh, elha 
 nos; V0S!elhs,elhas 
 
 de J delor 
 
 je; tu; U, elle 
 nous; VOUS; lis, elles 
 
 de } d'eux,d'elles 
 
 eu : tu ; el, ia 
 noi; vol; ei, ia-le 
 
 alnostru,vostru ; alor 
 
 canto, vendo 
 cantas, vendes 
 canta, vende 
 cantamos, vendemos 
 cantais, vendeis 
 cantao, vendia 
 cant-ava, vendia 
 cant ei, vendl 
 cant-aste, vendeste 
 cantasse, vendesse 
 cant-ando,vend- endo 
 cant-ar, vender 
 
 chant-i, vend-i 
 chant-as, vend-es 
 chant-a, vend 
 chant- am, vend- (Sm 
 chant-atz, vend-etz 
 chdnt-an, v6nd-on 
 oliantava, vendia 
 chantei, vendei 
 chant-est, vend-est 
 chant- es 2, vend-es 
 chant-an, vend-en 
 chant-ar, vend-re 
 
 chante, yend-s 
 chant-ea, vend-s 
 chant-e. vend 
 chant-ona, vend-ons 
 ohantez, vend-ei 
 chant-ent, vend-ent 
 chantais, vend-ais 
 chant-ai, vend-is 
 chant-as, vend-is 
 chant-asse, vend-isse 
 chant ant^ vend-ant 
 chant-er, vend-re 
 
 c^nt, vind 
 CTjnt-zi, vinz-i 
 cunt-^, vind-e 
 cunt^m, vind-em 
 cunt-atzi, vind etzi 
 ct^nt-^, vind 
 cuut-&m, vind-eam 
 ci^nt-ai, vind-ui 
 ci^iit-asi, vind-usi 
 cynt-4sem,vind-iisem 
 cynt-i^nd, vind-i^nd 
 ci^nt-a, vind-e 
 
 tenhc' cantado 
 temos cantado 
 so cantado 
 somos cantados 
 
 ai chantat 
 
 sui chantatz 
 sem chantat 
 
 ai chants 
 avons chant6 
 suis chants 
 sommes chant^s 
 
 am cijntat 
 am c^nt 
 
 1 i. e. temo. Spanish also uses the corresponding ten^o as an auxiliary. 
 
 2 A fuller fonn is retained in 2 sing, chantesses, and in plur, 
 
 3 In phrases like en voyant. 
 
 * The preposition before, the article after the noun. 
 
26 Classification of Languages. [chap. 
 
 of dare are found in phrases like inventwm daho, Ter. And. iv. 
 I. 59, vasta daho=vastabo, Virg. Aen. i. 63; ix. 323. Habere 
 and tenere {avoir, avere ; Spanish tengo, tenere) must have been 
 so used in the provincial speech of the later Empire ; we have 
 perhaps an anticipation of this in the classical exjpertum, cogni- 
 twm Tiabeo. The passive auxiliary construction with sum, etc. is 
 obtained by an easy resolution of any tense in that voice : but 
 the propriety of the active habeo or ieneo is not so obvious. It 
 may, however, have been extended by analogy from cases in 
 which such analysis was correct to others in which it could 
 not be so employed with strict accuracy. 
 
 (iii.) Next to these changes, foimded on pronunciation and on 
 the substitution of prepositions and auxiliary verbs for noun and 
 verb inflections, the usage of the definite and indefinite article 
 seems the most considerable step in the transmutation of Latin 
 into its derivative languages. The development of the definite ar- 
 ticle from a demonstrative pronoun, which is seen in the Romanic 
 forms il, lo, etc., derived from Latin ille, took place in Greek at 
 an early period, but within historical observation ^ ; for we see 
 it beginning in the Homeric poems in the use, beside the demon- 
 strative OS, of a parallel form 6 also demonstrative, but in certain 
 collocations suggesting the later use as definite article, e. g. v 8' 
 op a/iei^cTo UaiWas 'hBrjinj, etc. The Greek language thus gained 
 an important element of precision, and facility for the combi- 
 nation and grammatical handling of abstract ideas, e. g. by the 
 article with infinitive or neuter adjective ; and though little or 
 no attempt seems to have been made in the literary dialect of 
 Home to create a corresponding means of precision by an 
 analogous employment of the Latin demonstrative pronoun, 
 there are not wanting signs that the necessity for it was felt 
 and partly acted upon in popular language, by the employment 
 of ille and unus with the force of a definite and indefinite article 
 respectively". "Were this not the case, the evidence of the 
 
 * On the history and usages of the Greek article see Curtius' Greek 
 Grammar, §§ 365-391 ; Clyde's Greek Syntax, §| 3-9. The latter book 
 is a very valuable aid to the student of Greek grammar. 
 
 ' The theory of grammarians in this matter seems to have gone contrary 
 
II.] Classification of Languages. 27 
 
 Homauic languages would be sufficient proof that, at all events 
 in the provincial idioms of the later Empire, this usage had 
 become more or less established. The same development of 
 definite article from demonstrative seems to have taken place 
 in the Teutonic languages; for in German dm- (like 6y, 6) is 
 demonstrative, relative, and definite article ; and in English 
 that and which are often interchangeable. 
 
 For further suggestions upon the relation of the Komanic 
 languages to Latin, the reader may consult Max Mailer's Lec- 
 tures, Series I. Lecture v. and Hallam's Middle Ages, chap. IX. 
 part I. 
 
 to the practice of those who spoke and used the language. Quintilian 
 (I. 0. i. 4. 19) says, 'Noster sermo articulos non desiderat;' and Scaliger 
 called the article ' otiosum loquacissimae gentis instrumentum,' ' articulus 
 nobis est nullus et Graecia superfluus,' 
 
CHAPTEK III. 
 Classification op Sounds. 
 
 Principles The division of sounds and of the letters representing them in 
 
 of phonetic _ . 
 
 change. the alphabets of different languages, according to the organs of 
 
 the human voice by which the sounds are produced, is the basis 
 upon which enquiries into the mutual connection of languages, 
 and all etymology, must ultimately rest. In tracing the original 
 form or the common element of words or their inflections in 
 one or more languages, we are retracing the course of ' phonetic 
 change;' the changes i.e. in the sounds and the letters repre- 
 senting them, by which, while languages are in daily use as 
 media of oral communication, variety or degeneration from 
 simple and primitive forms have been produced. The principle 
 of this phonetic change is the endeavour, conscious or uncon- 
 scious, to secure ease of a/rtieulation. ' All articulate sounds are 
 produced by efibrt, by expenditure of muscular energy in the 
 throat, lungs, and mouth. This effort, like every other that 
 man makes, he has an instinctive disposition to seek relief from, 
 to avoid : we may call it laziness, or we may call it economy : 
 it is in fact either the one or the other, according to the circum- 
 stances of each particular case. It is laziness when it gives up 
 more than it gains ; it is economy when it gains more than it 
 abandons.' 
 
 Ease of articulation is secured in the majority of cases by 
 substituting a sound easier to pronounce for one which is found 
 difficult — a weaker for a stronger sound : and (with some few 
 
.Classification of Sounds. 2<) 
 
 exceptions) it is a safe rule in etymology that harder sounds are 
 pot derived from easier, nor a word which has retained a strong 
 sound from one which exhibits a correspondingly weak sound ; 
 nor, therefore, a language in which iadividual forms retain 
 strong sounds from a language whose corresponding forms re- 
 tain weaker sounds. Thus (to take a simple instance) such 
 forms as siZi/a, sus, video, vinum beside vXr], ()s, JSfic, olvos, go 
 far to prove what has already been demonstrated upon the 
 evidence of inflections (above, p. 17), that Latin cannot have 
 been derived from Greek, having retained in these words the 
 sounds s and v {F), which Greek has lost, or represents only by 
 an aspirate. 
 
 But what are hard or strong, and easy or weak sounds 1 and 
 how is the relative strength of sounds determined ? Obviously 
 by the physical conditions of their utterance. Sard sounds are 
 those which require greater physical effort on the part of the 
 organs of speech, easier sounds those which require less effort. 
 The table given on p. 31 exhibits the sounds arranged according 
 to the physical conditions of their production : and without a 
 ininute investigation of those physical conditions (for which the 
 student is referred to Max MUUer's Lectures, Series II. Lect. iii. 
 on ' The Physiological Alphabet'), a brief statement of them is 
 joecessary for the explanation of the terms employed. 
 
 The material of speech is breath, i. e. a continuous stream of Physical 
 ?iir from the lungs, modified by the different positions, or the human 
 interrupted and compressed by various actions of the uvula, 
 tongue, palate, teeth, and lips, which thus become organs of 
 voice*. If the glottis, or aperture through which the breath 
 
 ' For a fuller description of the instrument? of the human voice, see 
 Max Miiller's Lectures, Series II. Lect. iii. (pp. 109-1 14, 2nd ed.), and 
 (Farrar's ' Chapters on Language,' oh..vii. pp. 84, 85 : ' When we are speak- 
 ing we are in reality playing on a musical instrument, and a more perfect 
 instrument than ever was invented by man.' ' The larynx, with its carti- 
 lages and muscles, forms, in point of fact, a combination of musical instru- 
 ments ; it is at once a trumpet, an organ, a hautboy, a flageolet, and an 
 Aeolian harp. The air passing upwards and downwards through the 
 larynx and trachea forms its analogy with the wind-instruments ; the 
 'vibration of the chordae vocdles, its resemblance to the stringed.' See 
 ;also Dr. Carpenter's 'Animal Physiology,' p. 538 ; and Whitney, 'Life and 
 Growth of Language,' ch. iv. p. 59. 
 
3b Classification of Sounds. [chap. 
 
 passes from the trachea gr windpipe, be fully open, what passes 
 into the mouth is mere breath, made afterwards into sound by 
 the organs of the mouth. If however two ligaments at the 
 sides of the glottis, called chordae voeales, approximate to each 
 other so as to narrow the glottis, and vibrate as the breath 
 passes through, this vibration changes the breath into voice — 
 makes it vocal sound. And accordiug as mere breath, or vocal 
 sound, is emitted from the windpipe, the same position of the 
 organs of the mouth gives a different result. If it be only 
 breath that is checked or modified by their contact or approxi- 
 mation, the sound produced will be what is variously called 
 tenuis, 'hard' or 'surd:' if, on the other hand, voice or vocal 
 sound be checked by contact or approximation of the organs, 
 the sound produced will be media, ' soft' or ' sonant.' 
 
 Sounds are divided generally into Votoels and Consonants. 
 The physiological difference in their formation is as follows : — 
 
 Modification of the stream of vocal sound, without interrup- 
 tion or compression by the organs of the mouth, produces 
 Vowels {voeales, (jxavfievra), so called because they have a sound 
 of their own, being various modifications of the vocal sound 
 produced by the 'chordae voeales.' All vowels, therefore, are 
 'soft' sounds. 
 
 Interruption of breath or voice by complete contact, or 
 compression by approximation of the organs, produces Con- 
 sonants {con-sonantes, (rviKJxova), so called because they have 
 no sound of their own, but must be accompanied with a vowel 
 sound'. (Thus, in the Sanskrit character the vowel a is never 
 written after a consonant, because a, the primitive vowel sound, 
 is supposed to be inherent in every consonant.) Consonants are 
 either ' soft' or ' hard,' tenues or mediae. 
 
 The subjoined table illustrates the classification of Sounds as 
 applicable to Greek and Latin : — 
 
 (The Greek and Latin characters are given). 
 
 ■ The Arabic grammarians call a vowel motion, and a consonant u, 
 harrier, because in forming vowels the voice is not interrupted, whereas 
 in forming consonants it is stopped at certain iixed positions. 
 
m.] 
 
 Classification of Sounds. 
 
 31 
 
 TABLE IN ILLUSTRATION OF THE CLASSIFICATION 
 OF SOUNDS. 
 
 
 Mutes 
 
 'Explosive' 
 
 Sounds. 
 
 } 
 
 1 
 1 
 
 Semivowels : 
 'Fricative' Sounds. 
 
 Vowels. 
 
 Tenuis. 
 
 Media. 
 
 Spirants. 
 
 Liquids. 
 
 Pri- 
 mary. 
 
 Secondary. 
 
 Tenuis. 
 
 Media. 
 
 Guttural 
 Palatal 
 Dental 
 Labial 
 
 ckq 
 
 7 
 
 g 
 
 X 
 
 7* 
 
 h 
 
 ■ 
 
 
 a 
 a 
 
 !■■■ 
 
 03 
 \ 
 ( 
 
 
 
 
 
 
 i(i) 
 
 
 i 
 
 T 
 t 
 
 S 
 d 
 
 9 
 
 V 
 n 
 
 a 
 
 s 
 
 z 
 
 1 1- 
 
 
 IT 
 
 P 
 
 18 
 b 
 
 <t> 
 
 1* 
 m 
 
 f 
 
 F 
 
 V 
 
 
 V 
 
 u 
 
 (/and w are sometimes classed separately as ' Labio-dentals,' 
 being formed by contact of under lip and upper teeth.) 
 
 N.B. In Latin, c, g represent the hard sound of <, 7 : i (J), 
 semivowel, the sound of English y in you: v (and Greek F) 
 that of English w. 
 
 The vowels in Greek and Latin are pronounced nearly as in 
 Italian. (Cp. the table of the Greek and Eoman alphabets, 
 p. 42.) 
 
 [Sanskrit, besides the sounds expressed in Greek and Latin, 
 has letters expressing — 
 
 ' Palatal' mutes, ten. and med. [k g] ; spirant [s] ; nasal [n] 
 
 'Cerebral' „ „ „ [td]; „ [sh]; „ [n] 
 
 (These sounds are varieties of guttural and dental respect- 
 ively.) 
 
 A complete set of aspirated mutes, tenues, and mediae; e.g. 
 kh, gh ; kh, gJi ; etc. 
 
 Short and long sounds of all the vowels, + the dental vowel 
 sounds ri, li.] 
 
 • e.g. in «77J», Syyt\o$. 
 
33 Classification of Sounds. ^ [chap. 
 
 SoTofCon- Consonants are classified' (see the Table, p. 31) — 
 
 L Mutesand !• -^7 the Completeness or incompleteness of contact of the 
 
 vowik 'vocal organs.' 
 
 a. Mutes (acjiava, mufae), where there is a complete interrup- 
 tion of the passage of the breath or vocal sound. These 
 are 'Consonants' proper, having no sound of their own, and 
 depending for articulation upon the vowel sound which follows 
 when the stream of breath or vocal sound is released from 
 the ' check' or interruption. They are sometimes called 
 'Momentary' or 'Explosive' sounds [kg, td,pbj. 
 
 b. Semivowels^ {fiiil<jiava, semi-vocales), where the stream of 
 vocal soimd is not interrupted by complete contact, but only 
 compressed by approximation of the 'vocal organs,' so that a 
 continuous sound is heard from the friction of the breath or 
 vocal sound against the partially closed organs. They are 
 sometimes called 'Continuous' or 'Fricative' sounds [s, «, I, r, 
 f, V, etc. J. 
 
 ^dMe^ae ^^' ^^ *'^® accompaniment or absence of vocal soimd. (See 
 above, pp. 29, 30). 
 
 a. Tenues^ (^CKa, 'voiceless:' also called 'sharp,' 'hard,' 
 ' surd'), when the contact or approximation of the organs takes 
 place with the voeal chords (see above, p. 29) wide apart, so 
 that only a whisper takes place [k, t, p, s,/]. 
 
 b. Mediae* {jitaa, 'voiced,' also called 'flat,' 'soft,' 'sonant'), 
 when the contact or approximation of the organs takes place 
 with the vocal chords close together and vibrating so as to cause 
 sound^ during the approximation or contact \_g, d, b, z, v, etc.]. 
 
 • The teacher will find the comprehension and recollection of these 
 claBsifications and the terms employed much facilitated by oral illustration, 
 pronouncing himself, and asking his pupils to pronounce, each sound as it 
 is mentioned. 
 
 ^ For the more limited use of this term, see below, p. 35, note 2. 
 
 ' K, T, IT were called (tiAd ypiniuiTa ('bald,' slight, or thin letters) by 
 the Greek grammarians in distinction from the. aspirates x, S. <t>^ which had 
 a rough or shaggy sound. Hence ^|/lKSjs ypdtpuv ^to write with a tenuis 
 instead of an aspirate (fiiirvs for ^ixpva), Ath. 369 B. 
 
 * The mediae {iiiaa) 7, 5, J3 were so called because they were pronounced 
 by the Greek grammarians with more aspiration than the tenues and with 
 less than the aspirates. 
 
 ° On the general causes of the distinction between iermes and mediae. 
 
in.] Classification of Sounds. 33 
 
 in. By the part of the mouth at which, and the ' vocal ni. Guttu 
 organs' between which the contact or approximation takes 
 place. 
 
 a. Guttural, by the back or soft palate (uvula) and root of 
 the tongue \k, g\. 
 
 b. Palatal''; by the middle or hard palate and the tongue 
 (i.e. the guttural 'check' or contact pushed a little further 
 forward). 
 
 c. Dental, by the upper teeth and front part of tongue [t, d]. 
 
 d. Lakial, by the lips [p, 6], or under lip and uppers teeth 
 [/, v\. The latter are sometimes classed separately as Labio- 
 dental. 
 
 Somewhat outside of I. and III. come Nasals and Liquids. 
 
 Nasals are a variety of Explosive Mediae: i.e. when the Nasals. 
 organs are in position for pronouncing g, d, h, but the stream 
 of breath passes into the nose, ng, n, m are respectively pro- 
 duced. Accordingly,, if we try to pronounce n or m either 
 holding the nose, or when it is stopped by a cold so that the 
 breath cannot pass that way, the result is the original sound 
 of d or b, e.g. Tnoon becomes hood^. 
 
 see Prof. Helmholtz, as quoted by Max Miiller, Lectures, II. iii. p. 131, 
 2nd ed. Prof. Whitney insists upon the use of the terms 'surd,' 'sonant;' 
 see 'Life and Growth of Language,' p. 63. 'Hard' and 'soft' are more 
 familiar in English writers on language. 
 
 * The various consonantal sounds which in Sanskrit and other languages 
 are called palatal are formed by placing the tongue in a position inter- 
 mediate between the guttural and dental contact, and are modifications, 
 sometimes of gutturals, sometimes of dentals. In Sanskrit they approach 
 nearer the former, and are often represented, the tenuis by English ck (in 
 oJiurch, or Italian cielo), the media by j (i.e. as in our pronunciation of 
 GermoM, Oeorge). Many Sanskrit scholars, howerer, prefer to denote the 
 palatal series by the guttural signs, k, hh, g, gh, modified either by ' 
 \h', k'h) or by difference of type (k, fc; g, g), because this helps them to show 
 the easy transition between e. g. nom. ^T^ (vak) and ^^ (vftft) the stem of 
 the oblique cases. 
 
 ^ The following stanza from a jeu ct esprit, entitled ' The Lay of the 
 Influenzed,' may serve as an illustration of this : — 
 'Dever bore bedeath the bood 
 
 Shall byrtle boughs edtwide ; 
 Dover bore thy bellow voice 
 Bake belody with bide.' 
 This is incorrectly termed 'speaking through the nose.' A person who 
 has a cold ought to speak through his nose, but cannot do so, in pro- 
 nouncing m or M : and therefore he sounds nearly h ov d in. attempting 
 to pronounce the nasals m, a. 
 
 S 
 
34 
 
 Classification of Sounds^ 
 
 [chap. 
 
 Liquids. Liquids ('lingual' sounds or 'trills') are caused by the breath 
 
 passing over the sides of the back of the tongue [f\, and over the 
 tip of the tongue [r]. They may be classed with Semivowel 
 [Fricative] sounds, to which they have most affinity. 
 
 Aspirates. Aspirates are variously classed with 'explosive' and 'frica- 
 tive' sounds. The sound denoted by Ji {spiritus asper''-) is a 
 mere expulsion of breath^, unchecked by the vocal chords, which 
 remain wide apart without vibrating so as to produce vocal 
 sound. This 'breathing' (to adopt the term familiar in Greek 
 grammar), when it follows an explosive consonant, gives siich 
 sounds as Greek x Qa + Kj, 6 {t+K), (p {p + K). In pronouncing 
 the tenues k, t, p, the vocal chords are apart and in a natural 
 position for aspiration: but with the mediae g, d, h they are 
 close together, and not in position for aspiration. Hence gh 
 ig+h), dh (d + h), hh (b + h) are more difficult sounds, which 
 perhaps existed in the earliest forms of Indo-European speech, 
 but have only found expression in the Sanskrit and (to a certain 
 extent) in the Keltic languages. 
 
 The shght sound or 'breathing' heard before any vowel, and 
 best caught when two vowels come together (e.g. go over), is 
 rarely expressed by any sign, except in Greek by the spiritus 
 lenis or ' soft breathing.' 
 
 If the breath emitted for spiritus asper or lenis be modified 
 by certain narrowings of the mouth forming barriers which hem 
 it in, various distinct soimds are produced. Eight such ' bar- 
 riers,' with corresponding modifications of the spiritus asper and 
 lenis, are enumerated by Professor Max Miiller ; of which only 
 those for which signs are given in our table of sounds are here 
 given, viz. 
 
 ' The distinction between ^iritus asper and lenis is regarded by Prof, 
 Max Miiller as that which is denoted in consonants by the terms tenuis and 
 media, the glottis being in one case open, in the other closed. 
 
 ^ Others, however, regard A as a genuine consonant, produced very near 
 the glottis, so that it combines very readily with a following vowel, and 
 seems to be produced in the same act of enunciation. Mr. Peile, holding 
 this view, thinks 'that there may be a soft h which dififers from the 
 ordinary A almost as much as any soft consonant from the corresponding 
 hard ; and that this soft h differed infinitesimally (if at all) from the breath 
 heard after the momentary sound in the original aspirates {gh,, dh, bh).' — 
 'Introduction,' pp. 69-73. 
 
III.] Classification of Sounds. ^^ 
 
 1. The barrier produced by advancing the tongue to\^fards the 
 teeth modifies spiritus asper into s, spvritus lenis into z. 
 
 2. If the lower lip be brought against the upper teeth, the 
 barrier produced modifies spiritus asper into /, spiritus lenis 
 into V, as heard in English live, halve. Hencey, v are sometimes 
 called 'Labio-dental' sounds. 
 
 3. If the lips be slightly contracted and rounded, spiritus 
 asper becomes wh in wheel, which; spiritus lenis the English 
 w, which is apparently represented by Greek F and Latin v. 
 
 These sounds, s, z,f, v, etc., are called spirants. This name and Spirants. 
 the physical fact which it denotes (that the sounds so called are 
 modifications of the ' breathings'), will at once suggest the 
 explanation of such phonetic varieties as sedes, e8os ; sus, is ; 
 suh, viro; or the correspondence' of Latin y to various aspirate 
 sounds in Greek, e.g. tO' (j), x {x^^Vi f^^'i to 6 {6rjKvs, femina, 
 6rjp, ferd) ; and to ', originally F, in ptyea, frigus. 
 
 Vowels and Diphthongs ^ 
 
 I. The three primitive vowel-sounds are A, 1,11, (pronounced Vowels ■. A, 
 as in Italian). Of these a is formed nearest to the guttural 
 point of contact (with the lips opened wide); i nearest to the 
 palatal; u nearest the labial contact, the lips approaching 
 each other. 
 
 i and u pass into the cognate consonantal (or semivowel") 
 sounds of y and v (w). 
 
 a can pass into no cognate consonantal sound : it is the vowel 
 par excellence, occurring in Sanskrit', and probably in the 
 
 ' For a more elaborate analysis of vowel-sounds than appears necessary 
 to give here, see Peile, Introd. pp. 90-100 (3rd ed^ : also BeJl, 'Principles 
 of Speech,' and ' English Visible Speech for the Million.' 
 
 ^ 'Semi-vowel' is here used in the limited sense, which often attaches 
 to it, of the consonantal sounds of i, u. The reader will take note that it 
 has been applied above (p. 32) in a wider sense to the whole class of 
 'Fricative' consonants, as distinguished from Mutes or Consonants proper. 
 To avoid confusion it would be better either to describe the y and w sounds 
 as i and « consonantal, or to give up the wider application of the term to 
 Fricative consonants, but the inconsistency of usage is too confirmed. 
 
 ' In Sanskrit o following a consonant is never written, because it is 
 supposed to be inherent in every consonant (e.g. patara is written ptr) ; 
 and the Sanskrit alphabet, which has two separate characters for each 
 vowel-sound according as it is initial or in the middle of a word, has no 
 character for S, medial. 
 
3 6 Classification of Sounds. [chap. 
 
 earliest form of Indo-European language, much oftener than 
 i or u. 
 E, O. 2. H and are phonetic variations of the A sound. If we 
 
 compare kindred words in Sanskrit, Greek, and Latin, we find 
 that Sanskrit a is represented by a, e, o i^ Greek, by a, e, 5^ in 
 Latin, e.g. Sanskrit navas, Greek ve(F)os, Lat. novus=:i(n,ovos) ; 
 
 pitS.(r), Ttarfip (=:7raTf(j-s), pater j bharS.nai, (jjepa ( = ^e/ja>-/ii), 
 
 /ero. In some oases the variety secured by this weakening of a 
 has been turned to account, to indicate differences of meaning ; 
 thus Sanskrit padas, which is gen. sing., nom. and ace. plur. 
 of pad, a foot''', becomes in Greek iroSos, ?ro8ej, iroSas ; the 
 Greek thus gaining in distinctness what it loses in strength 
 of sound. 
 
 Diphthongs. 3. When two vowels follow one another so rapidly as to melt 
 into one sound we get a diphthong. Of the primary vowels a 
 alone can thus form the basis of a diphthong ; for i and u, if a 
 vowel-sound follows, pass into the 'semivowel' sounds of y and 
 V. e and 0, being varieties of a, can also serve as ' diphthongal 
 bases.' We thus get as diphthongal sounds, in Greek ai, av, ei, 
 fv, o», ov; in Latin ai, au, ei, eu, oi, ou: though, for reasons 
 which will appear afterwards, the Latin diphthongs, with the 
 partial exception of au and eu in a few words, became weakened 
 to the simple sounds cb (e), o, t, u, ce (e), U, and we must go back 
 to the archaic remains of the language for such forms as aidilis, 
 deieere, foidus, joudex. 
 
 Original a. Another vowel-sound is sometimes added, viz. the inde- 
 
 finite or neutral sound (' original vowel,' ' Ur-laut,' ' Ur-vocal '), 
 
 ' The vowela are originally short in quantity (as e.g. in most roots), 
 lengthening being generally the result of 'vowel intensification,' as in duco 
 (root due-), or contraction, as amas^ama-is. Vowela which are naturally 
 long must be distinguished from vowels which are naturally short, but 
 long by position, e.g. drma {&), nox (8). In speaking of vowel-sounds 
 generally we mean (unless otherwise specified) a, e, i, S, u. 
 
 ' The accent would vary in Sanskrit : but the point here la to note the 
 uniformity of the vowel in the three forms. Accentuation in Sanskrit 
 is only marked in the Yedas ; but it is sometimes desirable to mark it in ' 
 transliteration, for the light which it throws upon apparent anomalies of 
 Greek accentuation. See, for iiistance, Max MiUler, ' Chips from a German 
 Workshop,' vol. iv. p. 34, on intvai, Uvai : and below, chap, vi, on the 
 Vocative Singular. 
 
 voweJ 
 
in.] Classification of Sounds. 37 
 
 variously defined as ' the natural vowel of the reed/ ' the voice 
 in its least modified form,' etc. This is the sound heard in such 
 words as hut, dust ; and it has been said that in such words as 
 e. g. spurt, assert, bird, fatal, dove, oven, double, blood, but one 
 and the same indefinite vowel-sound is heard. However this 
 may be, there is no doubt an indefinite sound to which un- 
 accented vowels in most modern European languages have a 
 tendency to return, e. g. in the last syllable of beggar, nation, 
 jPaddington, Geima,n lieben; or the first syllable of French tenir. 
 Physically, it appears to result from leaving the tongue in its 
 most natural position, opening the mouth easily and emitting 
 vocal breath ; and it approaches the. sound of all the vowels. 
 It is this indistinct vowel, combined with r and I, that produces 
 the Sanskrit vowel-sounds ri and li (ri, li). 
 
 It should be borne in mind that sounds are distinct from Relation of 
 the signs used to represent them, i. e. the letters. The number letters. 
 of possible articulate sounds is greater than any nation ever 
 employs ; and the ' alphabet ' of some languages will express 
 sounds whicli that of others does not. Again, the use of letters 
 in time reacts upon sounds. They do not always fit each other 
 exactly to start with ; and while pronunciation is always 
 changing, spelling in a literary language becomes more or less 
 fixed. Thus in time letters become symbols of other sounds 
 than those proper to or originally denoted by them, and carry 
 their new sounds into other words or other languages. For 
 example, in the Eoman alphabet, which is common to most 
 nations of modem Europe, c, g, representing to a Eoman of the 
 classical period the hard sound of k or Greek y^ before all 
 vowels, in the pronunciation of the later Empire and in the 
 languages of modern Europe came to signify difierent sounds 
 before the vowels i, e ; and these new sounds are carried back by 
 each nation into their pronunciation of classical Latin, leading 
 to such anomalies as the identical pronunciation of sectis and 
 caecus, or the difierent pronunciations of locus, loci, loco, parts 
 
 ■ The evidence for this statement as to the pronunciation of c, g will be 
 found summarised in Boby's ' Latin Grammar,' vol. i. Preface, pp xliii-lii, 
 or Wordsworth's 'Fragments and Specimens,' Introd. ch. iii. 5§ 22-28. 
 
38 Classification of Sounds. [chap. 
 
 of the same word. . Again, j and v in Latin, the modem repre- 
 sentatives of consonantal i and u, have acquired, and carry back 
 with them into the modern pronunciation of consonantal i and u 
 in Latin words, quite different sounds from those of our y and w, 
 which are in reality much nearer representatives of the sounds 
 in question. Whatever, therefore, may be the practical value to 
 modern nations, in reading or pronouncing a dead language, 
 of attempts to reproduce the ancient pronunciation, it is of the 
 utmost importance, for philological and etymological enquiry, to 
 realise as accurately as we can what sounds, in the mouths of 
 those who spoke the Greek and Latin languages, are represented 
 by their written character ; and this not only for the philology 
 of those languages, but for that of all the modern languages 
 which, as we have seen, are connected with them. 
 s™skrit The only people who have ever attempted to express in their 
 
 written character almost every known gradation of sound, are 
 the Hindus, those who employed the Sanskrit language. The 
 Sanskrit alphabet has fourteen vowels, each (except a) with two 
 symbols, one initial, the other medial ; thirty-three simple con- 
 sonants ; and upwards of 400 or 500 compound consonants, of 
 which 133 are given in Professor Monier Williams' Sanskrit 
 Grammar as ' the more common ' of such consonants : while 
 Prof. Max Muller (Sanskrit Grammar for Beginners) specifies 
 257 compound consonants. Sanskrit, in fact, in its whole struc- 
 ture, is an elaborate process of combining letters according to 
 fixed rules. ' Its entire grammatical system, the regular forma- 
 tion of its nouns and verbs from simple roots, its theory of de- 
 clension and conjugation, and the arrangement of its sentences, 
 all turn on the reciprocal relationship and interchangeableness 
 of letters, and the laws which regulate their internal combina- 
 tion^' These laws, too, are the key to the influence which 
 Sanskrit has exercised upon the study of Comparative Philology. 
 That influence is due, not to its being (as is sometimes said) an 
 older language — though approaching on the whole nearer to the 
 primitive type whose existence we infer from a comparison • of 
 the various branches of the Indo-European family — but to the 
 
 ' Monier Williams' ' Sanskrit Grammar,' Preface to 2nd ed. p. xv. 
 
III.] Classification of Sounds. 39 
 
 fact that its elaborate system of phonetic combination of sounds 
 supplies illustrations for the different phonetic rules -yvhich de- 
 termine the variation, in different languages, of the elements 
 common to all. Owing to the transparency of its construction, 
 the nicety of its laws and its great antiquity in many respects, 
 and especially that of its vowel system, Sanskrit was soon found 
 to be more adapted than any other language to open men's eyes 
 to the nature of the connection of all the sister languages : and 
 in the first rejoicings of the students of language over its dis- 
 covery, its importance was for a time overrated. ' The prepos- 
 terous idea that Sanskrit must have preserved in every case the 
 oldest form ' is now however generally discarded ; and those 
 philologists whose labours rest upon the most thorough know- 
 ledge of Sanskrit, are the first to allow that even in its sounds 
 there are weaknesses and corruptions peculiar to it which pre- 
 vent it fii'om serving in all cases as the starting-point for com- 
 parison, and even send us to other languages to recover the 
 primitive form. Thus (to quote the remark of Curtius '), ' now 
 that this language has for a long time served exclusively to 
 throw light on others, the light begins to shine back from the 
 other languages upon Sanskrit.' With this limitation, however, 
 the pre-eminence of Sanskrit as the central point in the study 
 of Comparative Philology may remain accepted ; for (to quote 
 again from Professor Curtius), ' the exuberance of the old Indian 
 literature ; the antiquity of its most revered monument the Rig- 
 veda; the perfection of its alphabet ; the remarkable acuteness 
 and diligence of its native grammarians, who have prepared the 
 most valuable assistance for the study of Etymology, if only by 
 their discovery of the conception of roots and their careful index 
 of roots ; all these are claims on the part of Sanskrit, which 
 only during the last half-century has become the field of such 
 fresh and important investigations, to retain permanently the 
 prominent position of importance for the study of the whole 
 Indo-Germanic (Indo-European) stock of languages'*.' 
 
 ' 'Principles of Greek Etymology,' Introd. § 5 (p. 37, English trans- 
 lation). 
 2 Ibid. p. 30. 
 
APPENDIX TO CHAPTER III. 
 
 The Geeek and Koman Alphabets. 
 
 Greek __ A. Cfreeh Alphabet. It is universally admitted that the 
 Greeks learnt the art of writing from the Phoenicians, with 
 whom, as the chief traders of the Mediterranean, they were 
 brought into contact at an early epoch of their national his- 
 tory. In adopting the Phoenician alphabet they seem to have 
 retained both the forms and the names of its letters, slightly 
 modified, in the order in which they originally stood; the 
 Semitic terms, Aleph, Beth, Gimel, etc., being transformed 
 into names more euphonious to Greek ears, but of course 
 unmeaning except as signs. These names, through the influ- 
 ence of Greek civilisation, have become identified with the 
 practice of writing in all ages and countries ; and the word 
 'Alphabet' (from the first two Greek letters. Alpha, Beta), is a 
 lasting memorial of the obligations of modern literature and 
 science to primitive Oriental ingenuity. The old explanation, 
 that the name of each letter was . the name of some familiar 
 object, the first sound of which was the element to be repre- 
 sented, Aleph {Jl) being Phoenician for 'ox,' Beth (S) for 'house' 
 (cp. Beth-el =' House of God,' Beth-horon, etc., familiar to us 
 from the Bible), Gimel {G) for ' camel,' etc., is now discredited ; 
 and the Phoenician alphabet is no longer regarded as the ultimate 
 source of the world's alphabets, but is itself traced back to an 
 Egyptian source, being in its origin hieroglyphic *. 
 
 ' See Max MiUler, ' Chipa,' vol. iv. p. 486 ; and eapeciaJly Lenormaut, 
 ' Introduction & une m^moire sur la propagation de 1' alphabet Phenicien 
 dans I'Ancien Monde' (published 1866). 
 
The Greeh and Roman Alphabets. 41 
 
 The names of the letters were but little changed either in 
 Greece or the East, though their forms must have undergone 
 some alteration. The original community of form between the 
 classical Greek characters and the later Phoenician may be 
 traced in the older inscriptions of the two languages. The 
 whole Phoenician alphabet of twenty-two letters was adopted 
 by the Greeks with certain variations of power and order, as 
 appears from the subjoined table', in which column I gives 
 the Phoenician alphabet, as a representative of sounds, and 
 as a numeral system (this latter usage being also adopted by 
 the Greeks) ; column II, the whole number of letters ever used 
 by the Greeks in their earliest forms (twenty-one Phoenician, 
 in their original order, and five of native Hellenic invention) ; 
 column III, the classical Greek alphabet ; column TV, the cor- 
 responding numeral system; column V, the Roman alphabet, 
 with the probable pronunciation of its different letters. 
 
 The old Phoenician alphabet consisted only of consonants ; the Phoenician 
 pure vowel-sounds (like a medial in Sanskrit, p. 35, note 3), being 
 considered as subordinate aids to pronunciation, and included in 
 the power of each consonant. In Greek etymology, however, 
 the vowels were of almost equal importance with the consonants ; 
 and required to be as exactly distinguished as these, in a lan- 
 guage which depended so much upon poetry and music for its 
 full formation. But for this purpose they had not to invent 
 altogether new characters ; for several of the letters of the Phoe- 
 nician alphabet, though technically classed as consonants, were 
 more properly semivowel in character, and were appropriated 
 by the Greeks to denote the vowel-sounds to which they had 
 respectively affinity. These letters were Aleph, He, Jod, Gin, 
 which were adopted as the simple vowel-sounds A, E, I, 0; 
 while Vau, which, on analogy of the others, should have been 
 converted into U, retained its original power, as the expression 
 
 ' For the general plan of this Table, and some of the information about 
 the Phoenician alphabet, I am indebted to Col. Mure's ' History of the 
 Language and Literature of Ancient Greece,' Book L oh. iv. § 8. The 
 information about the Greek alphabet is derived irom Kirohhoffs ex- 
 haustive little treatise, ' Studien zur Gesohichte des Griechisohen Alpha- 
 bete' (Berlin, 1867) ; that on the Eoman alphabet mainly from Corssen. 
 
42 
 
 The Greeh and Roman AlpJiabets. 
 
 [chap. 
 
 Table showing the Coeeespondencb between the Phoenician, 
 Gbbek, and Eoman Alphabets. 
 
 Phoenician 
 
 Alphabet and 
 
 Numerical Value. 
 
 II. 
 
 Full No. of 
 
 Greek 
 
 Letters 
 
 (earliest form •). 
 
 III. 
 
 Classical 
 
 Greek 
 
 Alphabet. 
 
 IV. 
 Numerals 
 correspond- 
 ing. 
 
 V. Eoman Alphabet. 
 
 Characters. 
 
 Probable 
 Pronunciation. 
 
 I Aleph = I 
 
 Alpha A 
 
 A 
 
 I 
 
 A 
 
 ah 
 
 2 Beth = 2 
 
 Beta B 
 
 B 
 
 2 
 
 B 
 
 I 
 
 SGimel = 3 
 
 Gamma ^ P 
 
 r 
 
 3 
 
 = K 
 
 k (c in cat) 
 
 4 Daleth = 4 
 
 Delta A 
 
 A 
 
 4 
 
 D 
 
 d 
 
 5 He =5 
 
 E E 
 
 EitiXoK 
 
 s 
 
 E 
 
 ay (0 in whale) 
 
 6 Vau = 6 
 
 Vau F 
 
 
 6(.') 
 
 F (older 1') 
 
 f 
 
 7 Zain = 7 
 
 Zeta I 
 
 z 
 
 7 ' 
 
 (G) 
 
 g (in give) 
 
 8 Heth = 8 
 
 Heta B 
 
 H 
 
 8 
 
 H = ' 
 
 h (in hat) 
 
 9 Teth = 9 
 
 Theta 
 
 
 
 9 
 
 
 
 10 Jod = 10 
 
 Iota pi 
 
 I 
 
 10 
 
 jJvowel 
 [consonant 
 
 ee (in feet) 
 y (in yes) 
 
 II Kaph = 20 
 
 Kappa K 
 
 K 
 
 20 
 
 (K) 
 
 k 
 
 1 2 Lamed = 30 
 
 Lambda \ 
 
 A 
 
 30 
 
 L (old U) 
 
 I 
 
 13 Mim = 40 
 
 Mu |W 
 
 M 
 
 40 
 
 M 
 
 m 
 
 14 Nun = 50 
 
 Nu rj 
 
 N 
 
 50 
 
 N 
 
 n 
 
 15 Samech= 60 
 
 Sigma Q 
 
 B 
 
 60 
 
 
 
 16 Ay in = 70 
 
 
 
 lUKpOV 
 
 70 
 
 
 
 ' 
 
 17 Pe = 80 
 
 Pi r 
 
 n 
 
 80 
 
 P (P inscr.) 
 
 P 
 
 18 Tsade = 90 
 
 M 
 
 
 
 
 
 19 Koph =100 
 
 Koppa Q 
 
 
 90 (0) 
 
 Q (old 0) 
 
 k 
 
 20 Eesoh = 200 
 
 Rho P 
 
 p 
 
 100 
 
 R (old P) 
 
 r 
 
 21 Shin =300 
 
 Xi t. 
 
 s 
 
 200 
 
 S 
 
 s ^ 
 
 22 Thau =400 
 
 Tau T 
 
 T 
 
 300 
 
 T 
 
 t 
 
 
 T YV 
 
 •r 
 
 400 
 
 V vowel 
 
 00 (in fool) 
 
 
 Phi <J) 
 
 * 
 
 500 
 
 V consonant 
 
 w, or Fr. on in oui 
 
 
 Chi X + 
 
 X 
 
 600 
 
 X=H 
 
 X 
 
 
 Psi NKy 
 
 V 
 
 700 
 
 (Y)-i later in 
 (Z)J repres 
 
 troductions to 
 
 
 n 
 
 n liiya 
 
 800 
 
 ent Gk. </, f. 
 
 
 
 
 900 ( ) 
 
 
 
 ' The Greek characters are simply the Phoenician characters reversed to suit the mode of 
 writing from left to right, which in Greece superseded the earlier right-to-left ('links lauiig ') 
 writing of the Phoenicians. Examples of this latter method are found in the oldest Greek 
 inscriptions : while in those of the sijcth century B. c. is found a mixture of the two modes 
 ('furchenfdrmig' or ' ridge-like '), in which the writing is read backwards and forwards in 
 alternate lines, e.g. I S T I A '• ^' 'Io"''ia(ros) 
 
 n T 3 )l H (.dvie)riK€ Ti- 
 
 n A A n -irdWaiyi) 
 
III.] The Greek and Roman Alphabets. 43 
 
 of the consonantal sound of U ; though it may have served Greek 
 
 ..„.,, , •" ° •' . alphabet, 
 
 originally lor both vowel and consonant sound. A new sign 
 
 Y or V J expressing the vowel-sound of U, took its place 
 at the end of the twenty-two Phoenician letters. No known 
 Greek alphabet, however, is without this sign, or has less than 
 twenty-three letters. The invention, therefore, and adoption of 
 
 V was probably contemporaneous with the adoption of the 
 Phoenician alphabet ; and we may regard the original Greek 
 alphabet as consisting of twenty-three letters, identical with that 
 which appears on the oldest inscriptions of Thera and Melos 
 (circ. 620 A.c). The Vau (better known under its later name of 
 hiyanfia, from its form F) did not long remain in use, its form 
 being only known to us from early inscriptions. It reappears, 
 however, in the Latin F, occupying in the Koman alphabet the 
 place of Vau in the Phoenician, but denoting a different sound. 
 
 These five vowel-characters at first denoted both the long and 
 short vowel-sounds, E and O having besides the functions of the 
 diphthongs (i, ov, which were not until comparatively late times 
 expressed by combinations of simple vowel-characters. Much 
 earlier, before about 620 A.c, the attempt was made to dis- 
 tinguish long and short e by the adoption of a sign 8 fo*" 5, 
 E remaining for e and ei. This sign 8 was apparently a modi- 
 fication of the Phoenician Q (Heth, Hebrew 'Cheth'), which in 
 its open shape H was used to denote the spiritus asper. H02 
 occurs on inscriptions^ or, and H stands for iKarov: and this 
 usage of course survives in Latin H. "When H was taken as 
 a vowel, it was cut in two so that (- represented the spiritus 
 asper, -{ the spiritus lenis ; whence came the signs " ' for the 
 breathings. 
 
 Somewhat later (according to Kirchhoff, about 550 A.c.) arose 
 the distinction between and d, by the introduction of a new 
 character Q \ which took its place at the end of the alphabet, 
 after three new characters expressive of double consonantal 
 sounds. The names O fii/cpoc=o, and Q /xeya=oo or 5, were 
 
 • n and H were introduced into the Athenian alphabet in 403 B.C. 
 (arohonship of Eucleides) ; but their iuventiou must be placed much further 
 back. 
 
44 The Greek and Moman Aljohabets. [chap. 
 
 ^■^H . given after this to distinguish what had hitherto been combined 
 alphabet. . ° 
 
 m one form. The character a (i.e. oo), introduced about 
 
 Hadrian's time, made its form, like its sound, a lengthened 
 o (cp. our w or ' double m'). 
 
 The Phoenician Teih=:t sound, and Tkau=ih, were retained 
 'in their places by the Greeks, but their respective powers were 
 interchanged, and the names slightly altered to correspond; 
 Teth becoming Thsta and representing the aspirated t, and Thau 
 becoming Tau for the unaspirated tenuis. 
 
 The Phoenician alphabet was remarkable for its number of 
 sibilants— viz. Nos. 7.( I )> "5 ( H )' i8 (M), 21 ( J )' (I in 
 its later form, 2 ) w^as adopted at first to denote the double 
 consonant sound of ts or 8s, peculiar to Greek, and retained 
 this its original force throughout. To express the simple 
 sibilant the Greeks had the three remaining characters to choose 
 from. At first it was denoted by M (in the alphabets of Thera, 
 Melos, Crete, Corinth, Corcyra, &c., Olymp. 40-80); then by 
 %, or (written in a shorter form) 5, whence the Latin form S. 
 The superfluous character M then disappeared from the alphabet; 
 but the later form oi f^ (Phoen. alph. 13) exactly resembled it 
 in shape : hence the apparent anomaly of the same form de- 
 noting at different periods such unconnected sounds as s and m. 
 The remaining character B escaped extinction, because the 
 Ionic alphabet, which finally prevailed in Greece, had employed 
 it (as I ) to denote the compound sound ks (|'). A later form 
 still of 2 (sigma) was C > whence in late authors the orchestra is 
 called TO ToO Sfarpov a-lyfia, and sigma^a. semicircular couch 
 (Martial, x. 48, etc). [The Latin form S arises from the attempt 
 to write 5 in a single stroke. In the classical alphabet it has 
 interchanged places with S.] We also hear of a-av^, a Doric 
 
 ' The numberg refer to the Table on p. 42. 
 
 ' Herodotus (i. 139), speaking of the Persian names, says they all end 
 in the same letter, rd AaipUes jikv SAv icaKiovffi, ''iuves dk Xiyfjta. (Tafi(p6pas 
 (Ar. Eq. 603, Nub. 122, 1298) is a horse marked with the old letter a&v; 
 cp. KOTtnarlas Itnros, Nub. 23. Col. Mure assumes adv to have been derived 
 from the Phoenician Zain, and places it in col. II. of his table between 
 Vau and Heta, supposing that the Dorian usage of aAv = a alluded to by 
 Herodotus was a mere provincial anomaly. Liddell and Scott regard it as 
 a 2nd sibilant, which : Phoen. Shin : : 2 : Samech. 
 
III. J The Greeh and Roman Alphabets. 45 
 
 form of ori'yfia, which only remained as a numeral =900, under Greek 
 ,, V • j_i i- -k alphabet. 
 
 tne name tsa^mi, in the lorm ?j, 
 
 Koppa, Q, disappeared from the classical Greek alphabet, its 
 sound being so like that of K, that one sign sufficed for both. 
 It remained however as a numeral =90, and is found in old 
 Doric and Aeolic inscriptions : and it survives in Q, which the 
 Romans adopted from the Dorian alphabet of the Greeks of 
 Cumae. 
 
 Xi was originally written X2 (chs). The original Greek alphabet 
 had no sign for the guttural and labial aspirates (M, ph), 
 nor for those combinations of a mute with a following sibilant 
 {ks, ts, ips) which seem to have been regarded by the Greek ear 
 as single indivisible sounds, requiring a corresponding ex- 
 pression in writing. The dental aspirate was from the first 
 expressed by © ( (g) © O ) ! and the Phoenician character J 
 in its later form Z was, as we have seen, adopted as the sign of 
 dental mute + sibilant (ts). The other combinations, nh, kJi, 
 KS, ITS, are expressed on the oldest inscriptions by juxtaposition 
 of the signs for their component parts : but the range of 
 the alphabet was afterwards increased by three new signs, cp , 
 X (4-), V ( Y ), which took their place after V- Their introduc- 
 tion must have been early ; for no alphabet but those of Thera 
 and Melos is without them. The order, however, and signifi- 
 cance of the new signs varied in the two main groups of Greek 
 alphabets. Thus (i) in the Eastern group (including Argos 
 and Corinth in Greece proper) the order was cfj X ^) signify- 
 ing as in classical Greek wh, kA, jts respectively : the sound of 
 KS being denoted by I ( |-f| ), a variety of the Phoenician ffl 
 (Samech). (2) The Western alphabets put X before 4) , and 
 gave it the value of ks, denoting kA by V, and using for n-j the 
 old expression tts or (f>s. 
 
 This latter usage (of "Western Greek alphabets) represents, 
 according to Kirchhofi; the original order of these signs, super- 
 seded by the ultimate prevalence of the Ionian alphabet, as 
 exhibited on inscriptions of the Aegean islands. The Eoman 
 alphabet, derived from a Dorian source (see below), has pre- 
 served the force of X ='«s (^)i ^^^ (as a numeral sign only) that 
 
4^ The Greek and, Roman Alphabets, [chap. 
 
 of V=kA: but in the classical Greek alphabet all trace of 
 these values has disappeared. On the other hand, in Roman 
 inscriptions of all periods we find XS for X ', which looks as if 
 the idea of X as=cA (x) still influenced to some extent its written 
 value. Its place in the Roman alphabet was of course deter- 
 mined by that of Greek X (x). 
 Sphabet. 'B. Roman Alphabet. The history of the Roman alphabet will 
 
 be found fully treated in such books as Wordsworth's 'Frag- 
 ments and Specimens of Early Latin' (Introd. chap. II), and 
 Roby's Latin Grammar, and need only be briefly noticed here ^. 
 It was derived from the Dorian alphabet of the Chalcidian. 
 colony of Cumae, as is shown by the form of ;S=?, and the use 
 of p (Koppa) ; and in its oldest form seems to have consisted 
 of twenty-one letters, viz. A, B, 0, B, E, F, Z, H, I, K, L, M, 
 N, 0, P, Q, R, S, T, V, X. The three aspirates th, jph, ch (in 
 the Dorian character ©, <p, .4^) were never used by the Romans, 
 but the characters were retained as numeral signs. ^ (x) was 
 written _[_, and abbreviated into L=5o; Q, with its circle in- 
 complete, became C=ioo, and was then identified with the 
 initial of centum : and cp became cio or ffl, and finally M= 
 1000. D = 5oo is the half of q) ; X=io is perhaps from ®, 
 an ancient form of 6 ; and V=5 is the half of X. 
 
 X went out of use at a very early period, its occurrence in 
 the ' Carmen Saliare,' and on a single coin, being in fact 
 the only evidence for its having been used at all ; but in 
 the time of Cicero it was reintroduced for the transcription of 
 Greek f. 
 
 E went out of use, pi-obably before the XII Tables, except in 
 a few old abbreviations (Kal.=Calendae, K=Kaeso, etc.), and 
 C was used to denote both the guttural tenuis and media, until 
 a modified form, Gf, was taken to represent the media. . The 
 earliest inscription upon which G is found is the epitaph of 
 
 ' Mr. Eoby quotes as instances eiestrad (S, C. de Baooh., see Appendix 
 I), taxsat, lexs, proxsumius, exsigito, deixierU. 
 
 ^ The most complete account (from which the others referred to are 
 mainly derived) is to be found in pp. 1-29 of Corssen's ' Ueber Ausspraohe,' 
 etc., vol. i. For illustrations and examples reference must be made to 
 Corssen. 
 
III.] The Greek and Roman AVphdbets, 47 
 
 Scipio Barbatus, i.e. not later than 240 B.C. (Appendix I). TheKoman 
 old G was retained in G, Ci\''= Gains, Grnaeus. alphabet. 
 
 In the time of Cicero, Y and Z were introduced for repre- 
 senting V and f in Greek words : and the combinations ih, ph, 
 ch, were written for the first time (see below, chap. iv). 
 
 The Emperor Claudius tried to introduce three new letters, 
 viz. an inverted digamma d, for the consonantal sound of v; a 
 reversed Greek sigma D, for hs or fs ; and the sign of the 
 Greek spiritus asper \- (see above, p. 43), for the middle sound 
 between i and m, i.e. Greek v. The first and third of these 
 letters are found on monuments of the period, but did not remain 
 in use : while for the ' anti-sigma,' as it was called, there is no 
 evidence even of contemporary monuments. 
 
 Double consonants ^ were not written till the time of Ennius, 
 who is said to have introduced the practice in imitation of 
 Greek. The earliest instance upon inscriptions is about 186 B.C. 
 (see Appendix I), from which time to that of the Gracchi 
 (about 120 B.C.) usage fluctuates: after 120 B.C. doubled con- 
 sonants are general. Another method of denoting stress upon 
 a consonant was by the 'sicilicus' (so called from its shape, 
 sica), e.g. ser'a, as'eres=:serra, asseres. 
 
 The example of Greek 17 and a led Roman literati to try 
 various methods for expressing long vowel sound. 
 
 (a) Doubling the vowel ; said to have been used by the poet 
 Accius, and found on inscriptions of his time between 130 and 
 75 B.C., always with the vowels A, E, F(vootum is found on an 
 earlier inscription). A double / was used after Cicero and 
 Caesar to express the semivowel J {Aiiax, Maiia, etc.). 
 
 (6) The length of I was in earlier Latin expressed by writing 
 it ei; but after Sulla's time by making the i taller than the 
 other letters (vicus, lIbbei, etc.). This tall i was also used for 
 the semivowel or _;' sound, especially at the beginning of words 
 
 ' The date of the introduction of doubled consonants is of some im- 
 portance for the scansion of Plautus. Plautus died fifteen years before 
 Ennius ; and but few instances of double consonants in his plays can be 
 really genuine. Supelectile, simUiimae, saielites, sagUa, etc. are generally 
 correct ; and ille, esse must often be scanaed ile, ese. See Wagner, Intro- 
 duction to AvZularfa, pp. xli-xliv. 
 
48 The Greek and Roman Alphabets. [chap. 
 
 (Its, Ivbbto, etc.) ; and in later inscriptions it is used con- 
 fusedly for both I and i. 
 
 (c) By an accent or 'apex' (') from about 63 B.C. onwards. 
 In Republican inscriptions this apex is found over A, E, EI, 
 0, V (for examples, see Corssen, I. p. 22); in the Augustan age 
 it is almost universal. 
 
 The Romans devised a simple nomenclature for the letters, 
 which has superseded the Graeco-Phoenician names of the Greek 
 alphabet. The vowels were denoted by their own sound ; the 
 explosive (mute) consonants and A by a vowel after them ; the 
 fricative and nasal sounds by a vowel before them. The vowel 
 employed for this purpose was e, except that k, h were called 
 ha, ha, q was called qu, and a;, ix. 
 
CHAPTEE IV. 
 
 Changes and Modifications of Sounds'. 
 
 These must be noticed here so far as they affect Latin and 
 Greek, and further illustrations of' their operation will appear 
 hereafter in the explanation of inflections. Sounds, we must 
 remember, are interchanged and modified (a) between two or 
 more languages [Smpv, lacrima ; duo, two, zwei ; Tr/j/re, quin- 
 que]; (6) in the same language [e.g. tego, toga; rpetpa, BpeiJAai]. 
 These changes and modifications proceed according to regular 
 rules, varying sometimes in different languages ; and the correct 
 investigation of these rules is the basis of all true etymology. 
 
 Two general principles affect etymology : — 
 
 (a) Harder sounds are not derived from easier ; or words Harder 
 which have retained a stronger letter from those which exhibit derived from 
 a correspondingly weak one. (In our Table of sounds Gutturals 
 are stronger than Palatals, and so downwards j Tenues stronger 
 than Mediae; Explosive stronger than Fricative.) Thus koIos 
 must be older than ttoXoi ; iquus than ittttos (Ikkos, asvas), svs 
 than vs. 
 
 Apparent exceptions are often examples of the general 
 phonetic law that change arises from weakness of articulation : 
 e.g. frag-or, frac-tus. It is easier to pronounce tenuis and 
 tenuis together, media and media, aspirate and aspiratfe : so in 
 
 ' In this chapter I have gone over somewhat the same ground as that 
 covered by Peile's ' Introduction to Greek and Latin Etymology,' adopting 
 in the main his arrangement of the phenomena of phonetic change, 
 with many of Iiis illustrations. 
 
50 Changes and Modifications of Sounds. [chap. 
 
 Greek, XfXTor from Xtyu, ru^^els from riir-ra, jrXex^els from 
 
 TrXeKck), 
 
 So hiem-j>-s appears stronger than hiem-s ; but p is merely 
 
 phonetic, inserted because it is difficult to sound s after m. 
 
 Again, the reduplicated form Bldijut is changed by Greeks to 
 
 TtStjiu, which is easier to pronounce, though t by itself is stronger 
 
 than 6. 
 
 Sounds usu- (p) Generally, only letters pronounced at the same part of the 
 
 changeable mouth are etvmologically interchangeable — dentals with dentals, 
 only at the . . . ° r^ 
 
 samepart of labials with labials, etc. [Gutturals, however, the strongest 
 
 the mouth. , , „ . . , , ■, , 
 
 sounds of all, sometimes give way altogether to dentals or 
 labials ; and in Sanskrit, Greek, Latin, and Gothic, are found 
 less frequently than tjiese latter.] 
 
 Apparent exceptions may often be explained by the existence 
 of both letters in the original form ; e. g. his and Si's : cp. 
 Sanskrit dvis. Here the Latin 6 represents the v (w) sound. 
 In tTTTTor, Lat. equus, Sanskrit asvas, tt represents the ■» sound. 
 uvv and cum are the same word; but from |w=kot5v: so leaTr-ror 
 and va/p-or are reconciled by Lithuanian Tceap-as. 
 
 The latest and most comprehensive explanation of such 
 changes, however, is that which refers them to the influence of 
 weah articulation. One or two examples of its effects are 
 subjoined. 
 'Lahiaiism.' I- Labialism, or change from k to w, p, is supposed by 
 Curtius to be due to a parasitical v (w), unconsciously produced 
 by lazy articulation of k ('labial after sound'). That v {w), 
 following k, could change it to p, appears from Indo-European 
 akva, Sanskrit asva, equos, inrros (vdiich must=4Kfor). Here kv 
 has become in Greek tttt. 
 
 That the v in these cases was merely phonetic, not a suffix, 
 appears from instances where Latin has kv (qu), as well as k (c). 
 e.g. sequ-or, sec-undus ; coqu-o, coc-us : Greek eiroftat, iriirav 
 show that V must be parasitic in Graeco-Italian time, and re- 
 tained by Latin in some words while dropped by Greek (Peile, 
 286, 7'). So with the change from g to j9, h ; Latin gu gives 
 
 ' Corssen ('Ueber Ausspraohe,' etc., i. pp. 71-75^ shpws that qu was a 
 mode of denoting the labial ' after sound,' or modification of the guttural 
 
IV.] Changes and, Modifications of Sounds. 51 
 
 the middle step. In urgtiere, urgere, tinguo, reyyo), v is parasitic ; 
 but it is less often so after g than after Tc, g being an easier 
 sound. 
 
 2. BentaMsm : h changed to t, probably from influence of y Dentaiism." 
 sound (t or _/), as in transition from -no to -tio, where i is 
 semivowel. Here it is part of a suffix; but this proves the 
 power of y sound to change a guttural to a dental, and hence 
 philologists assume a parasitic y where they find the change 
 without any apparent reason. There are, however, but few 
 certain examples : t/s, quis, Sanskrit kis ; Indo-European hatvar, 
 TiTTapes { = TeTFap(s), quattuor. 
 
 These two instances of change from one class of sounds to 
 another are given to ' show that some reason can generally be 
 found for the apparent non-observance of our rule (6). We 
 may now pass to the consideration of the two main heads under 
 which all changes of sounds seem to fall ; viz., 
 
 (a) Dynamic change, which is voluntary, and intended to 
 express change of meaning; the formative principle in lan- 
 guage. 
 
 (6) Phonetic^ change, which is involuntary, and due mainly to 
 lax articulation ; the destructive principle in language. 
 
 "We need only here enumerate, with a few examples under 
 each head, the principal changes of sounds that aifect Latin and 
 Greek ; referring the student for a fuller illustration to such 
 books as Schleicher's ' Compendium,' and Peile's ' Introduction 
 to Greek and Latin Etymology.' 
 
 A. Dynamic change. 
 
 I. Reduplication. This appears to be the earliest and most 'Dynamic' 
 
 -^ ^ ^ , cnange. 
 
 natural device of language to strengthen the expression of anHedupiica- 
 
 idea', observed most frequently in the language of savages and 
 
 tenuig, and so a transition from guttural h to labial p. In English, a 
 similar labial modification of the dentals is expressed in letween, dwarf, 
 and palatal modification is heard in the pronunciation of nature, verdure 
 (ty, dy). Tha labial modification of d (dw) is expressed in Old Latin 
 duellwm, but passed into the simple labial in classical Latin, tellum. Cp. 
 bis with Sanskrit dvis, quoted above; and duon:oro(m)=honorwm on old 
 inscriptions. 
 
 1 'Phonetic' is sometimes applied in m wider sense to any change of 
 sound, voluntary or involuntary: I have restricted myself to its more 
 limited application. 
 
52 Changes and Modifications of Sounds. [chap. 
 
 Eedupii- children, and commoner in the earlier than in the later stagfe 
 cation. p t • ' • 
 
 of highly developed languages such as Greek and Latin. In 
 
 these it is gradually superseded by more refined and subtle 
 
 modes of expressing the required change of meaning; and traces 
 
 of its application remain only in occasional and (for the most 
 
 part) exceptional phenomena: — 
 
 (a) In imitative words, e. g. ululo, dXaXafm ; or names of 
 
 animals expressive of their sound, e. g. euculus, furtur ; eiroyjr, 
 
 (&) In Alliteration \ a favourite device of early Latin and 
 Greek poetry (and also among other nations), to strengthen the 
 expression of an idea by mere repetition of the sound of letters 
 and syllables. Ennius and Naevius exhibit constant examples 
 of its use; as also does Plautus, with whom however it 
 becomes more of a trick of composition. Lucretius also em- 
 ploys it with considerable effect (for examples, cp. Munro's 
 'Introduction to Notes,' II. pp. io6, 107, ist ed.), and Virgil 
 does not disdain this among other poetical artifices, e. g. Aen. 
 
 vi, 834 :— 
 
 'Neu patriae validas in viscera vertite vires.' 
 
 Alliteration, of course, does not prove the use of reduplication 
 as a formative principle in language; but it illustrates the 
 natural tendency to intensify an idea by the repetition of sound, 
 (c) As a formative principle, Reduplication is commonly 
 employed in Indo-European languages to produce 'frequent- 
 ative ' and ' desiderative ' verbs. In Sanskrit such verbs are 
 regularly formed from every root, by reduplicating the initial 
 consonant and vowel of the root, and sufiixing in one case ya, in 
 the other ish or sh. Thus from the root budh (=' toTinow') 
 are formed bobudhye" (frequentative or intensive), bobudi- 
 sMmi (desiderative); from vid (='to know,' Greek fideiv, Latin 
 vid-ere), vividye (frequentative), vividiah^mi (desiderative). 
 Similar formations in Greek and Latin are fuipfjudpeiv (root mar', 
 
 ' On the use and effects of alliteration in Latin poetry, see Munro-'s 
 ' Lucretius,' Introduction to Notes, II. (vol. ii. p. 106, Ist ed.). 
 
 ' Greek lu-ixiofuu = /ii-fii-yo-iiai is perhaps analogous to ho-bhxid-ya. 
 
 ° The various ramifications of this root mar are exhaustively traced in 
 Max MiiUer's ' Lectures,' II. vii. 
 
iv.J Changes and Modifications of Sounds. 53 
 
 originally =' to grind down,' 'rub,' and so 'polish'), 'to flash;' 
 irajKpatvfiv (root <l>av, as in e-(l>dv-rjv), nomvia, heihia-a-ofiai ; ^if/)/ii)- 
 pi^eiv, cp. Lat. me-mor-ia, eta.; or in nouns, XmKa^ (root Xaft as 
 in e-Xa^-ov), SaiSaXfos, TramdKSeLs, dfiaifuiKeTos. The same force 
 appears in the reduplicated 2 aorist. 
 
 (d) Eeduplication is also employed in the formation of some 
 'present stems' (denoting, apparently, protracted as distin- 
 guished from momentary action), e. g. SiSaiu, rWrfiu, yiyiioiim 
 
 {=: yi-yev-Ofiai), irmra (jn-ner-io, root ttet, as in c-irecr-oj') ; Lat. 
 sisto, gigno, sera {=se-so, root sa, as in sa-tunC). More com- 
 monly still (in Greek almost universally), to form ' perfect 
 stems,' e. g. XiXonra (root XtjT, as in c-Xot-oi/), pepuli : such 
 forms, indeed, are too familiar to require illustration for the 
 present. 
 
 2. Vowel Intensification ('strengthening' or 'raising'), e. g. Dynamic 
 to strengthen the idea of a root for the formation of Noun ' Vowei-in- 
 or Verb stems Xot-, Xeot-w ; fid-, fid-o, foidus {foedus). It ap- tion.' 
 pears that Indo-European speech expressed these and similar 
 modifications of ideas, by strengthening or raising the vocal 
 sounds, in a regularly ascending scale of each of the three 
 primitive vowels, a, i, u. This ' raising ' or ' strengthening ' 
 was produced by allowing a stronger current of air to pass 
 from the lungs before souiiding the radical vowel of a word 
 — thus, in effect, producing the sound of a before such vowel. 
 We thus have three ' scales : '— 
 
 a ; I. a -H a= a 2. a -I- a = a (no distinction between ist 
 
 i : I. a + i = ai (e) 2. a + ai = ai. and 2nd), 
 
 u : I. a + u = au (o) 2. a + au=au. 
 
 The vowel sounds thus gained were used by different branches 
 of Indo-European peoples, according to different phonetic laws, 
 with more or less regularity. Sanskrit exhibits it most clearly'; 
 
 ' The two stages of vowel increase in Sanskrit are known by the names 
 of Guna (JtJS, 'quality') and Vriddhi (^^, 'increase'). Thus from 
 -V^vid, ' know,' is formed by Guna, "Veda ; by Vjiddhi (with addition of a 
 suffix -ilea) Vaidika=' belonging to the Vedas.' Hence ' Vaidic' is now 
 often employed by English writers as more correct than 'Vedic' In con-, 
 
54 Changes and Modifications of Sounds. [chap. 
 
 except that a is sometimes weakened to i and u, and that the 
 first raising of i, u, is e, o; the scales are employed as we have 
 given them (see Peile, chap. VI., and Schleicher for illustration), 
 in the formation from roots of noun and verb stems. 
 
 Indications of a similar employment of vowel scales are ex- 
 hibited by Gothic and Lithuanian (Peile, pp. 19 1-2), and also 
 by Greek and Latin ; by Greek most fully, the vowel system in 
 that language being far stronger and less liable to corruption 
 than in Latin. 
 
 Vowel- 
 scales : 
 
 The scales in Greek would be : — 
 
 
 Greek. 
 
 A scale '. a, t, 0, raised ist to 0, 0, 1;. 
 
 2nd to 17, a. 
 
 
 (e to 0, to a or ij). 
 
 
 e. g. ffv, yovri, -ff-yov-a. 
 
 ^■qyvvfUf epptuya. 
 
 
 \ey, \6yos, 
 
 ■nrliaam, TTii(. 
 
 
 tl>9tp, <p86pos, 
 
 
 
 ay. Idya, | Perfect stems. 
 
 
 
 I scale'. 1 raised 1st to El (ai). 
 
 and to 01. 
 
 
 e. g. Kiir, XEiTTOf, 
 
 A.^A.oiira, , 
 cf/ti, of/ios, Ol^IJ 
 
 
 TTiff, reiSu, 
 
 ■ndSu, itiitoiSa, 
 
 
 IS, clSai, 
 
 etSojf otfia. 
 
 
 XJ scale. V raised Ist to fv (ot/). 
 
 2nd to ov (av). 
 
 
 e. g. e\eve, (^A.fifl-oi'), 
 
 eiA.i}A.ou9a, 
 KiKfvSoi, iuoKovOiw, 
 
 
 mv, nveC/M, 
 
 ffeva. aovftai. 
 
 
 XU, X*"", 
 
 airivdai, avovS^, 
 or as a (xin/vviu). 
 
 jugation, vowel intensification strengthens verbal bases in those persons 
 whose terminations are called weak. Thus from V i, ' go,' is formed in 
 sing. €mi, ^shi, €ti ; in plur. i-Ta&a, &c. : cp. Gk. iliu, tX, (Tat, iftev. 
 
 '■ In the A scale Greek availed itself of the variation of the a sound 
 into a, (, u, originally merely phonetic but capable -of being applied to 
 distinguish different meanings (see p. 36). ■ o is a heavier sound than e ; 
 so that while f is employed for the present stem, the greater intensity' of 
 idea implied in the completed action can be expressed by j as e.g. in 
 irepSo), niiropBa ' (Peile, p. 194). It is however in the formation of nomi- 
 nal bases that this change of sound is most frequently employed, as in 
 the examples given above. 
 
 ' It will be observed that the I scale both in Greek and Latin affords 
 the best examples of the regular working of vowel-intensification. 
 
iv.J Changes and Modifieations of Sounds. 55 
 
 Vowel scales in Latin : — 
 
 
 Vowel, 
 scales ; 
 
 A scale, e, o, a, raised rst to o, e, a. 
 
 and to 5. 
 
 Latin. 
 
 e. g. i.mensj moneo. 
 
 2. sonus, persona. 
 
 
 nec-8, noceo. 
 
 v8cs, vocem. 
 
 
 seq-ui, socius, 
 
 v8co. 
 
 
 prec-ari, proous, 
 
 datorein, Sorrjpa, 
 
 
 fer-o, fors. 
 
 (Skt. dataram). 
 
 
 toga, tegula, 
 
 
 
 ISgo, legem, 
 
 
 
 plaoere, placare. 
 
 
 
 I scale, i raised to [ei], i, §. 
 
 2nd to [oi],oe, u.. 
 
 
 [ai], oe. 
 
 
 
 e. g. 1 . die, dico. 
 
 moenera, moinera, munera. 
 
 
 fid-es, fido. 
 
 foidus, foedus, 
 oinos, unus. 
 
 
 TJ scale, u raised ist to [eu]. 
 
 2nd to [ou], u. 
 
 
 au, o. 
 
 
 
 due, ducere. 
 
 douco (inscr.). 
 
 
 but here the distinction between the Steps is quite lost : for eu, and ou both 
 came back to u. 
 
 3. Nasalisation may possibly also be reckoned as a process of Dynam'<! 
 ' Dynamic Chanaie,' as beinar an extension of consonantal sound Nasalisa- 
 
 ■^ ^ ' " . tion. 
 
 parallel to that of vowel sound just described, e. g. in tundo 
 (root tiid), where tund : tud : : neid- : md. In its origin the nasal 
 is a variety of another sound (see above, p. 33), and the gut- 
 tural nasal is found (except in Sanskrit) only in combination 
 with the guttural which produced it {ayxovr), angustus). Nasal- 
 isation occurs chiefly in the formation of a certain class of pre- 
 sent stems in Greek and Latin ; and in the verbs so formed it 
 is (like vowel-intensification for the same purpose) confined to 
 the present stem. This formation will be found discussed in 
 chap, viii; and if the view there adopted, that the nasal is 
 dynamic, and not merely part of a grammatical suffix, be cor- 
 rect, we possibly may reckon Nasalisation with Reduplication 
 and Vowel Intensification as a means of ' Dynamic change ^.' 
 
 * Consult especially Curtius, 'Tempora und Modi," pp. 53-66; 
 Verbum,' i. pp. 240 sqq. ; also Peile, oh. vi. pp. 215-220 (3rd ed.). 
 
56 
 
 Changes and Modifications of Sotmds. [chap. 
 
 Phonetic 
 change: 
 
 Its effects. 
 
 Vowel- 
 change : 
 Substitu- 
 tion, 
 
 B. Phonetic Change : — 
 
 ,The difficulty of tracing vowel-intensification in Latin points 
 to the opposite process of weakening or decay, which in vowels 
 is particularly characteristic of that language. This brings us 
 to the second head of change, viz. that which is involuntary, 
 or ' phonetic ' strictly so called, the principle of ' Phonetic 
 Decay,' which plays so large a part in the history of language. 
 As has already been pointed out (chap, iii.) the cause of such 
 change is the conscious or unconscious effort after ease of 
 articulation. 
 
 Its effects may be traced (i) in the substitution of a weaker 
 for a stronger sound; (ii) in the loss of letters representing 
 certain sounds ; (iii) in ' assimilation ' to a neighbouring sound, 
 where the concurrence of two dissimilar sounds causes difficulty 
 of pronunciation ; (iv) in ' dissimilation,^ where the concurrence 
 of two similar sounds causes a like difficulty. We may look at 
 these effects I. in vowels, II. in consonants. 
 
 I. Yowels : (i) Substitution of weaker for stronger sound. 
 The Latin language, we saw, retained fewer traces than Greek 
 of the system of Vowel Intensification. On the contrary, it is 
 especially distinguished by weakness and decay of vowel sounds. 
 Thus of six diphthongs (ai, ei, oi, au, eu, ou), once in use as 
 Latin sounds, and traceable on inscriptions, five had dwindled 
 down to simple sounds by the time of Plautus, au being the 
 only one generally preserved (with the exception of at in a few 
 words, neu, ceu, neuter, heus, etc.) ; while here we have a 
 weakening to o, and forms with this weaker sound side by 
 side with those retaining au ; e. g. lautus, lotus ; Claudius, 
 Clodiua ; plaudo, explodo. The other diphthongal sounds were 
 entirely superseded by the weaker forms ; e.g. quaistor by 
 quaestor ; coirare by coerare, then curare (cp. foidus, foedui) ; 
 deivus by divus, omneis by omnes ; joudex hy judex. In vowel 
 sounds again, Latin shows a constantly progressive degradation 
 of sounds from stronger to weaker, as represented in the fol- 
 lowing scale : — 
 
IV.} Changes and Modifications of Sovmds. 57 
 
 A to O ... U ... B ... I Substitution- 
 
 O to U...E...I of Vowels: 
 
 TJ to E . . . I 
 
 E to I, U. 
 I to E. 
 
 A comparison of Greek and Latin with Sanskrit shows that 
 the original vowel a (largely predominant in Sanskrit) has been 
 changed first by breaking up into the three sounds of a, e, 
 (e. g. jToSti?, TTo'Ser, nohas, all = Sanskrit padas), and then by 
 further weakening of each of these three sounds. While, how- 
 ever, in Greek the process of vowel change was (speaking gene- 
 rally) confined to the breaking up of original a into a, c, 0, {l 
 remaining unaltered,) and a vowel of one scale but seldom pass- 
 ing into another (a to i, or d'): in Latin such further changes are 
 so frequent as to assume the character of special phonetic laws 
 of the Latin language. A few examples under the heads of 
 formative elements, composition, and reduplication, will make 
 this clear, 
 
 {a). Formative elements: i. Verbs. Sanskrit bhara-mas be-InPorma- 
 comes Greek <l)epo-fiev (Doric), the thematic '■' a passing to the ments. 
 sound and there remaining. Latin weakens to m in vol-u-mus, 
 and generally weakens it still further to i, e. g. Jer-i-mus. The 
 apparently unsubstantial character of the second a in tohara,- 
 mas, as a mere link between stem and termination, has perhaps 
 helped its decline : and we see that in the termination -mus it 
 has not sunk so far. In the termination again of 3rd pers. plur. 
 (Sanskrit anti, Greek -ovTt (Doric), Latin -unt), the weakening 
 of a in Latin, though it goes one step further than Greek, stops 
 at u ; while such forms as dederont, found on inscriptions, and 
 the retention of after u in quom, loquontur, etc., show that the 
 sound was not entirely superseded. In ferentem, and ferendmn, 
 however, the a before nt sinks down to e; old legal forms like 
 jure dictmdo, res repetimdae, also morihundus, oriundus, secundus 
 
 ' Schleicher (comp. § 32) gives a few examples of a of primitive Indo- 
 European roots weakened to i or u in Greek, e. g. dad&mi, Si5a)fit; /iv\os, 
 from root mat (mar) ; vvict-, Lat. Twct-, Skt. naktar. See also Peile, eh. vii. 
 (P- 275. 3r<i ed.). 
 
 ^ The 'thematic' or, as it is sometimes called, 'connecting vowel' is 
 explained in oh. viii. 
 
58 Changes and Modifications of Sounds. [chap. 
 
 SaMi\tvAvxa.(=.seque/ndus), etc., remaining as evidence of an earlier stage in 
 the process of decline. The Greek participle stops short at o, 
 in <l>epovTOi. 
 
 In final syllables the vowel usually sinks to e, e. g. mowueront, 
 -unt ; then {nt being lost by the lax pronunciation of the final 
 syllable, which was characteristic of Latin) motmere; uteris, 
 ufere/ ipsus, ipse; cp. ablative in e of 3rd declension, origi- 
 nally I. The reason for i, the weakest of the vowel sounds, thus 
 passing to e in termination, is suggested by Corssen to be, that 
 in pronouncing e the organs of speech vary least from their 
 position when at rest. 
 
 2. Nouns. The masculine terminations -os and -us (in early 
 Latin -os) represent Sanskrit -as, the a sinking in Greek to 0, 
 in Latin a step further to u. So neuter terminations ; Sanskrit 
 ganaa, Greek yevos, Latin genus ; in oblique cases sinking to -es, 
 yive{a-)os, gener-is ; but in some words retaining the stronger 
 sound, e. g. corpor-is = corpos-is, from corpus. In gen. plur. 
 duonoro{m) = bonorum,, shows that -wm is a weakening of -om 
 (cp. Greek -av), representing an original -am, the older vowel 
 sound being apparently retained in provincial Latin, and trans- 
 mitted to modern Italian — e. g. lore := illorum. 
 
 tfou °™^^" (^) Composition : a, 0, u, in Latin frequently weakened to i, 
 the lightest vowel, from efibrt after lightness of sound ; e. g. 
 causidicus [causa), armiger {armo-), comiger {cornu). Especially 
 is this the case in composition with prepositions, where such 
 weakening is the rule with but few exceptions, cp. e.g. capio 
 with its compounds, and with auceps, cestus with incestus, etc. 
 In Greek compound verbs, on the contrary, the original form 
 remained generally intact — cp. aya, (rvvaya, napdya, Kardya, with 
 ago, redigo, suhigo, etc. Where this prevailing tendency did 
 not obtain in Latin words, the exception may sometimes be 
 accounted for by the particular meaning, e. g. in tepefacere, cale- 
 faeere, etc. The idea of causation, obviously represented by 
 facere, may have prevented the sinking of the vowel which takes 
 place in eonfioere, perfice/re, etc. Other exceptions, for which it 
 is not easy to see a reason, are postJiabere, cp. with prohibere, 
 p&rfaciles with difficiles, expando (perhaps to distinguish it from 
 compounds oipendo, e.g. expendo), etc. 
 
IT.] Changes and Modifications of Sounds. 59 
 
 (c) EedupUcation. Here Greek and Latin both weaken the in Eedupii- 
 vowel in the reduplicated syllable in most cases to e, as in 
 Terv0a (root tvtt), tetuli. Latin in some words retains a 
 stronger vowel, e.g. poposci, cucurri ; but side by side with 
 these are found such forms as peposci, showing the tendency to 
 uniformity, regarding these syllables as mere grammatical forms. 
 And Latin goes further than Greek in weakening the vowel of 
 the radical syllable also, e.g. pepigi {root pag, seen in pac-tum), 
 cecidi (root cad). 
 
 (ii) Loss of Vowel Sounds. Uncommon in Greek, except in a Vowei- 
 few verbs which form a present stem by reduplication, and dropLss.^*' 
 the root vowel, m-n{€)T-at, yi-y(e)v-onai, fu-/x(e)i/-(a, etc. ; and 
 sometimes in formative suffixes before an inflection, e.g. ira- 
 
 T(e)por, /iTjT^efpos. 
 
 In Latin : a drops in virgo {^virago), elarus and clamor 
 (root cat), palrna (Greek ■n-dKafo)), cupressos (jamdpia-a-os) ; o in 
 vict{o)rix, nep{o)tis, doci{o)rina, etc. ; u before I (this consonant 
 and vowel having an affinity for each other, as being produced 
 near each other in the mouth), in vinc{u)lum, peric{v)lum, 
 saec{u)lum, etc! ; and in words formed with the suffix -ulo-, the 
 preceding consonant (especially n or r) then assimilating itself 
 to I and producing the terminations -ello, -olio, -Ulo, -ullo, e. g. 
 ocellus (= ocululus), libellus (= liberulus), asellus (asinulus), 
 homullus (liomonulus), corolla (coronuld), bacillus (bacululus), 
 pupillus (jiupilulus), Stella (ster-ula) : e before r (its most cog- 
 nate consonant), especially in the suffixes -ero, -bero, -tero, etc. ; 
 lit(e)ri, inf{e)ra, ag(e)ri, laieb{e)ra, sac{e)ro, soc(e)rus, etc. 
 Far more frequent is the loss of i, the thinnest of the vowel 
 sounds, and the most frequent substitute for the stronger vowels. 
 It seems capable of dropping out from almost any position, as 
 e.g. in such familiar words as quaes{i)tor, audac(i)ter, val{i)de, 
 gaudeo (cp. gavisus), fer{i)t ; dixti {dic-si-sti), and similar con- 
 tractions ; teg{i)men, repos(i)tus ; eo{i)go, sv/r(i)go, porgo (beside 
 the longer form porrigo), etc. 
 
 That this decay of vowel sounds was caused by the vowel Effect of the 
 
 Till. c TiiiT-i accent upon 
 
 gradually dymg out 01 unaccented syllables, is the most recent vowel-loss. 
 and most probable explanation. This is not the place for a 
 
6o Changes and Modifications of Sounds. [chap. 
 
 discussion upon the Latin accent, such as may be found at 
 length in Oorssen's great work, ' Ueber Aussprache,' etc., and 
 briefly summarised in Peile's 'Introduction.' We need only 
 notice (i) that the decay first in quantity^ and then in form of 
 jSwaZ syllables, which marks the history of Latin speech, seems 
 most fully connected with the known law of Latin accen- 
 tuation^, never to accent the final syllable : (2) that many of 
 the apparent metrical irregularities in the lines of Plautus and 
 Terence (lines which, as intended to be spoken, are naturally 
 subject to the practice of ordinary pronunciation), are best 
 explained by the neglect in rapid pronunciation of sounds in 
 syllables upon which no stress was laid'. We have only to 
 pronounce the words ourselves to understand the Plautine 
 prosody of voluptdtem, ferentdrius, seneetuti ; and in these and 
 numberless other cases of comic prosody, the vowel sound is in 
 a kind of intermediate stage between full pronunciation and 
 total extinction — written, but scarcely heard, and liable to be 
 pronounced more or less distinctly according to chance. 
 
 N.B. The change of quantity from long to short in final 
 syllables is a loss or weakening of vowel sound, just as raising or 
 lengthening a short vowel is a gain or 'intensification of sound.' 
 Loss of quantity is an intermediate step to extinction of a final 
 sound ; and the gradual decay of sounds can often'be historically 
 traced through distinct stages of decline — a syllable with a 
 vowel naturally long becoming short in ordinary usages (as 
 am&t, monet, cp. amdre, monere ; honSr, cp. honoris), then 
 losing its final consonant, and finally, perhaps, disappearing 
 altogether. 
 
 ' As the lengthening a short vowel is a, process of raising or increasing 
 (or ' intensifying,' see p. 53) the vowel sound, so the shortening a long 
 vowel is a decreasing or diminishing, and the result a decay in quantity. 
 
 ' The rules of Latin accentuation (little familiar to us as rules from the 
 fact that they coincide so nearly with our English accentuation of Latin 
 words) are given by Quintilian, I. O. i. 5. 22-31. See Koby's 'Latin 
 Grammar,' vol. i. § 296 sqq. ; and Wordsworth, 'Fragments and Speci- 
 mens,' Introd. oh. iv. 
 
 ' On this question of Plautine and Terentian prosody I may be per- 
 mitted to refer to Introd., Part IV. of my (new) edition of Terence, 
 Andria (Kivingtons, 1875). Eeference is there made to other and fuller 
 sources of information. 
 
IV.] Changes and Modijicatioiis of Sounds. 6i 
 
 (iii) Assimilation of Vowels: (i) by Consonants, througli Vowei- 
 their phonetic relationship to particular vowels (see above). Assimiia- 
 
 a, the fullest and most independent vowel sound, is subject 
 to no such influence. It passes into o by weakening of articu- 
 lation, and so down the scale of descent to w, e, i. None of 
 these however rise to a, nor do u, e, i rise to o. The difierence 
 in strength between a and o was clearly felt, as also between o 
 and u ; but between u, e, i there was no such strongly marked 
 difference, and in their case the order of descent is sometimes 
 stopped or varied by the influence of neighbouring sounds. 
 Thus u, by its affinity to the labial nasal m, was retained at an 
 early stage of the language in sumus, volumus, the vowel which 
 in Sanskrit is a (bharamas), and in Greek o {<l>epofiev), sinking 
 generally in Latin to i (ferimus). To the same influence (of 
 labial 6) perhaps are due the forms Hecuba (older Hecoha, Greek 
 'EKa/S;;, and triumpus (Greek dplan^os). I however is the sound 
 which, especially when followed by another 'consonant, had the 
 greatest tendency to produce «. Thus e rises to u in pulsus 
 from pello, sepultus from sepelio, cp. mulgeo, dfniKym ; stultus, 
 stolidus, sulcus, oKkos, scopulus, trKon-fXos ; pessulus, irda-craXos ; 
 crajmla, KpanrdXrj. 6 had an especial afiinity to r' .■ so in oblique 
 cases of neuters in -us (= Greek -or, Sanskrit -as), where the s 
 becomes r, the u sinks to e, e.g. funus, funeris; genus, generis 
 {=igenos-is). Sometimes i rises to e, from influence of r, e.g. 
 pulvis, pulveris ; and e is the commonest vowel in Latin before 
 two consonants or a double consonant (a;); ^.g. judex, but 
 jiuMcis ; au^eps, but aucupis — remaining sometimes where one 
 of the two has been dropped, e.g. mile(i)s, milit-is, dive{t)s, 
 divitis. Cp. also the participial forms, -en{t)s and -endus, the u 
 surviving in euntis, etc., and in old legal forms, e.g. jure 
 dicundo. 
 
 i, as the thinnest of vowel sounds, and the point to which all 
 vowel sounds naturally tended to sink, can hardly be said to be 
 the result of assimilation, so much as of the absence of any assi- 
 milating tendency which would retain the vowel at an earlier 
 
 ^ Eoby, ' Latin Grammar,' vol. i. § 39. 
 
62 Changes and Modifications of Sounds. [chap. 
 
 Assimilation stage of decline. It seems however to have a. certain affinity 
 of Vowels. «, - . T'7/.v • 
 
 for dental sounds; e.g. before n in maehvna, oal{i)nea, trutina 
 
 (cp. lirixdmj, ^oKduiiov, Tpvrdvr^, and before the suffix -no in 
 
 dominus, etc. (cp. with Greek mBavhs, havbs, etc.); in a large 
 
 class of genitives in -inis, from stems originally in -on, e.g. 
 
 turbinis, imaginis, Tiominis (old form Jiemones), Apollims (old 
 
 form Apolones) ; before t, d, in verbal conjugation, agite, agito 
 
 iSyere, dyirto), gemitus (gemere), domitus (doma-re), debitus 
 
 (dehere); or in formations like candidus, frigidus {frige-re), 
 
 morbidus (morbo-). 
 
 (2) Assimilation of vowels by other vowels is seen in the 
 
 tendency of two vowels coming into contact to approach each 
 
 other. Thus a root vowel i, in cpntact with a, 0, and u, becomes 
 
 e in qiieam, gueunt, eo, earn, eundi, etc., but remains unmodified 
 
 in forms where there is no such contact, nequit, nequihat, imus, 
 
 ibo. The oblique cases of is show the same change. Again, 
 
 where two vowels are separated by a consonant, the latter 
 
 (especially i) tends to assimilate the former^. This is seen in 
 
 many proper names, Duilius for DueUius, Lucilius (Lucullus), 
 
 Popilius {populus) ; in derivatives, such as consiliv/m {consul), 
 
 fadlis {facultas), inqvMinus {incola). e assimilates o in bene 
 
 (originally bono, then bone), i in illecebrae (root lie, of allieio) ; u 
 
 is assimilated by in soboles {svholes), e by «* in tugurium {teg.). 
 
 Vowel- (3) Dissimilation is of less frequent operation, occurring only 
 
 ■pissimiiap in some cases where, from whatever cause, the same vowel 
 tion. _ 
 
 sound occurred twice, and acting then as a bar to further 
 
 change. Sometimes the two vowel sounds coalesced into one : 
 thus, when quom tended to become quum, by substitution of u 
 for 0, the two vowels often coalesced, with the result cum {q not 
 being written without u) ; but the principle of dissimilation 
 retained the old spelling even in the Augustan age, and we have 
 quom or cum, equos or ecus, linquont and lincunt, etc., in- 
 differently^. The same principle operated in retaining the older 
 forms evntis rather than eentis, ipsius, iUius rather than ipsiis, 
 illiis (the genitive ending -us, Greek -os, regularly sinking to 
 
 ' Eoby, ' Latin Grammar,' vol. i. § 41. 
 
 ' See Munro's 'Lucretius,' Introd. to Notes, I. (vol. ii. p. 27, ist ed.). 
 
IV.] Changes and Modifications of Sounds. 63 
 
 -is), and in avoiding eeis (dat. plur. of is), by the foTm eis or 
 ieis. 
 
 II. Changes of Consonants : The respective characteristics of Consonant 
 Latin and Greek are here reversed. We have seen that the General ten- 
 Latin vowel-system is weaker and has been subject to greater 
 degeneration by phonetic change than the Greek : but the 
 Latin consonants are stronger, and (as will be seen) are com- 
 paratively free from assimilation, which obscures the radical 
 form of many Greek words'. Bearing in mind what has been 
 already said of the relative strength of sounds, and of the general 
 principle which governs all phonetic change — the desire to secure 
 ease of articulation, the following general rules of consonantal 
 change will be intelligible : — 
 
 1. 'Explosive' (or 'Momentary') sounds change to 'Pro- 
 tracted ' (or ' Fricative ') sounds, not vice versa, e. g. c (^) to 
 c (s), centum to French cent ; t to s, tu, tv, (tv ; d to I, Su/cpu, 
 lacruma, 'OSva-a-^vs, Ulysses; b to v, habere, Italian avere, 
 French avoir ; f to v, sajpere, savoir : fh, ch, th, to /', ^ipa, 
 fero, x°^hi fcl, 6rjp (Aeol. <t>rip), fera ; ch to h, x"/""') humi, 
 XavSdvm, jpre-lmndo. 
 
 2. Gutturals change to dentals and labials, not vice versa. 
 
 3. Tenues change to mediae in their respective classes, not 
 vice versa (except where influenced by other sounds), e. g. frag-, 
 frac-tus (see above, p. 49). 
 
 4. Eules (2) and (3) apply most obviously and uniformly to 
 Explosive Sounds or Consonants proper. Among 'protracted' 
 or momentary sounds it is not so easy to trace definite rules of 
 change. The contact of the vocal organs being less complete, 
 in fact, an approximation only, the sounds are much less defi- 
 nite ; and their strength depends more or less upon the length of 
 time during which they are sounded. The spirants y, s, v do 
 not seem to interchange much, but neither s nor v pass into y, 
 which, according to order of pronunciation, would naturally be 
 the strongest sound. Of the liquids, r seems to be older than I, 
 Greek and Latin often giving I where Sanskrit has r; and 
 
 ' E. g in (ppdrrffw, as compared with Lat. farc-io, see below, p. 75. 
 
 ' For other examples see Roby, ' Latin Grammar,' i. § 99, and Corssen. 
 
64 Changes and, Modifications of Sounds. [chap. 
 
 Consonant hence some philologists consider that I arises always from a 
 General ten- weakening of original r, pointing in illustration to the fact 
 that many children are unable to sound r, and substitute the 
 easier I for it. Schleicher, e.g. ('Compendium,' § 147, 156), 
 refers to Xcukos, luc-eo, from root ruh (appearing in Sanskrit as 
 rty .) : loc-utus, XaK-ety, cp. with Sclavonic reh-a, ' I speak ;' 
 re-lic-tus, Xar-iiv, with Old Indian ri'A', etc. : but there seem to 
 be other roots in which I is invariably found (see Peile, p. 85). 
 And in the Romance languages I and r interchange both ways ; 
 e. g. peregrinus becomes jpelkgrino, and Tibu/r, Tivoli ; but hi,s- 
 ciniolus becomes rossignuolo, and apostolus, apdtre. s in Latin 
 always passes to r between vowels, except in some cases ^ where 
 s is not original but a substitution (e. g. for ss in causa, for d in 
 esuries, etc.), or in compounds of words with initial s (desilio) ; 
 and in Greek it frequently passes into the sjpiritus asper ' (aks, 
 sal; eSos, sedes ; vXrj, silva, etc.). This h sound in Greek is 
 always a remnant of one of the spirants, and weaker than any 
 of them ; in Latin it represents an original gh, and seems to have 
 been more strongly sounded. 
 
 We may instance the effects of phonetic change upon conso- 
 nants, under the same heads as those of vowel-cliange : — 
 Consonant (i) Substitution of Weaker for Stronger sound, (a) media for 
 Subsfitu- tenuis, g for k, pac-iscor, Trriy-mijn, pag-us, pango ; Kn/Sepi/am, 
 guberno ; curculio (Plautus), gurguUo ; negotium (nee-), t 
 seldom passes into d. In Greek Soirtr (Aristoph. Vespae 676) 
 perhaps = tottiji ; venoSes. (Hom. Od. iv. 404), perhaps = nepotes. 
 In Latin the confusion between t and d in the MS. spelling of 
 words like haud, apud, sed, is to be referred to the general 
 weakness and uncertainty of Latin final sounds, p passes to 
 6 in a few Greek words (e. g'. v^pis from imcp), and in rather 
 more Latin (e. g. carbasus, KapTraa-os ; la/mbo, lab-ium, Xairrew, 
 \a<j)i<Taa) ; and b in Latin frequently represents Greek ^, nuhes, 
 ve^os ; orbus, 6p^av6s ; and the suffix -ber = -<p6pos (cl>epm), salu- 
 ber, ctmdela-hrvm. 
 
 (J) Any further substitution of the mute or explosive sounds,, 
 e.g. t, d, p, b, belongs rather to the head of Assimilation. 
 * Eoby, ' Latin Grammar,' i. § 193. 
 
^•] Changes and Modifications of Sounds. 65 
 
 There are, however, in Latin a few instances of d passing into Consonant 
 
 7j/T-i_ii Ao change: 
 
 I and r (probably due to some peculiarity in the Latin pronun- Substitu- 
 ciation of d, bringing it very near the point at which I and r 
 were sounded), e.g. olere from root ad, cp. odor, oSaSa ; lacrima 
 =8dicpv (cp. Gothic tagr, our' tear ')j levir (see I'orcelIini),= 
 SaFrip, cp. Sanskrit devar. Similarly lingua may have been 
 dingua, cp. Gothic tttggd, German munge, English tongue; and. 
 Festus states that Livius Andronicus actually wrote dacrima^. 
 The change of c? to r is chiefly found in the preposition ad, in 
 composition before v, f, arvocatos, a/rfuerunt, arvorsum, arvena, 
 arfines, etc. ; cp. also arbiter {ad-beto), arcessa (ad-cesso). This 
 change is sometimes reckoned as an effect of assimilation ; but 
 more probably arose from a weak pronunciation of d, near the 
 point at which r is produced. And the appearance of these 
 words with the d in classical Latin seems to show that this 
 carelessness of pronunciation had only just begun to produce an 
 effect upon orthography, when it was checked by the literary 
 epoch of the language; and the inference is, that it was an 
 accidental and isolated phenomenon in the Latin pronunciation 
 of that particular sound. 
 
 (c) Changes of Spiramts {y, s, v). These have especially Changes of 
 affected Greek, and in a great measure produced the distinctive 
 feature of accumulation of vowels without a consonant, e. g. 
 Siji'oio [once 8a<r-y6-tryoj. 
 
 Y. This symbol is unknown in Greek from the earliest times, t. 
 no doubt because the sound denoted by it had disappeared or 
 become changed. It appears in Greek (i) as t, especially in 
 the BufiSx yo(io); thus avSpi-to^s, eicrejSeta (^cwo-c)3e(r-ya), yevereipa 
 (^^yeverep-ya), Tervcjivia (^:=Terv(fioT-ya^, hfijioio (Homeric genitive 
 =8^o-tryo); (2) as e, in Ktveos [Epic form of Kev-y6, 'empty'], and 
 in the contracted future forms, <^ev^ovpai=:<peuy-(Tio-pjii,-:=ip€vy- 
 ayo, the t remaining in the Doric form Ttpa^iop.es^'n-pay-o-io-iies; 
 
 ' Bergk suggests that in Ennius' well-known lines (Epigr. i. 4) — 
 'Nemo me lacrumis deooret neque funera fletu 
 Eaxit. Cur ? volito vivos per ora virum ' — 
 the poet may have writtea daorumis, which would obviously increase the 
 force of the alliteration. 
 
66 CJianges and Modifications of Sounds. [chap. 
 
 Changes of (3) as the ' spiritus asper,' ^irap, jecur; Syios, Sanskrit yagyas ; 
 T. va-iilvr], from root yvdh {v6 becoming ia- before /i). Or (4) it is 
 altogether lost, as e.g. in Attic genitive mnov='hrrroo, cp. with 
 Epic "mnoio ; TrXeoxsrirXtioi/, ^ua>=:an older (pvto) (Sanskrit bliu- 
 ya-), and )(d\Kcos (Doric) :=;(dXK€toy, aXa5ea=dXa5«a. 
 s. S in Greek usually passes into the spiritus asper at the begin- 
 
 ning of a word, as we see from many familiar examples where 
 the analogy of other languages shows that the word once began 
 with a-. {cSos, sTTTa, eiroimt=::^sequor ; os^irFos^suus). Some- 
 times a spiritvs asper is due to a lost o-, which was not initial, 
 e. g. e'uTTTiKeiv = e-iariiKeLv = t -cf oriyKctv (the root sta being, redu- 
 plicated to sa-sta, o-c-ora); rififis = d(TiJ^K, cp. the form &fifies= 
 SxTfies. At the end of roots and words an original s generally 
 retained its place in Greek, and so in formative and case 
 suffixes, being, in fact, one of the few final sounds which Greek 
 euphony tolerated : at the beginning of words also it is some- 
 times found, e. g. myaa, <T€\rpn] ; but only regularly when a hard 
 consonant follows, and protects it from the usual change to ', 
 as in aropcvwfu, trroa, (rx'T"") ^tc. 
 
 In Latin s between two vowels is almost always changed to r, 
 e. g. lares = lases (Carmen Arvale), feriae =fesiae (cp. festus) ; 
 heri, hesternus, x^^l 6ram, ero, from stem es of esum; gero, 
 ges-tum ; oneris, generis = ones-is, genes-is, from stem onus, 
 genus ; gen. plur. suffix -r'Mm:= an original sam, and r of passive 
 voice =s«. The only exceptions to this general law of change 
 are (a) where s is not original, but a substitution for other 
 sounds, e. g. for ss in causa, for d in esuries (ed-o), ausim 
 {avd-eo) ; (6) in compounds of words where s was initial, de- 
 silio, po-siiura, prae-sentia, bi-sesctus, etc. ; (c) in certain other 
 words, viz. asinus, hasium, caesaries, casa, caseus, cisium, fusus, 
 laser, miser, nasus, pusillus, quasillum, quaeso (also quaero, rosa, 
 vasa) ; and in some proper names, Caesar, Kaeso, Lausus, Pisa, 
 Sisenna, Sosiae. 
 
 In Greek, on the other hand, o- between two vowels drops out, 
 e.g. yiveo'-os, yeveos, yivovs, TVTrrri-iTai, Tvrmm, Tmrg ; and this is 
 almost universal, except in cases where the loss of o- would have 
 created confusion, and an artificial efibrt was therefore made to 
 
IV.] Changes and Modifications of Sounds. 67 
 
 , retain it, e. g. rams would become tcus. In derivatives of this Chanses of 
 
 c • , . n . ,. Spirants: 
 
 torm, in -ns, and m some inflections like Tidrja-i, la-Taaat, the 
 tendency to drop o- was resisted ; the intellectual or instinctive 
 desire of retaining the part of a word, which was characteristic 
 of its meaning, in these cases triumphing over the physical 
 tendency to reduce the effort of articulation. Similarly in Latin, 
 the tendency to drop final -s, seen in the constant change of 
 forms like amabaris, amaberis, to amahare, amahere, is to a great 
 extent resisted in the form amaris, because amare would lead 
 to confusion with the pres. infin. active ; and ab is much 
 seldomer changed in composition than sub, because of the 
 danger of confusion with ad. 
 
 In Latin fi,nal -s (like final -m) seems to have been faintly 
 Bounded in pronunciation^, and thus was often omitted in 
 writing also. In the scansion of early Latin poetry it was 
 ignored before an initial consonant (a fact noticed by Cicero, 
 Orat. 48. 161), e.g. turn lateralis dolor certissimus nuntius 
 mortis j Ennius 601 (Vahlen), and so often in Lucretius (e.g. 
 i. 1S9, 186) and once in Catullus (116. 8. Ellis). From Terence 
 Wagner instances in the Hecyra auctus sit 334, defessus sum 
 443, incerfus sum 450, expertus sum 489, nuUus sum 653, 
 usus sit 878 — all endings of iambic lines. Virgil (Aen. xii. 115) 
 in imitating Ennius's funduntque elatis naribus lucem, trans- 
 poses thus (to suit a stricter pronunciation of final s) : lucemque 
 elatis nwribus efflant. The tendency recurred in the fourth century 
 A. D., and remains in Italian, Spanish, etc. It is also, of course, 
 illustrated by such forms as nauta (cp. vairrfs), ille, ipse (for 
 illus, ipsus), puer, famul {puerus, famulus), pote, m/zge {potis, 
 magis), ammbare {amabaris). 
 
 V=F. This spirant (pronounced^ something like English w) vorf. 
 was known to the Greeks at an early period by a distinct 
 symbol, the Digamma (F), found chiefly upon old Doric and 
 Aeolic inscriptions', and traceable in its effect upon the scansion 
 
 ' Boby, 'Latin Grammar,' § 193. 
 
 ' On the pronunciation of u consonantal (v) see Wordsworth, ' Frag- 
 ments,' etc., Introd. iii. §§ 10-15; ^°^y> 'Latin Grammar,' Preface, 
 pp. xxxii-xlii. 
 
 ' e. g. those of Boeotia, Phocis, Ldoris, Laconia, Argos, Corinth, Cor- 
 
 F 2 
 
68 Changes and Modifications of Sounds, [chap. 
 
 Chansesof of Homer; but evidently passing out of use at the earliest 
 Vot'f. period to which such inscriptions carry us back. It appears 
 in ordinary classical Greek as v, e.g. bvo (Sanskrit dvau, Gothic 
 tvai, German zwei), vmis (cp. nav-is, Sanskrit nav-as), and the 
 Aeolic forms x'""''j irvtia-^x'^^''' '""^Fia. In these latter Attic 
 Greek has lost it, as also at the beginning of many words, 
 in which, from the analogy of kindred forms in other languages, 
 it must once have existed {ohos, vinwm; ol&a, ISetv, vid-eo ; 
 epyov, German Werk, English worJc). It also appears as spiritus 
 asper (on the evidence, again, of analogy with other languages), 
 
 e. g. eanepos, Vesper/ cvuviu = ecr-Wfu, ves-tis ; UT-Tap = 18-T0p-, 
 
 from f i8- 6the verb forms having lost it altogether, and la-Tiup 
 in time coming to have the spiritus lenis). In a few cases 
 
 V (f ) seems to have been hardened or strengthened to /3, e. g. 
 the Laconian forms, ^ctos^^tos (cp. Latin vetits, veter-nus), 
 ^epyov:=Fepyov i and /3oij\o/iai, with its various forms, ^oXXo/iot 
 (Aeolic), ^aXoiiai (Doric), the original consonant of which must 
 have been v (f ), cp. vol-o, Sclavonic vol-i-ti (inf.), Gothic vU-jan, 
 German willen, English will ; for here the evidence of so many 
 languages for the v sound prevents us from regarding Latin v 
 as a weakening of /3. The occasional confusion between b and 
 
 V in Latin, and the representation (chiefly in Plutarch, a Boeotian 
 Greek, and an indifferent Latin scholar) of Latin v by Greek /3, 
 has been pressed as an argument against the pronunciation of 
 Latin v like w, and in favour of the labial sound of English v. 
 Even in Plutarch, however (ist cent. A.D.), ov is almost twice 
 as common as ^ for Latin v {OvdKrjpios, etc.) ; in Dionysius of 
 Halicarnassus (Augustan age) /S is only occasionally found ; 
 while in Polybius (2nd cent. B.C.) ov is the regular equivalent 
 for V. It seems therefore highly probable that the translitera- 
 tion of V by j3 is connected with a dialectical tendency to con- 
 fuse V and b in Latin, which appears in rare cases like ferveo, 
 ferbui, and afterwards more commonly on inscriptions of the 
 
 2nd century A. D. and onwards. The v in all such cases was 
 
 cyra, etc. (KirohhofF.) The Romans, taking a Doric alphabet (see p. 46), 
 found this character, but changed its value, thinking the w sound si^- 
 ficiently represented by V, 
 
IV.] Changes and Modifications of Sounds. 69 
 
 possibly the 'labial v^,' passing irregularly but not permanently Consonant 
 into h : and the safest conclusion from the evidence of trans- 
 literation appears to be that Latin 'i; generally =w, but some- 
 times dialectically a labial v^. The substitution of /* and y for 
 F is also assumed in a few cases, of very uncertain etymology; 
 e. g. iJ.6(rxos, Sa-xos, d/ufiriv, avxr/v, fiiXSofiat, eKSo/uu, etc. 
 
 In Latin, just as j/ is represented by i (consonantal), and 
 sometimes disappears (e. g. in min{i)or, ero = esio), so v is 
 represented by u (consonantal), and sometimes disappears as 
 in s{v)ibi, t(v)ibi (roots sva-, tva-). It is also occasionally 
 represented hj/, e.g. frango, Fprjyvvfu ; frigus, Fpiyim ; and the 
 sign F is of course the old digamma, adopted by the Latins, but 
 to denote a different sound. 
 
 (d!) Changes of the Aspirates, especially the aspirated mediae Changes of 
 bh, dh, gJi in Latin. General rules : — These aspirates {gh, dh, bh), ^"'^ 
 vrhen they occur in the middle of a word, are generally 
 represented by the corresponding unaspirated letters ;, when 
 initial they can all be represented by the single sound /. This 
 sound is not itself an aspirate, and has e.g. no power of assimi- 
 lating a preceding nasal like the other mutes in Latin {in-ficio, 
 but im-petus), so that it may be different in sound from Greek 
 (j) (c/i^mVm). Priscian's account of the difference between the 
 two, that ph is pronounced Jlxis and / non fixis lahris, is 
 explained by some to mean that ph ij an explosive or momen- 
 tary, / a fricative or protracted, sound. If this be true, / must 
 be considered as only a spirant or breathing, pronounced with 
 a strong breath, and taking the place of h strongly sounded 
 after h, d, g, the distinction between these letters being obscured. 
 
 * Labial (as distinguished from labiodental or English) v is formed by 
 bringing the outer edges of the lips together, while the voice escapes 
 laterally. This sound is said to be heard in Central Germany (e.g. in weg), 
 and in Spanish h, and modern Greek (Peile, ch. iv. p. 80. 3rd ed.). 
 
 ' For an admirably full discussion of the pronunciation of Latin v, see 
 Eoby's 'Latin Grammar,' vol. i., Preface, pp. xxxii-xlii; and cp. Peile, 
 ch. viii. pp. 355-357. Corssen (Aussprache, i. p. 310 sqq.) maintains that 
 V had not a ' weak vowel sound like English w, but a consonantal tone like 
 German w' — meaning the labiodental sound of English v. He much 
 exaggerates, however, the extent to which /3 represented Latin v (see 
 Eoby, I. c). 
 
70 Changes and Modifications of Sounds. [chap.' 
 
 Changes of and only one part .of the respective combinations h+Ji, d + h, 
 g + h being retained. At the beginning of a word the first part 
 of each fell away, leaving only the latter under the form of/ (or 
 h) : in the middle of a word, Latin generally retained the first 
 part and the latter or aspirate fell away. "We thus have/= hh 
 in fa/ri, root hhd, whence cfidvai; fui, root bhu, whence -bo, -bam 
 of future and imperfect : f = dh in firmua, root dha/r; fores, 
 root dhva/r, whence Sanskrit dva;ra, Greek 6ipa, German Thiir, 
 English door: f=gh 'va.fa-mes, ^a-rk (Sanskrit ga-hft-mi); /oms, 
 root fu = Greek ^u in i-x^-Srjv and forms of \(Fei, = originally 
 ghu, cp. Gothic giutan =.German giessen (whence ' Giessbach ' 
 the name of a water&ll) ; forrmis, ' warm '= Sanskrit ghar-maa, 
 Greek 6cp-ii6s ; f el =: Greek x"^- / also = Greek 6 in femina, 
 6rj\vs, fera, dfjp, and in other words, in some of which however 
 it and the 6 may represent an original bh or dh, as in fores. In 
 some cases, side by side with the form in which the aspirate has 
 sunk to /, is found another with h, used in the classical dialect ; 
 thus haedus, Sabine foedus [originally gh, the g remaining in 
 ' goat '] ; Jiariolus, Sabine fariohts (Greek xop-^v)- So hircus, 
 fircus ; hostis, fostis (root ghas, in Gothic gas-t-s, English guest): 
 and fordewm, foedos, attributed by Quiatilian (i. 4. 14) to the 
 old Romans. 
 
 F occurs most frequently as representative of bh, with which 
 it has the labial element in common; less often of dh, with 
 which it has only the use of the upper teeth in common ; least 
 often of gh, with which its only connection appears to be, as 
 already mentioned, the strong breath with which it and the h 
 of gh were each pronounced. 
 
 oSg?™' (^) '^°^^ °^ Consonantal Sounds : — 
 
 Loss. (a) Initial sounds, s and v (f ) are most frequently subject 
 
 to loss in both Greek and Latin, especially before the nasals m, 
 n and liquids I, r. Thus ijUpifwa, cp. Sanskrit smar-a,-mi, 'I 
 remember;' m6s, cp. Old High German snwr; p^a, root pv=<Tpv 
 (the s appearing in stream, strom, etc., and in the first p of 
 eppeva-a, fppirjv, etc. by assimilation from e-a-pev-a-a, i-a-pi-rjv) ; 
 show a loss of initial o- in Greek. In Latin, again, no native 
 word begins with sm, sn, or sr, and even borrowed words some- 
 
iv.] Changes and Modifications of Sounds, 71 
 
 times lose tte s, e. g. myrrha = a-iiipva ; cp. fwnda, <r(j)ev86vri, Consonant 
 fides (a string), o-^i'Si;, folio, ^(fiaWa, etc.^ v (F) is lost before p in Loss. 
 pifa (German Wurzel, English wort), and prjywfu, Latin frango 
 (where the f was a weakening of an earlier hh, traceable in 
 Gothic hrihan, English break) ; and before vowels in all cases 
 where it has not passed into the spiritus asper — e.g. ohos, 
 vinum, oIkos, vicus, epyov, work, IS-etv, vid-ere, etc., etc. 
 
 Loss of other consonantal sounds, when initial, is generally 
 ' sporadic,' i. e. confined to stray instances, which do not offer 
 sufficient evidence of any general phonetic tendency — e.g. the 
 loss of c in uhi, unde, preserved in ali-cubi, ali-cunde ^ 
 
 (6) Medial sounds are rarely lost in Greek, except in avoiding 
 difficult or impossible combinations of sounds, such as would be 
 Terv<p-a-6€, (ardKvTai (Ionic eWaXarai). t falls out regularly in 
 the oblique cases of certain nouns, e. g. Kepa(T)-os, and perhaps in 
 the 3 sing, of verbs (Ti7rTei=:Ti7rr£-Ti) — but another explanation 
 is more probable here^ In Latin, medial sounds are more 
 often lost. Corssen and Schleicher give a number of examples 
 of such loss, chiefly before spirants and nasals, e. g. miles = 
 milits, cesor cosol = censor consul ; ma{g)-ior, pe{r)-iero ; 
 re{s)mus=retmus, Greek ipcriiov; exa{g)-men, de{c)-nus, po{s)- 
 no (cp. pos-ui). Before momentary sounds such loss is rare 
 except before dentals — e. g. tor{e)tus (torq-ueo), id{o)-tus, ju{s)- 
 dex, i{s)dem. Schleicher considers that in all or most of these 
 and similar examples the lost letter has first been assimilated, 
 and then omitted, from the practice in old Latin of not writing 
 the same consonant twice * : e. g. res-mus, rem-mus, remus : an 
 ingenious attempt to bring under a uniform rule a number of 
 scattered examples, which may or may not be true, but is hardly 
 capable of proof. 
 
 1 Corssen ('Kritisohe Beitrage,' p. 428) suggests that Boma = Srouma 
 (root sru), the ' stream-town ;' a term applicable to the old 'Roma quad- 
 rata ' on the Palatine Hill, before the Tiber was kept within its banks. 
 This of course is but one among a number of competing etymologies for the 
 
 name. ,_.,,... , „ 
 
 2 Other examples are given by Peile, ch. viu. pp. 370-1 ; and Corssen 
 
 ('Kritisohe Beitrage," pp. 2, 57-H> H^)- 
 
 * See below, ch. viii. 
 
 * On double consonants in Latin, see above, p. 47. note. 
 
"]% Qhanges and Modificmtiom of Sounds. [chap. 
 
 Consonant (c) Loss of Final Sounds, i. e. of the consonant or consonants 
 Loss. ' of the final syllable. The tendency of all languages to throw 
 back the accent from the final syllable, gave this syllable a 
 weaker pronunciation, and made it liable to phonetic corrup- 
 tion, the extent of sucti corruption varying in different lan- 
 guages with the inability to accent the final syllable. Thus 
 in Latin, which never accents the final syllable, there is more 
 extensive loss of final consonantal sounds than in Sanskrit or 
 Greek : just as we have already seen that its final vowel 
 sounds are peculiarly liable to corruption, either by the 
 shortening of sounds originally long, or by total loss. The 
 operation of this common tendency to weaken or drop difiicult 
 sounds in final unaccented syllables varies with the phonetic 
 laws of individual languages by which certain final sounds are 
 accepted or rejected. The Greek ear, for example, allowed 
 no consonantal sound to end a word but v, »•, and less frequently 
 p; the only exceptions being ovk and « (^ or ■^ of course 
 include a) : and when any other consonant appears etymolo- 
 gically at the end of a word it is usually rejected — e. g. f»eXt = 
 stem jXiKiT, as seen in fniKvr-os, cmfta =i (tS/xot- (crii/iaT-or), ^<rav, 
 Latin erant, and all participles in -a>v, the stem of which is 
 -ovT, as in oblique cases Tmrrovr-os. In Latin -nt is an admis- 
 sible final sound (amant, erant, etc.), though in participles t 
 is changed to s in the nominative sing, {mnans, amantes) : and 
 the different treatment in the two languages of this participial 
 stem termination -nt is a good illustration of the direction 
 given to general tendencies of phonetic change by the phonetic 
 laws of individual languages. The paucity of admissible final 
 Bounds in Greek leads also to corruption of the final syllable 
 even when accented, e.g. riBfls = ndevrs. In Latin, as has been 
 shown, the tendency is for the final vowel to sink to a uniform 
 sound of e, but there is considerable variety of consonantal 
 termination : s, m, t, r, c, d being all found, besides many 
 combinations impossible to Greek (which avoids the accumula- 
 tion of consonants at the end of a word), e. g. in ferunt, Jmnc, 
 volt, fert, scrohs, ars, puis, hiemps. Almost any combination, 
 in fact, that could be pronounced was allowed, with the excep- 
 
IV.] Changes and Modifications of Sounds. 73 
 
 tion of doubk consonants (e. g. oss-is, but nominative os ; fellis, Consonant 
 fel) or two explosive mutes, e. g. lact-is, lac ; cord-is, cor. As Loss. 
 far then as the language of the classical Eoman writers is con- 
 cerned, there is less deterioration of final consonantal sounds 
 than in Greek : but there is good reason for supposing that in 
 the pronunciation of ordinary life, in the spoken language of 
 which the plays of Plautus and Terence are the chief written 
 representatives, 'neglect of final sounds^' was more the rule 
 than the exception ; so much so, that upon old inscriptions they 
 are often actually omitted. This is most often the case with 
 the most common final letters s, m, t. The case of final s has 
 already been noticed (p. 67) under the changes of spirants. 
 
 Final m, as is evident from its regular disregard in Latin Pinal m in 
 poetry, must have heen weakly pronounced ; and this is con- 
 firmed by the statements of grammarians, and the evidence 
 of early inscriptions, on which we find such forms as oino 
 (unvm), viro {virum), etc. (cp. Appendix I. Insor. i. 2), and dono 
 dedit^donum dedit. The omission is however rare in legal 
 inscriptions, where greater accuracy was desirable, and in others 
 after 130 B.C., when literature began to insist on precision 
 of grammar afid form; but is found in the vulgar wall in- 
 scriptions at Pompeii, and towards the end of the third century 
 A. D. becomes frequent again. The Italian forms meco, died 
 {mecum, decern) and the Uke, show how completely it must have 
 become ignored in pronunciation in the later Empire : and 
 the history above sketched of its appearance on inscriptions 
 shows how the natural tendency of pronunciation towards 
 phonetic decay was checked for a while during the predomin- 
 ance of a classical literary dialect, only to assert itself more 
 completely in the end. 
 
 (^ Consonantal Change — ^Assimilation. Consonant 
 
 Sounds which require very difierent positions of the vocal Assimiia- 
 organs, or which are respectively tenues and mediae (see above, 
 ch. iii. p. 32) are obviously difficult to pronounce close 
 together; and when two such incompatible sounds would 
 
 ' See Wagner, Introd. to AuLiA., pp. xxix-xxxv, and my Introduction, 
 IV. to Terence, Amdria. 
 
74 
 
 Changes and Modifications of Sounds. [chap. 
 
 Consonant 
 ohttnge : 
 Assimiiii- 
 tion. 
 
 Complete 
 Assimila- 
 tion. 
 
 otherwise come together, the principle of euphony operates to 
 produce such a change in one or the other of the two sounds as 
 wUl make them easy to pronounce in close contact. These 
 changes are included under the general head of ' Assimilation,' 
 by which is implied the change of one of two neighbouring 
 sounds to a sound either the same as or sufficiently like the 
 other to be ' compatible ' with it, and therefore easy of pro- 
 nunciation in close contact. It may indeed happen that the 
 recurrence of the same sound twice is unpleasant to the ear, 
 in which case euphony requires 'Dissimilation,' or change to 
 a sound different from, but compatible with, the sound whose 
 repetition offends : but as there are naturally but few cases in 
 which such repetition of the same sound is unpleasant. Dis- 
 similation plays but a limited part in phonetic change. 
 
 Assimilation is either (a) of the first sound to the latter 
 {regressive, assimilation) ; or (6) of the second sound to the first 
 {progressive assimilation) ; or the two sounds pass into (c) a 
 third (doubled) sound ; or {d) into a single letter. It is also (i) 
 complete, where the assimilated letter becomes the same as the 
 other ; (2) partial or incomplete, where it passes into a similar 
 sound. 
 
 (i) Complete Assimilation: — 
 
 (a) Of the first to the second sound. — In Greek o- to k or p., 
 e.g. evvvij.i=: Fecr-wiM {ves-tis), f/i/ii (Aeol.)=£(r-/ii, a/i/iej (Dor. and 
 
 Aeol.) = a-(7/ie J = ^/i€is; labials to nasals, S/ifia^Sn-iia, ypd^/Ka:= 
 
 ypd(f>-iia, TcTviJ.iiai,:^TeTV(j)-iuu : nasals to liquids (especially a-vv 
 and iv in composition) — (ruWa/ji^dva), wppico, etc. So woaa-i 
 (Epic)=7roS-(ri. In Latin ^ siimmus=^sup-mus, flamma=flag- 
 ma (Jiagrare), puella^=jpuer{^la, esse=ed-se {edo) : and so with 
 prepositions in composition: ad in appello, agger 0, etc., 06 in 
 occurro, officio, etc., sub in summoveo, etc., ec-{eK) in effero, etc., 
 dis in diffiigio, etc., com in corruo, etc. 
 
 (6) Of the second to the first sound. — In Greek (chiefly in' 
 
 Aeolic forms) : Kxivva = Krivyai, eareKKa = tortX-o-a, ivefifia ^ 
 
 fvcficra. In Attic imTos:=t7rFos, cp. 'Ikkos^IkFos (Sanskrit a^vas). 
 
 I Other examples are given by Boby, 'Latia Grammar,' § 34. 
 
IV.] Changes and Modifications of Sounds. 75 
 
 In Latin issirrms^is-tumus^ : so cehrrimus, facinimus^=celer-^^^°''"^^ 
 
 •' change: 
 
 timus, facil-tumus : ferrem, vellem=fer-sem, vel-sem. Assimila- 
 
 (c)' The two sounds pass into a (double letter) third sound. 
 In Greek the sound o-o- (or tt) ^ seems in many cases to have 
 arisen from the combination of the y {j) sound with dental and 
 guttural mutes (i.e. from ry, 6y, Ky, yy, xu)- Thus with 
 dentals we have Kpeatrmv^zKper-yav {icpaT-os), \i<T(Toiw.i=KiT-yofmi, 
 (X1V-7;), Kopia-cro)=Kopvd-ya). In these and similar cases the y pro- 
 bably, through influence of the preceding dental, passed into 
 the dental sibilant o- (our «, or s in 'rise'), which then was 
 either assimilated by, or assimilated, the preceding dental : e. g. 
 \LT-yo-p.ai became \iT-a-o-imt, which, by regressive assimilation, 
 (a) became Xia-a-oimi, or by progressive (6) XiTTOfiai. With 
 gutturals, the y sound changed the guttural to a dental (Denta- 
 lism, see above, p. 50), which then produced the result just 
 described. Thus ^tra-aiv, c\dcT(raip^rJK-ya>v, cXapf-imv (cp. iJK-iaros, 
 e\d)(^-ia-TOs), ava<T(Ta^=avaKya, oa-(ra=:oKyaj VOX; 7reiT(ra>=:iTeKya> (root 
 rrcK, coq.), and so with many verbs whose present tense ends in 
 -o-o-o), but the stem in a guttural — e. g. npda-a-a, (npay-), ^paa-a-a 
 ((j>paK-, Jj3,tin far c-io), nTvira-ai (tttvx-'j), dXXdo-cro) (dXXay-ij), Kqpiairai 
 (Kr]p-vK-os), Tacro-co (ray-oy), \eia(Ta> (XeuK-os), nTija-dto (orraK-oi'), 
 Tapd<T<Ta> (rapax-rj). 
 
 In Latin the t of the suffixes -tus (participial) and -tor with 
 the final letter of the root (especially if a dental) passes into ss, 
 e.g. fissus—fid-tus, cassus (Cic.) =.cad-tus, divissum (Cic.)=: 
 divid-twm, fossor=fod-tor. As to the exact process of the 
 change, there are two different views : — (i) Corssen, Schleicher, 
 Curtius, and other leading philologists, assume that it is the 
 result of progressive assimilation, the dental of the root being 
 first weakened to s (because the Roman ear did not tolerate 
 two dental mutes coming together), and the following t as- 
 similated to this s ; the change of < to s in cases like Tner-svm,, 
 la/p-sum, etc., where there is no dental at the end of the stem, 
 
 * On ttis and a rival explanation see below, ch. vi. 
 
 ^ On the origin of aa (tt) see Peile, oh. viii. pp. 387-390. A fuller, but 
 (the book being out of print) less accessible discussion of the point is given 
 by Curtius, ' Tempora und Modi,' pp. 99-no (on the formation of verbs in 
 -aaa, -ttoi). 
 
7^ Changes and Modifications of Sounds. [chap. 
 
 consonant being due to false analogy. (2) The other view (expounded 
 AssimUar ^7 Mr. Roby in the Preface to ,his Latin Grammar, pp. Ivii-lxi) 
 is that tt, dt became first ts, ds, and then ss or s; this second 
 change being due to the fact that ts, ds were in Latin 
 ' unstable ' combinations likely to be soon changed, whereas st 
 (the assumed result of the first stage in the process of change of 
 tt, dt on the other view) is a perfectly ' stable ' sound, easy 
 to pronounce and very common in Latin, for any further change 
 of which there would be no phonetic reason. If, for example, 
 tond-tvm, had (as on the other view) become tons-tii/m, this latter 
 need have undergone no farther change (except perhaps to tos- 
 tum, which in fact did result from tors-twm, the supine of torreo, 
 stem fors-)'. Other arguments urged against the first view, 
 are (a) that it does not account (except on the arbitrary sup- 
 position of ' false analogy ') for the supine in -sum fii'om stems 
 ending in Ig, rg, II, rr^ — cmmmi (curr-d), 'mul-su'm-=mMlg-twm, 
 etc., and from a few other verbs {labor, jubeo, premo, maneo, 
 Jiaereo, etc.) whose perfect active is found with -si : (5) that the 
 progressive assimilation which it supposes, though possible, is 
 very rare in Latin ; (c) that stems originally ending in s do not 
 follow the prescribed change from st to ss : e. g. ges-tum does 
 not become gessvm,. 
 
 (d) Two sounds coalesce into one letter in Greek, when 
 dental and guttural mediae (S, y) are followed by y : e'. g. c^o/iat 
 =e8-yo-fuu (root sed- as in Latin), S^a (root 08-), a-xif'"='''X''^-2/<» 
 (cp. Latin scid- in scindo), Zeis:=Ayeiis, Sanskrit Dyfi,us. f is 
 thus a compound letter = Sy and then Ss, s being the weak 
 sibilant («), and hence in prosody lengthens a preceding short 
 vowel. As with xy (see above, p. 75) so yy became 8y, and 
 this C' thus o-oKm^a^a-aKmy-ya, (r(j)aCa=ar<f)dy-ya) (stem as in 
 2 aor. e<T(j)dy-rjv), Trefos is TreS-iii, pi^a is Fplhfa, <f)pa^a> is <^pa&-ya 
 (TTe-cjipaS-ov). In the Boeotian dialect y was assimilated to 8, 
 producing 88 — e. g. ^paS-Sa, traXn-i88o) — or 8 initial, Aeis, Zeis. 
 
 Incomplete (2) Incomplete Assimilation :— 
 
 As8imiJa> ' 
 
 tion. (a) "Where the two sounds only approximate to each other, 
 
 the change not being so fully carried out. This includes 
 ' Mr. Peile (Introduction, p. 396) prefers this view to Corasen's. 
 
IV.] Changes and Modifications of Sounds. 77 
 
 (a) all those ' euphonic ' changes by which the final letter Consonant 
 of a root or stem is made to correspond with the first Assimiiet- 
 letter of a termination either as tenuis, media, or aspirate — 
 it being easier to sound two tenues, two mediae, or two as- 
 pirates together. Thus in Greek before dental tenues, mediae, 
 or aspirates (r, 8, 6) only tenues, mediae, or aspirates re- 
 spectively of other organs can stand; and the only allowable 
 combinations are kt, wr, ySy 08, x^> </>^- Consequently we have 
 
 from ffXeKcB, irkex^rivat not ■nktuBrivai. : from Xeyffl, XfKTOs not 
 Xey-Tos, \ex^^^"-i not ^eydrjvai : from 8ex<'/««> Sf"""* not Sexros : 
 from ypa(j}a, ■ypaTr-ros not ypa^TOS, ypi^&rjv not ypoKpSriv. So in 
 
 Latin from ago we have actus not ag-tus : from traho, tractus 
 not trah-tus : from lego, lectus not leg-tus. In most of these 
 cases, actual pronunciation of the words will make the phonetic 
 reason for the change clear. 
 
 Before the tenuis a-, y and x become k, and and (j) become 
 jt: ko- is then written |, and iro-, i/r. Thus from aya, Sy-a-a 
 becomes oKo-a {a^a), cp. recsi [rexi)=reg-si from rego : Sex'>f""' 
 SiK-croimi (Sc|ofiai), cp. traxi^=trah-si from traho: and so too 
 with the futures of Tpl^a and ypd(j>o, or perfect of scribo. 
 
 (6) Nasals often influence the preceding sound. Thus in 
 Greek before p. a guttural tends to become y, a dental to become 
 o- (the dental spirant). So we find 8«ay/i6s not SmK-pos (Skok-oj), 
 fic^peypai not ^i^pexp^ai (/Spe'^o)), 1(r-fiev not iS-fiev (vfiS of o'So), 
 ^wa-jj/u not ^WT-pai. (di/ijTi»), TTemurpat. not irhruB-pai (jreid-O)). 
 A labial before ;« becomes /x by complete assimilation (see above, 
 p. 74). In Latin so»i-«Ms:=s()p-nMS, *S'aniMiM»i=:<Sa6(i)mw»i. 
 Nasals again are affected by a following consonant : thus in 
 Greek v before gutturals becomes the guttural nasal y (troyKoXcai 
 for (fw-KoXem) ; and the labial nasal p, before labials [epireipos, 
 e/xi/fuxos) ; before liquids, as we have seen, it is completely 
 assimilated (p. 74). So in Latin n before a labial mute 
 or nasal became m (impello, imhuo, immotus), though this 
 tendency of pronunciation did not at once affect the orthography 
 of classical Latin '. 
 
 • On this point see Munro'a ' Lucretius,' Introd. to Notes, I. (vol. ii, 
 p. 26, 1st ed.). 
 
78 Changes and Modifications of Sounds. [chap. 
 
 ChanRe of t (c) The change of t to o- before t in all Greek dialects but 
 Doric [e. g. in 3 singular -n, cjirjo-l, Doric ^oti,- cp. ia-rt; in 
 abstract substantives in -a-i-s = n-s, (fida-is, Homer and trage- 
 dians ipans; when suffix ya{Ja), to follows t, e. g. iTKov(r-ios fifom 
 jrXoCr-or, hiiaitrios, yepova-ia^yepovr-ia ; and in forms like etKoal, 
 Doric FiKOTi, Sanskrit Vinsati, Latin mginti, and 3 plural 
 <})tpovin^4'^pova-i=<l>epovn (Doric)] is perhaps a case of assimila- 
 tion, occurring first in cases where i with a vowel following 
 represented the semi-vowel y{j) sound (e.g. jrKoiKnos^TrXovT-yo-s) 
 and exercised an assibilating influence upon t, and then 
 extended to all cases of t followed by i, in a preference for 
 the softer sound o-. A similar change of 5 to o- before i in 
 the Laconian and Boeotian dialects is evidenced by Aristo- 
 phanes (Lysistrata 86 vai rm a-im, cp. also Ach. 906). Similarly 
 in late Latin, and in the modern languages derived from it, i 
 following t, c, d, g assibilated the preceding consonant, so that 
 by the seventh century a. d. -tio, -cio were both pronounced -sho 
 (whence our pronunciation of words like nation, musician). 
 The Italians, again, pronounce ci like English ch, gi as j, and 
 have Marzo from Martins, palazzo from palatiwm, mezzo from 
 medius ; while the French have assibilated c before other 
 vowels also, e. g. chamhre from camera. This assibilation of 
 ci, ti is sometimes assumed to have taken place ia classical 
 times, from the confusion between -cio and -tio found in the 
 MS. spelling of such words as eondieio ; this confusion being 
 further applied as an argument for the soft pronunciation of 
 Latin c before i^. But this variety of spelling in MSS. is 
 due partly to doubts as to etymology, partly to the assibilation 
 of' ci, ti in popular pronunciation at the time when the extant 
 MSS. were written. Inscriptions (by far the most trustworthy 
 guide in orthography) show no such variety of spelling till 
 comparatively late times, the . change of ci and interchange of 
 ci and ti not appearing much before the seventh century A.D., 
 and then chiefly in Gallic inscriptions. The change of ti (to si) 
 was earlier and more general in the vulgar Latin and other 
 
 ' Eoby, 'Latin Grammar,' Preface, pp. xlviii-1; Wordsworth, 'Frag- 
 ments,' Introd. iii. §§ 23-26. 
 
IV.] Changes and Modifications of Sounds. 79 
 
 Italian dialects ; but (according to Corssen, who has gone most Consonant 
 elaborately into the evidence ') it was not established in the Assimiia- 
 speech of educated Romans till the fourth century a.d., though 
 traceable much earlier in isolated forms, e. g. Acherunsius for 
 Acheruniios, Hortensius (in' old Latin Hortentius), and many 
 names of towns in -Msio, -edo, cp. with others in -entio, -untio ; 
 compare also viciens from vicesiens=vicensiens for vicentiens. 
 There is no variety, in the most trustworthy inscriptions of 
 earlier periods, in the spelling of such words as dido, condicio, 
 solacium, patricius, tribunicius, contio, nvmtius, indutiae, otium, 
 negotium, setius. 
 
 (4) Dissimilation. 
 
 Dissimilation, or the euphonic change of one of two simZor Consonant 
 sounds whose concurrence displeases the ear, is, as has been Dissimiia- 
 said, of comparatively rare occurrence. One regular case in 
 both Greek and Latin is the change of a dental mute before 
 another dental mute at the beginning of a sufe ; tt, 8t, and 
 6t becoming or; rd, S6, 66 becoming o-fl. Thus a.m<TTos^= awT-ros 
 (avvTo) \ i(rTa>p= f'lBTOip ; mmos, 7rciiT6fivm=7ri6T6s, 'irfid-6rivm : in 
 Latin claustrum=^claud-trwm,, equester =: equit-ter, est^ed-ti 
 (edo). In Greek, again, one of two aspirate sounds close 
 together is 6ften dissimilated : e. g. 6l-di)ixi becomes Tt-6rjiJ.t, 
 c-6v-6rjv becomes irvdrjv, and -61 of imperative Kkv6t becomes n 
 from the preceding aspirate in tu<^5?/ti, iTa>6r]Ti. In the redupli- 
 cated syllable of verbs beginning with two consonants, the 
 consonant sound is lost (e. g. ektoi/o for k^-ktovo, tyvioKa for 
 yiyvaKo) probably from the tendency to Dissimilation. Lastly, 
 in Latin the termination -alis is changed to -oris when an 
 I precedes : e. g. mortalis, lateralis, but puellaris, popularis, 
 volgaris ; and Pa/rilia a variety of Palilia {Pahs). 
 
 Besides the changes which result in the substitution of a changes due 
 
 -. ,, .. T . 1 . toindistinct 
 
 weaker for a stronger sound, there are others which seem to utterance. 
 be due to indistinctness of utterance, in the pronunciation of' 
 words without sufficient clearness and sharpness to give each 
 letter its proper sound. 'In this case,' says Mr. Peile, 'no 
 other recognised letter is at first heard; but an indefinite 
 ' ' Feber Ausspraohe,' etc., i. pp, 49-67. 
 
8o Changes and Modifications of Sounds. [chaf. 
 
 amount of indistinct sound is produced after the letter thus 
 slurred; which in time, if this relaxed pronunciation become 
 common, often takes the form of the nearest sound in the 
 existing alphabet^ Thus two letters grow out of one; and 
 a word is often actually increased.' As examples of this 
 introduction of additional sound through indistinct pronuncia- 
 tion, we have (following Peile's enumeration), (i) 'LabiaUsm,' 
 the change from k to p, (2) ' Dentalism/ the change from kio t 
 Parasitica [for both these phenomena see above, pp. .so, f,i\. (^) The in- 
 sertion of a parasitic a before y or i. ay, we have already seen 
 (p. 76), becomes f by partial assimilation of y to the weak 
 dental spirant z : and when we find in Greek ^vy-hv, but in 
 all the cognate languages y of root yug or its regular substitute, 
 the conclusion seems warranted that somehow or other a d 
 sound, not radical, became heard before the y, and that thus 
 this combination dy was avoided by passing to f, as in the cases 
 already noticed. Curtius ('Griechische Etymologic,' p. 551 sq. 
 second edition) gives examples of various forms arising, as he 
 thinks, from the combination of y with a parasitic d arising 
 from indistinct articulation : e. g. (a) C in fuyoi', in CvH'^" (root 
 yam), in fa/iAr and Cvfo) (Sanskrit ytisha, Latin ius), in none of 
 which is 8 radical. The double verb forms -afa, -cuo are also ex- 
 plained by Curtius on the same principle : -aa being a variation 
 from aya with the loss of y, it is assumed that before y feU out 
 it may have given rise to a parasitic 8 — a very ingenious and 
 not impossible explanation. (6) 8(, in the adjectival termination 
 -Sw-s, which Curtius regards as arising from the common -to- 
 or -yo-; this termination -810- being always preceded by a 
 vowel, after which the sounds 10 would be difficult to pronounce 
 clearly. (Others, however, consider that -810- is weakened from 
 original -ryo-: and the etymology is at best very doubtful.) 
 The same applies to a few terminations in -Seor, e. g dSekcjji-Seos, 
 where 8 is not radical, but an original ryo might also be 
 assumed, (c) dy doses the original y, so that parasitic d only 
 remains, e. g. in the Boeotian SuyAK for C^yov, iepdSSm for 
 Upabya (lepdfo)), or Upaya (lepam). bvyiai is strong evidence for 
 the theory of the rise of f in ivyiv. but we can hardly feel 
 
IV.] ; Changes and Modifications of Sounds. 8i 
 
 enough, certainty either as to orisfinal form or meaning to Parasitic <« 
 
 1 , , ., . , ^ . ,. beforej^ori. 
 
 pronounce a verdict upon other cases to wnich Curtius applies 
 his theory — e.g. the particles Si;, 8riv and suffixes -8e and -fe as 
 modifications of (p)ye from the pronominal root ya ; or adverbs 
 in -Sov, -driv, -8a, patronymics in -8a, and nominal stems in -a8- or 
 -j8-, as arising from the adjectival suffix ya (lo) with a parasitic 8. 
 The rarity however of d and great frequency of i/{j) as an 
 element in stem-formations of Indo-European languages make 
 it difficult to find any other way of harmonising these Greek 
 forms with those of kindred languages : and it is fairly urged by 
 Curtius and his supporters that a process which every one allows 
 in some cases (e.g. ^vyov and 8vy6v, cp. with iug-um) is at least 
 possible in others'^- 
 
 (4) The aspiration of unaspirated letters (in words where Aspiration of 
 
 ^ ' * '■ ^ unaspirated' 
 
 none of the cognate languages exhibit an aspirate or its sub- letters. . 
 stitutes) is found to some extent both in Sanskrit and Greek ; 
 a parasitic h being produced, most commonly by influence of 
 an adjoining nasal or liquid or preceding cr, as in 4)pm8os {ttpo), 
 K\ei6poi> (the suffix -Tpov), Tecj>-pa (Latin tep-eo), 'Xix-vos (Xuk-, 
 luc-eo), e^atipvrjs (i^amvrjs), (rxiC'" {scid in sci(n)do), and perhaps 
 a6k-va> (if a strengthened form of sta which in Sanskrit becomes 
 stha). In other cases no cause for the change is apparent 
 beyond mere laziness operatkig irregularly, and affecting only 
 some words permanently, e.g. ^\ecf>apov, <ro<^or, a-aip^s (sap- of 
 sap-io). In Latin the aspirates had early disappeared ; but 
 irregular aspiration at the beginning of a word seems to have 
 been known both in Latin and Greek. Both peoples left out the 
 aspirate where it ought to begin a word, and in both there was 
 a tendency to replace it where it had no right to be ; just as in 
 vulgar English the h is often regularly dropped, and almost 
 as regularly inserted before a vowel where it is not required. 
 There seems to have been a tendency in Greek to aspirate an 
 initial v, e.g. v8<op, imo, va-npos, a tendency which is intelligible 
 if we suppose the sound of v to have been something like 
 
 ' Examples of a similar phenomenon in other languages are — Italian 
 diacere, diacinto, maggiore (from Latin jacere, hyacinthus, major) ; Modern 
 Greek Sidm ^(oidxioy); Gothic daddja (0. H. 6. tajH ; Skt. dhayami). 
 
Sa Changes and Modifications of Sounds. [chap. 
 
 German m, which is difficult to pronounce without a breath 
 slipping out before it. The Aeolic Sfi/ies (17^"*) is probably 
 right : rj/iets resting on a false analogy from i/ieis, where ' repre- 
 sents y. In other cases, e.g. mrros, cp. with 'Ikkos, equus, etc; 
 fas in Attic Greek (other dialects having smooth breathing); 
 ifxapreiv (Attic); cp. with rjii^poTov (Epic) ; there seems nothing 
 to account for the aspirate, which is perhaps due to mere mistake. 
 lAspiration In Latin the insertion of h was of later date, never being 
 
 in Latin. -r. i t • • • 
 
 found, according to Corssen, upon Republican inscriptions. 
 After jp, e, t, r it occurs chiefly in Greek words, but not before 
 100 B.C. and not generally before 50 B.C. About this period 
 there appears to have begun a tendency to insert in pronuncia- 
 tion, and also in writing, a superfluous h : thus Cicero (Orat. 
 48, § 160) says that at one time he spoke as the old Eomans 
 did, pulcros, Cetegos, triumpos, but afterwards conformed to 
 the ordinary practice and said Pyrrhus, Phryges (not as Ennius 
 wrote, Burrus, Bruges) ; but still sepulcra, coronas, lacrimas, 
 etc. Catullus wrote a well-known epigram (Ixxxiv. ed. Ellis) 
 ridiculing the pronunciation of chommoda, hinsidias, etc. : 
 and according to QuintUian (I. O. i. 5, 20) some inscriptions 
 in his time had choronae, chenturiones, praechones. In late 
 inscriptions (fourth cehtury a.d. and onwards) the utmost 
 irregularity is seen,;^ being omitted and inserted almost at 
 random, e. g. hoc (ac), hornat, Jiextricata, haditus, hauctoritas, 
 omini, abitat, inospita ; from which we infer great confusion 
 and uncertainty in the use of the aspirate in the ordinary pro- 
 nunciation of those who cut the inscriptions. Pinally in modem 
 Italian the h is not sounded at all. 
 
 In MSS. of the best classical authors and in the writings 
 of grammarians there is a good deal of uncertainty in the 
 spelling of particular words, the errors being more often in 
 omission of h (from reaction, probably, against the tendency 
 noticed by Cicero and Catullus). The preferable spelling, for 
 example, of the following words is harundo, harena, heres, holus, 
 hordeum, haruspex, Tiedera, erus, uments, umor : but all these 
 ate constantly spelt otherwise in the best MSS.^ 
 
 ' See Munro's 'Lucretius,' Introd. to Notes, I; Kennedy's 'Virgil,* 
 Appendix E, pp. 607-609 (ist ed.). 
 
IV.] Changes and Modijieatioiis of Sounds, 83 
 
 (5) Auxiliary Vowels (prefixed or inserted^ Auxiliary 
 
 A M- /• • \ 1 , 1 .• • -i ■ • vowels. 
 
 An auxiliary (inorganic) vowel, purely pnon-etic in its origin, 
 
 is found most frequently before X, p, n, v, and rarely before 
 
 an explosive sound ; never before t, it, ^. A protracted or 
 
 fricative sound has something of a vowel character (see above, 
 
 p. 32) about it, and it is therefore easy for a vowel to slip out 
 
 before such a sound; whereas before a momentary (explosive^ 
 
 sound the vowel must be deliberately and consciously sounded. 
 
 This additional ('prosthetic') vowel is sometimes found at the 
 
 beginning, sometimes in the middle, of a word ; oftenest as a 
 
 or e, less often as o or i, seldom as v. Examples in Greek ; 
 
 (a) initial; d-arfip (cp. Latin stella>-=^ster-ula, Vedic Sanskrit 
 
 star, German Stern, our sta/r) : a-o-jralpa (an easier form of 
 
 o-naipa) ; i-'Xaxos (Sanskrit laghu-s, Latin levis=legu-is); i-iie, 
 
 i-pM, cp. with /xc, poi (stem ma) ; dBe'Xa, ^eXw ; the Homeric 
 
 iftpyeiv, iFfiKocri, iFeporj, etc, ; o-vopa (Sanskrit namau, Latin 
 
 nomen); Sp.ipa\os=:o-vd(j>a\os (navel); o-Boiis (stem oSovt-, Latin 
 
 dens, Sanskrit dantah). In these and similar cases (a limited 
 
 number in all) the vowel seems to be merely phonetic, the result 
 
 of careless articulation. 
 
 (6) Medial. Here the case is not always so clear, because 
 
 the fuller form may sometimes be the older and have lost its 
 
 vowel. Thus 6peya> quoted by Schleicher (Comp. § 46), as referable 
 
 to a root arg with e inserted, is as likely to be from a root rag 
 
 (Latin reg-o) with an initial prefix o. aK{e)yeivas, ^\-v-6ov (root 
 
 e\6-), aK-i-^a> (oXk!], Latin are-eo), are more probable cases. In 
 
 the conjugation of many verbs we find a secondary stem formed 
 
 by the phonetic addition of e alternating with the original stem. 
 
 Sometimes the enlarged stem forms the present, the shorter 
 
 stem the other tenses, as yrjO-, yr]6i-a>, yeyq6-a; 8ok-, SoKe-a, 8e- 
 
 Soy-p/u : sometimes vice versa, as pax-, pdx-o-pai, i-pax^-a-aprjv, 
 
 o'X-j o'Lxo-pai, olxri-a-opai. (Curtius' Greek Grammar, §§ 325, 6). 
 In Latin there is but little evidence of a vowel as a phonetic 
 
 prefix ; enim (cp. nam) and e-quidem {quidem) being almost the 
 
 only instances. 
 
 (6) Insertion of Auxiliary Consonants. Auxiliary 
 
 In Greek between vp, pp, p\; ap-S-pbs=dvpos (stem dvep-); 
 
 G 2 
 
84 Changes cmd Modifications of 8ov/nds> [chap. 
 
 fieoTifi^pla ^ iiearjiipia {fmepa) : aii^p&ros^anpoTOs (stem ppo-, 
 Latin mor-); /iEjuj3Xti)(ca=/iic/iXti)Ka (stem po\-). ^poros is for 
 fi^poT6s=:p.poT6s ; /3X/tt(b for it^Xirra^p^XiT-ya ; in both cases 
 the ^ is parasitic. 
 
 In Latin p between ms — hiemps, swmpsi. 
 
 In modern languages French gendre (gener), nonibre {nwne- 
 rus); English humble {hwmUis), Ambleside {=Hamal-seat ; 
 Hamal being a Norse name), are examples of similar phonetic 
 insertion of h, d. 
 
 In the foregoing pages an attempt has been made to refer 
 
 most of the changes of sound that have been noticed to one 
 
 uniform principle, viz. the tendency to weak articulation and 
 
 National the desire to secure the easiest pronunciation. But in tracing 
 
 peouUaritKS ^ ° 
 
 of utterance, the operation of such tendencies it must be remembered that 
 the diflSculty of uttering a particular sound varies with different 
 tribes and nations. It varies, as we know, with different 
 individuals sometimes from organic defect, sometimes from 
 want of practice ; and such varieties of pronunciation, unless 
 deliberately corrected or successfully fought against, become 
 permanent peculiarities '. Hence (to" take examples from Eng^ 
 lish) we have people who cannot pronounce r, who ' lisp' ' the 
 sound of s as th, who pronounce i; as w and vice versa. And 
 so with nations ; certain sounds or classes of sounds are pre- 
 ferred or avoided'', are more or less frequently or seldom 
 pronounced : and in this way, on the separation of different 
 tribes from a common stock, the same words take difierent 
 shapes among different tribes, the ambiguous or intermediate 
 sounds being differently fixed or differently developed. In 
 Professor Max Miiller's Lectures, Series II. Lecture iv. pp. 
 171-183, etc., will be found a number of illustrations (a) of 
 the absence or presence of certain sounds in the speech of 
 particular nations, (b) of the different shapes which the same 
 root exhibits in different languages ; from which a few selections 
 are here made. 
 
 (a) The dentals seem to be the easiest sounds ; they are the 
 
 ' See Max Muller's ' Lectures,' II. Leot. iv. 
 
 ' Whitney, 'Life and Growth of Laoaguage,' p. 72. 
 
ly.] 
 
 Changes and Modifications of Sounds. 
 
 85 
 
 most universally employed and are the first uttered by children. 
 But it is said that the dental media d does not occur in Chinese 
 nor in three American dialects. Again, some of the Polynesian 
 (Turanian) languages ' have no gutturals, and some North 
 American dialects no labials :• while in the language of the 
 Sandwich Islands the gutturals and dentals are indistinguish- 
 able. The tenues and mediae are not distinguished in the 
 Polynesian dialects, and are often confused by the Welsh, 
 who say Tavit for David, pet for led. Sanskrit shows many 
 weakened forms of consonants, due perhaps in some measure 
 to the effects of the eneirating climate of India: e.g. the 
 palatal sibilant ^ (s' or s) which arises from careless pronuncia- 
 tion of k without bringing the root of the tongue firmly against 
 the back of the palate; or the 'palatal' sounds ^, *I (Jc, g) 
 which are weakenings of k and g respectively. Sanskrit has 
 the aspirated mediae gh, dh, bh, which were difficult sounds 
 to most other Indo-European nations (see above, p. 34). Greek 
 retains the aspirated tenues Xt ^j 'P- Latin has neither. The 
 comparative peculiarities of Latin and Greek with respect to 
 final sounds have already been noticed (p. 7 2). 
 
 (5) The variation of the same root in different languages Grimm's 
 may be illustrated by 'Grimm's Law' of regular interchange 
 between (i) Sanskrit, Greek, and Latin taken as one group, 
 
 (2) Gothic and Low German dialects (including English), 
 
 (3) High German and its stock (including modern German); 
 the one having an aspirated mute (or fricative representing the 
 aspirate) where the second has a media and the third a tenuis. 
 The following formula will express this law : — 
 
 
 I. 
 
 II. 
 
 HL 
 
 (i) Sanskrit, Greek, Latm . . 
 
 Aspirate 
 
 Media 
 
 Tenuis 
 
 (2) Gothic and Low German. . 
 
 Media 
 
 Tennis 
 
 Aspirate 
 
 (3) High German, etc 
 
 Tenuis 
 
 Aspirate 
 
 Media 
 
 The following table gives a simple illustration, of its workings 
 initial letters being taken as freest from the influence of 
 neighbouring consonants, and dentals as offering the most 
 
Grimm's 
 
 86 Changes and Modifications of Sounds, [chap. 
 
 regular illustration. Fuller illustrations are given in the table 
 below, p. 91 : — 
 
 I. 
 
 rGreek B . . . . 
 
 ■ ILatin / . . . . 
 
 BvyiTijp 
 
 Btjp 
 
 Svpa 
 
 fxeBv 
 
 
 
 fera 
 
 fores 
 
 
 
 i. ■Rnglish d . . . 
 
 daughter 
 
 deer 
 
 door 
 
 mead 
 
 
 3. German t OTth,=t 
 
 tochter 
 
 thier 
 
 thor 
 
 meth 
 
 
 II. 
 
 
 
 
 
 
 rGreek S . . . . 
 'iLatin d . •, . . 
 
 6Sai/s 
 
 dafiav 
 
 Mo 
 
 ?5-«y 
 
 SSap 
 
 dens 
 
 domare 
 
 duo 
 
 edere 
 
 unda 
 
 ■i. English t , , . 
 
 tooth 
 
 tame 
 
 two 
 
 eat 
 
 water 
 
 3. German 3 or s ' . 
 
 zahn 
 
 zdhmen 
 
 zwel 
 
 essen 
 
 waeaer 
 
 III. 
 
 
 
 
 
 
 rGreek t . . , . 
 LLatin t . . . . 
 
 ri (ffu) 
 
 rpfts 
 
 
 t4 
 
 
 tu 
 
 tres 
 
 tenuis 
 
 is-tud 
 
 frater 
 
 i. English ih . . . 
 
 thou 
 
 three 
 
 thin 
 
 that 
 
 brother 
 
 3. German d . . . 
 
 da 
 
 drd 
 
 dimn 
 
 das 
 
 bruder. 
 
 The principal exceptions to this law of change are thus 
 classed in Perrar's Comparative Grammar, pp. 34—38. 
 
 1. Onomatopoeic and imitative words, and natural sounds : 
 e.g. v\aKTS>, English howl, German heulen ; itKayvfj (Latin clangd), 
 English cla/rik, clatter, etc.. Old Norse hlaha ; /idju/uj, awna, 
 {mamma, papille), English mamma, papa, German amm^, 
 (from the natural sounds ma, pa) ; Sanskrit t&ta (dear), Greek 
 Ttrra, TiTBt], &c., English tit, teat, Old High German tutto 
 (breast). 
 
 2. Borrowed words, in which the sound of the original 
 language was naturally retained. 
 
 3. Eegular exceptions in the consonantal groups sJe, st, sp. 
 Thus in Latin stella, English star, German stern, the st sound 
 is identical ; the hard s, in fact, not being easily pronounced 
 with any sound but a hard one. 
 
 4. Where sounds have been irregularly changed within the 
 same language. Thus the change from Sanskrit dv&ra to 
 
 ' So a Greek aspirate frequently corresponds to Latin s (see above; 
 p. 66). 
 
IV.] Changes and Modifications of Sounds, 87 
 
 English door, Grerman Thor (Old High Grerman tor), would be Grimm's 
 exceptional, did not 6vpa, fores show that d of dvHra represents 
 an original aspirate dh. So in Sanskrit budhna (depth), 
 (English bottom. Old High German bodam), b represents bh 
 of Indo-European bhudJma. 
 
 The process of this ' Lautverschiebung,' or Dislocation of 
 Consonants, between the languages in question, is thus traced 
 by Professor Max Mliller (Lectures, Series II. Lecture v). 
 
 T. The physiological analysis of sound shows, at each of the Original pro- 
 
 ji • i p 1 I, ..... cess of the 
 
 tnree points 01 consonantal contact, four possible varieties ofcliangesfor. 
 
 pronunciation — viz. a hard sound (tenuis), or a soji sound Grimm's 
 (media), or aspiration by an audible emission of breath imme- 
 diately after utterance of the hard or soft sound. Thus we 
 have : — 
 
 Guttural k, kh, g, gli. 
 
 Dental t, th, d, dh. 
 
 Labial p, ph, b, hh. 
 
 2. The development of, and maintenance of, the distinction 
 between these varieties of articulation is characteristic of the 
 increasing development of languages, in which new ideas are 
 constantly requiring expression, and the phonetic organs are 
 consequently driven to new devices which gradually assume 
 a settled and traditional form. There was probably a time 
 when the Indo-European peoples (as yet un-separated) had 
 no aspirates at all : and while some dialects never arrived 
 at more than one set of aspirates, others ignored them al- 
 together or lost them again in course of time. But it seems 
 likely that before the separation of the Indo-European peoples, 
 some of them at any rate had elaborated a threefold modifica- 
 tion of consonantal contact — tenuis, media, and aspirate — thus 
 securing in many cases (e.g. the roots tar, 'to cross,' dar, 'to 
 tear,' dhar, ' to hold') distinct utterances for distinct expressions. 
 The distinction thus gained was kept up in Sanskrit by tenuis, 
 media, and aspirated media (t, d, dh) ; and in Greek by tenuis, 
 media, and aspirated tenuis (r, 8, 6). But in Latin, where 
 the aspirates had not been realised at all, the distinct utterance 
 of the third (or aspirated) variety of consonantal sound would 
 
88 Changes and Modifications of Sounds. [chap. 
 
 Grimm's naturally be lost. Thus (to take a case -where only two roots, 
 one containing an aspirated sound, had to be distinguished) in 
 Sanskrit we have da-d.9,-mi, 'I give,' and da-dhfl,-mi, 'I place;' 
 Greek keeps up the distinction in Sl-Sa-fu and ri-ffrj-fu; Latin 
 is obliged to give it up, and retains only one of the two roots 
 in da-re, ' to give,' replacing the other by different words, such 
 as facere or ponere. But credere, eondere, abdere point back 
 to the root dhd, ' to place,' as having existed originally in Latin 
 as in other cognate languages. The Teutonic tribes again, 
 who had no aspirates, tried nevertheless to maintain the dis- 
 tinction between the threefold varieties of consonantal contact, 
 which had come to them as ' the phonetic inheritance of their 
 Aryan (Indo-European) forefathers :' and it is in their en- 
 deavours to supply the place of the aspirates in words common 
 to them with the other Indo-European nations that Professor 
 Max Miiller sees the first step in the progress of ' Lautverschie- 
 bung.' Where Sanskrit had aspirated mediae, and Greek 
 aspirated tenues, Gothic (like Celtic and Sclavonic) preferred 
 the corresponding mediae. High German the corresponding 
 tenues. None of these, however, borrowed from, or came 
 after, another; they are 'national varieties of the same type 
 or idea.' 
 
 3. Thus far ' Lautverschiebung' is the representation of 
 aspirate sounds by nations which did not possess them : but 
 the stock of common Indo-European words which began with 
 mediae {g, b, d) and tenues (k, t, p) led to further changes 
 in Gothic and High German utterance. These nations having, 
 as we have seen, already used their mediae and tenues respec- 
 tively to supply the place of the aspirates, found themselves 
 in a difficulty. The Goths, for instance, felt the distinction 
 between the two series of consonantal sounds which Sanskrit 
 kept distinct as gh, dh, bh and g, d, b ; but they had already 
 employed the second to denote the first ; and so, in order to 
 keep them distinct, fixed this latter series g, d, b in their own 
 national utterance as k, t, p. Then arose the same difficulty 
 of maintaining distinct the third series of sounds which Sanskrit 
 and Greek had fixed as h, t,^p; and the only remaining ex-» 
 
IV.] Changes and Modifications of Sounds. 89 
 
 pedient ■was to adopt the corresponding 'hard breaths' }i, <^, Grimm's 
 
 , . Law. 
 
 and/. 
 
 Similarly the High German tribes, having taken the sounds 
 ■which Greek took as aspirate tenues x, S, <p, to be h, t, f, were 
 driven to adopt the breaths ch, z, f as the second variety ; ■while 
 for the third variety nothing was left but the mediae, which 
 however in the guttural and labial series have constantly been 
 replaced by Gothic 7i andy; 
 
 If we denote the aspirates by (i), the mediae by (ii), the 
 tenues by (iii), and the breaths by (iv), the following table will 
 exhibit the process just described : — 
 
 
 Original Aspirate. 
 
 Media. 
 
 Tenuis. 
 
 A.'lSanskrit . . . 
 B.J Gothic . . . 
 
 . (i) gh, dh, bh. 
 . (ii) g, d, b. 
 
 (ii) S, d, b. 
 (iii) k, t, p. 
 
 (iii) k, t, p. 
 (iv) h, th, f. 
 
 A.-lGreek . . . 
 C.J High German . 
 
 • (i) X, e. f- 
 . (iii) k, t, p. 
 
 (ii) 7. S. e- 
 (iv) ch, z, f . 
 
 (iii) K, T, IT. 
 (ii) (g), d, (b). 
 h f 
 
 The chief objection to this theory of the changes of Grimm's 
 Law is, as Mr. Peile points out (in note i to his 5th chapter), 
 the want of sufficient motive for the second variation. The first 
 obviously arose from the desire to get rid of the objectionable 
 aspirates, which were expressed by the mediae in Gothic, the 
 other changes following from a desire for clearness. But it is 
 not so easy to see why, when the aspirates were gone, and only 
 the breaths 7i, th, f left, the High German speech should have 
 made any further change. Nor is it easy upon any theory 
 to account for the first variation in the Gothic and Lovy 
 German dialects, from the mediae or sonants g, d, b to the 
 tenues or surds k, t, p ; for all analogy of languages shows 
 that phonetic change is from harder to easier sounds (i.e. 
 from tenues to mediae) and not vice versa, so that ^w, "for 
 example, would be at once assumed to be an earlier not a 
 later form than genus, yivos. The facts are indisputable ; but 
 they seem to contradict one of the most invariable and best 
 authenticated laws of speech. Nor has any theory yet been 
 devised which is satisfactory at all points. We can only say 
 generally of the changes for which we cannot account that they 
 
90 
 
 Changes and Modifications of Sounds. [chap. 
 
 are possibly examples on a large scale of that tendency to main- 
 tain tbe distinctive features of a word against the influence of 
 phonetic decay, an example of which is seen in the ' compensa- 
 tory lengthening' of a syllable to make up for some loss of 
 sound ; and which may often be discerned in a struggle of the 
 intellectual or instructive desire to preserve those parts of a 
 word that are characteristic of its meaning against the physical 
 tendency to reduce the effort of articulation. 
 
 General Table of G-rimm's Law. 
 
 Original Sounds. 
 
 A. 
 
 B. 
 
 Gothic and 
 
 Low 
 
 German. 
 
 C. 
 
 Hisli 
 German. 
 
 Sanskrit. 
 
 Greek. 
 
 Latin. 
 
 I. Aspirates ■< 
 
 rKH 
 
 TH 
 
 .PH 
 
 eii(h) 
 
 dh(h) 
 bh(li) 
 
 X 
 
 e 
 
 h,f{g:v) 
 
 f(d.b)- 
 f(b) 
 
 g 
 d 
 b 
 
 k 
 t 
 P 
 
 II. Mediae • 
 
 Lb 
 
 d 
 b 
 
 7 
 S 
 
 g 
 d 
 
 b 
 
 k 
 
 * 
 
 cb 
 zz 
 f, ph 
 
 in. Tenues ■ 
 
 T 
 
 k 
 t 
 
 K 
 
 T 
 
 0, qu 
 
 t 
 
 t.g(f) 
 th, d 
 
 l>,g,k 
 d 
 
 
 Lp 
 
 P 
 
 IT 
 
 P 
 
 f, V 
 
 f.v 
 
 ' There are few really Saxon, and no Gothic (unless foreign), words 
 beginning with p. In Sansltrit, too, the consonant b, which ought to cor- 
 respond to Gothic p, is seldom, if ever, an initial sound, its place being 
 occupied by v. Hence this particular phase of Grimm's Law is inserted 
 without illuatratiou by both Bopp and Max Miiller, to complete the 
 scheme. 
 
IV.] 
 
 Changes cmd Modifications of Sounds. 
 
 91 
 
 !Z| 
 
 a H M 
 ^ % > 
 
 c 
 
 a' 
 
 1 
 
 3 
 
 : 
 
 t 
 
 > 
 
 1 
 
 52 
 1 
 
 (h)anser 
 heri 
 fel 
 fera 
 
 /ero 
 /rater 
 
 genus 
 genu 
 duo 
 domus 
 
 cor(d) 
 
 quia 
 
 comu 
 
 tu 
 
 tres 
 
 tenuis 
 
 pes 
 
 planus 
 
 1: 
 
 
 Q5* 
 
 w 
 
 II II ^ r 
 S3 
 
 do' 
 
 dor 
 
 yiru 
 pruoder 
 
 chunni 
 
 chniu 
 
 zuei 
 
 herza 
 
 hiruz 
 
 drt 
 
 fuoz 
 vol 
 
 B a 
 
 p 
 
 1 
 
 1 
 1 
 
CHAPTER V. 
 
 FOEMATION OF WOEDS. 
 
 Elements LANGUAGE is made up of articulate sounds combined into 
 words. These sounds, however, convey no meaning in them- 
 selves (except in a few cases of interjectional sounds) : and it 
 is only when words are formed that we have language properly 
 so called, the medium of communication between men, the 
 means of expression of human thought. Thus, although to 
 understand the changes and varieties in the outer form of 
 language, it is necessary to investigate the nature of sounds 
 and their production by the physical organs of voice — the 
 'Phonology' or 'Sound-Lore' of linguistic study; the ulti- 
 mate facts in language regarded as an expression of thought 
 or meaning are words — or rather, the elements, or several 
 combiuations of sounds expressive of meaning, into which a 
 careful analysis shows that all words can be divided — i. e. 
 Analysis of ' Morphology' or ' Word-Lore ^.' These elements are broadly 
 Eadicaiand divided into 'radical' and 'formative'— i.e. on the one hand, 
 elements. ^^^ portion of the word which gives its general meaning in 
 the simplest and most rudimentary form ; on the other, all 
 
 ■ Some references to books which treat more fiilly of these questions 
 than is possible here, may be of service to the student. Thus, on Pho- 
 nology : Schleicher, 'Compendium,' §§ 1-204; I'crrar, 'Comparative 
 Grammar,' ch. i-vi. §§ 1-86; Peile, 'Introduction to Greek and Latin 
 Etymology ;' Roby, ' Latin Grammar,' Book I. §§ 1-302 ; Ourtius, (The 
 Student's) 'Greek Grammar,' |§ 1-99; and 'Elucidations,' pp. 17-47. 0" 
 Morphology: Schleicher, §§ 205-241; Ferrar, ch. vii. viii. §§ 87-127; 
 Koby, Book III. §§ 740-999. In Curtius" 'Greek Grammar' and 'Eluci- 
 dations ' the formation of Noun and Verb stems is treated as a part of 
 Noun and Verb Inflection. 
 
•Formation of Words. 93 
 
 those additions wMch vary or define or restrict this general 
 idea, or adapt the word for its place among, and its relation 
 to, other words combined into a sentence for the expression 
 of thought. The radical element of a word is termed the 
 root : while under the term formative elements are included 
 (i) those modifications of the root either by 'dynamic change' 
 or by the addition of suffixes (themselves originally independent 
 roots), by which it becomes a Noun- or Verb-' Stem;' (2) 
 the inflections expressive of Case, Number, or Gender, Tense, 
 Mood, or Person, by which these Noun- or Verb-Stems are 
 enabled to express so many various shades of meaning when 
 placed in relation to each other as parts of a sentence. 
 
 [It should be noted here, that this division into Noun and Mvision of 
 
 L ^ ' ^ words into 
 
 Verb ('Nominal' and 'Verbal' Stems or Bases) is exhaustive Noun imd 
 of Indo-European words. In all Indo-European languages haustiYe. 
 (and therefore in Greek and Latin) there are originally only 
 two kinds of words distinguished as noun (Si/ofia) and verb 
 (pfjiia). The faculty of language in man leads him first to 
 give names {nomina, opofiara) as signs expressive of conceptions, 
 and then leads him to form verbs {verba, pruiaTo) to express 
 that which 'is said' about or predicated of the conception 
 expressed by names. All other ' Parts of Speech' designated 
 by grammatical analysis have been developed out of one or 
 other of the two main classes of Nouns and Verbs. This is 
 sufiiciently obvious with Adverbs, which are often merely case- 
 forms of existing nouns, substantive or adjective (e.g. SiKijr, 
 instar, torva iuens, irXeiov, n-Xfiora), and can generally be traced 
 back to archaic, or mutilated, or otherwise altered case-forms. 
 The same applies to Prepositions, which grammatical analysis 
 shows to have been originally adverbs \ separable alike 
 from the cases with which they are used, and from the words 
 with which they are compounded in classical Greek or Latin; 
 many prepositions being still used in those languages as adverbs 
 (e.g. ante, cirenm, contra, extra, etc.) So too with Conjunc- 
 tions and aU 'Particles,' though it is not always possible to 
 
 1 See Curtius' (The Student's) 'Greek Grammar,' §§ 444-446; 'Eluci- 
 dations,' ch. xvii. pp. 200-203, 
 
94 Formation of Words. [chap, 
 
 trace the original form in words which, being in very constant 
 use and not as the most essential words in a sentence, are 
 the more liable to corruption and decay in utterance. In words 
 however such as oti, quod, quia, quam it is obvious; que is 
 some case form of qui; hs fs evidently adverbial, and wt is 
 merely its phonetic equivalent; wow = we wnum ; and we, nei 
 is evidently a case form: and similarly, numbers of examples 
 might be produced, were we concerned now with more elaborate 
 proof of the statement here given '.] 
 Eoots. By a ^root' we mean the simplest combination of sounds 
 
 which expresses the general meaning of any word or set of 
 kindred words, e.g. da is the root of Sanskrit da-d&-uii (Si'Sa^i), 
 da-mus, da-tur, etc., Sanskrit da-tar (porrip), etc. : jug oiju{n)go, 
 jug-um (for the nasal sound n in present stem cp. Xafipdva, 
 ?-\aP-ov). 
 
 The formative elements, suffixes and inflexions, which form 
 words from simple roots, are originally independent roots. 
 Thus in SiSam, da-d£t-mi, mi is a weakened form of ma the 
 pronominal element of first person ; in vox {voc-s), Sanskrit v&k 
 (=v&k-s), s=sa demonstrative pronoun. , 
 
 Thus every I. E. word is a whole gradually sprung from 
 several, or at least two ' roots.' The first of these is the ' root ' 
 in the ordinary acceptation of the term, i. e. that which conveys 
 the meaning in general; the others have degenerated into 
 suffixes for expressing modifications of meaning. 
 
 In the 'Isolating' or 'Radical' stage of language, the roots 
 remain separate and distinct : i ma. 
 
 In Agglutinative languages the principal root remains the 
 same, but receives an addition in the form of a changeable 
 prefix, suffix, or infix : i-ma or i-mi. 
 
 The Inflectional, or highest type of language, alters the prin- 
 cipal root (by reduplication or by raising the vowel) for purposes 
 of expression : ad,mi (Sanskrit emi), elm ^■ 
 
 N.B. — A simple root without modification or addition of 
 suffix cannot form a word. 
 
 ■ See Appendix II. 
 
 ^ On the three ' stages ' of linguistic growth, see ch. ii. pp, 4-8. 
 
v.] Formation of Words. 95 
 
 Roots are always monosyllabic ; and are distinguished as :— 
 
 1. Primary ; e.g. i (go), ad, (eat), da (give), yu (join). 
 
 2. Seoonda/ry ; e.g. <M{i (strike), yv^g {jug, yoke, i.e. join), 
 yudh (fight, i. e. join battle), flu (flow), ard (hurt), spac (see). 
 
 These secondary roots are probably in all cases (as evidently 
 yug, yudh, cp. with yu) modifications of primary roots, by the 
 addition of a letter or letters, expressing usually some extension 
 or limitation of the idea ^ The additional element may have 
 been in some cases 'dynamic' (see above, p. 51), in others 
 ' phonetic ' — i. e. a mere change of sound, afterwards turned to 
 account for the expression of meaning, as e. g. the phonetic 
 variation of the a-sound into a, e, o (p. 36). 
 
 The primary roots are the most important in the history 
 of language, but their predicative power being generally too 
 indefinite to answer the purpose of advancing thought, they 
 were to a large extent encroached upon and supplanted by 
 secondary roots. 
 
 Philologists are not agreed upon the exact definition of a Definition of 
 '■root.' Professor Max Miiller (Lectures, I. p. 215) states a 
 root to be ' whatever cannot be reduced to a simpler or more 
 original form.' This, he says afterwards (Lectures, II. chap, iii.) 
 is objected to as making a root a mere abstraction, and so unfit 
 to explain the realities of language ; to which he replies that 
 in one sense a root is an abstraction ; for it is a cause, which 
 we only recognise and arrive at from its effects, viz. words. 
 These efiects we hear in language, but not the root itself. 
 At the same time Professor Max Miiller seems to maintain 
 the real force of roots — e.g. that the root da, in some way 
 or other which we cannot yet explain, has some necessary 
 connection with the idea of 'giving' — and regards them as 
 'phonetic types' which we cannot as yet explain, but which 
 alone give us definite forms on which to rest our enquiries into 
 language, standing between us and the ' chaos of onomatopeia 
 
 ^ e. g. from root tar (whence Teiptu, tero, etc.) we get the modified forms 
 (fro), tri (triticum), tru (rpioi, etc.), torh (torqueo, d-Tp6K-)js = 'not turned'), 
 tram (rpe/m)), trib (rptPco, tribula), trup ^Tpiiravov). Tra is a variety of 
 tar : tri and tru are secondary, by modification of a to i, «. 
 
9^ i formation of Words. [chap. 
 
 and interjections.' It seems however that, as we cannot know 
 with certainty the ultimate form of these roots, our speculations 
 as to the cannection between them and the ideas they express 
 can he little more than guess-work ^ 
 
 Those therefore who disbelieve in the existence of, or at least 
 in any chance of finding out, any necessary connection between 
 roots and the ideas expressed by them, prefer to consider roots 
 as ' mere abstractions,' as ' headings ' of common elements under 
 which to class words belonging to the same family, as ' labels ' 
 or ' tickets ' of classification. After all, they say, we can only 
 find out in many cases the simplest forms in individual 
 languages ; and if in different languages we find different 
 simplest forms or roots to express the same idea, it is hard 
 to assume in the case of any one of them any inherent or 
 natural power by which it is connected especially with that 
 idea. An example of this view may be seen in Peile's ' Intro- 
 duction to "Grreek and Latin Etymology,' where the definition 
 given^by Curtius of a root as 'that combination of sounds which 
 remains when a word is stripped of everythiag formative' is 
 accepted ^. 
 
 I am inclined to think this latter view is the safest, and 
 the most consistent not only with the present state of the 
 science of language but with its future prospects. As was 
 said above (chap. i. pp. 2, 3) we arrive by analysis of language 
 at certain primitive and elementary combinations of sounds, 
 which we cail 'roots,^ and which, forming as they do the 
 common element in groups of connected or kindred words, 
 we speak of as conveying such and such a meaning. But we 
 must now, and I think always, accept these simplest forms 
 as ultimate facts which Philology will never explain to us. 
 Comparison of languages and analysis of words may now 
 
 ' Mr. Peile puts this forcibly and clearly (' Introduction,' p. 42) : 'That 
 there was some connection (between idea and form) originally I believe; 
 but I do not believe that it is ever discoverable with certainty : and that it 
 was ever necessary, I deny.' And in support of this position he aptly 
 quotes M. R^nan's dictum (De I'Origine du Langage, p. 48), ' La liaison du 
 sens et du mot n'est jamais n^cessaire, jamais arbitraire, toujours elle est 
 motiv^e.' 
 
 ^ Ch. iii. pp. 41-44 j(3rd fid,). 
 
'^•J Formation of Words. 97 
 
 and then point to some simpler and more elementary form 
 than has yet been reached : but the prospect of finding out 
 the reason of such forms, and why they came to have the 
 meanings which they have in language, is so remote, if not 
 altogether visionary, that it may for all practical purposes be 
 disregarded. And therefore I prefer such a description or 
 definition of roots as assumes nothing with regard to their 
 inherent power of expressing particular meanings, and whether 
 under the title of 'abstractions,' 'labels,' or 'simplest forms' 
 takes them as facts, but unexplainable facts. 
 
 On this view, then, a root may be defined as ' the simplest Definition of 
 ascertained combination of sounds, which expresses the general 
 meaning of any word or set of kindred words in one or more 
 Indo-European languages ^.' 
 
 Koots are for the most p&rt ' predicative,' i.e. expressive of 
 ideas of action, state, etc. ; but there is a limited number of 
 'pronominal ' or ' demonstrative ' roots (expressive e. g. of such 
 ideas as ' here,' ' there,' ' this,' ' he,' ' I,' etc.), which cannot be 
 traced back to predicative roots and must be considered inde- 
 pendent of those ordinarily so called. These pronominal roots 
 enter considerably into the formation of inflections ^, as well as 
 of the pronouns and pronominal particles (i. e. conjunctions and 
 some adverbs and prepositions). 
 
 ' The distinction between roots, stems, and words may be shortly put 
 thus : — The root is the original part of the word, giving a certain idea ; 
 the stem is that idea more closely defined to a certain bearing of it ; the 
 inflected form (or word) is the complete word as used iii speech in connec- 
 tion with other words in a sentence. Compare ' Elucidations to Curtius' 
 Greek Grammar,' Translator's Preface, p. vii. 
 
 * A list of Indo-European pronominal roots is given in Leo Meyer's 
 ' Vergleichende Grammatik,' I. pp. 324-335 : cp. Perrar, 'Comp. Gram.' 
 § 95. The following are among the more important of these roots and 
 their derivatives : a, whence probably the augment (in Sanskrit a), and 
 possibly ifoi, aafii-s (^/lefs), etc.; i, in i-d, i-pse, i-ta, etc.; hva (who), 
 Skt. ka-s, Gk. ris, Lat. quis; irais. Ion. kois, ica-l (a locative), Kfv ; ta 
 (demonstrative), whence Gk. rhv, t^v, rb, oBtos, etc., Lat. ia-te, ipse, (for 
 ip-te), turn, tarn, item, etc. ; da-, whence iro-Sa-7ris, o-Si, qiumdo, qwi-dam, 
 unde, etc. ; sa (demonstrative), Gk. 6, r), &^a( ; na, an, ana, whence Gk. 
 vii, vlv, vii, vvy, hv, fv, avd, Lat. nos, ne, nwm, in, etc. . pa, in a-ir&, irapoL, 
 irepJ, ab, pro, per, etc.; bha, in Skt. inflections, -bhyas, -bhyam, and 
 -bhis, Gk. ^^t>l, and Lat. -his, -bus of dat. abl. plur. ; ma, tva, and sva 
 of 1st and 2nd pers. and reflexive pronoun. 
 
98 Formation of Words. [chap. 
 
 stems. Stems (also called 'bases' or 'themes') arise from roots by 
 
 modification of the root-vowel, or addition of formative suffixes. 
 Roots express a possibility (potentiality) of action. The stems 
 formed from them denote for verbs the action itself, for nouns 
 the person, state, or thing concerned in or resulting from that 
 action. Thus the root 6?a= giving (potential); da-d&-mi, di- 
 ha-y.1, cZo = ' I give ; ' So-njp, da-tor = the person giving, the 
 giver ; Sd-o-ii = state of giving ; do-nwm = thing given. The 
 stem of a word is most readily detected by observing what 
 remains when the ' inflections ' (i. e. declension or conjugation 
 ending) is withdrawn. 
 
 InSections. Inflections are the alterations in or additions to a word, to fit 
 it for difierent functions as parts of a sentence : the common 
 part which remains the same under these different uses being 
 the stem. Thus in Xdyos, dominus : — 
 
 N. \<57o s. 
 
 G. K6yo-syo, \6yo-iOf \6yo'0 (\.6yov). 
 
 J). K6yo-oif \6yip. 
 
 A. \6yo-y. 
 
 v. \6yo- (stem used inter] ectionally, and o sinks to e). 
 
 The common part \oyo- is the stem : the root is Xey- seen in 
 Xf-y-fl). 
 
 N. domino-s, dominus. 
 
 Gr. domino-i, domini. 
 
 D. domino-i, domino (or illo-i, illi). 
 
 A. domiuo-m, dominum. 
 
 Abl. domijio-o, domino. 
 
 v. domino-, doming (as above). 
 
 The common part domino- is the stem : the root is dom- 
 seen in dom-a-re, befi-av, etc. : -ino- is a suffix added to the root 
 to form a nominal base or stem. 
 Distinction [Note that the stem is distinct from the inflected word, and 
 Word. must not be confounded even with the Nominative Case, e. g. 
 aaxppov- (seen in oblique cases a-&<l>pov-os, k.t.X.) is the stem of 
 (ra<f>pav, npaynar- of Trpayfut ; and Latin words like consul, 
 rrmlier have dropped the final -s indicating the Nominative 
 Case.] 
 
v.j Formation of Words. 99 
 
 So vox:=vdc-s. The root is v8c- (seen in v5c-o) : the stem Analysis of 
 voc- by modification of the root-vowel. 
 
 (pa-Ti-s (speech, report), root (j)a- ; nominal suffix -n-, inflec- 
 tion -s. The same root (pa is lengthened to form a verbal stem 
 (prj-fil : the nominal stem is <^aTu. 
 So in the formation of Verbs : — 
 
 eifu (ibo) : root i (in tfiev) ; stem el, by modification of root : 
 inflection -/ii. 
 
 elju {su'm)=^i<Tfu {Aeolie), Sanskrit asmi. Root and stem (in 
 this case identical) ia-. Inflection -fu. 
 
 '6p-vv-iu, root op-, verbal suffix -m- to form the present 
 stem. 
 
 (ftevy-o) = (f>(iy-a>-iti. Root <f>vy- (in €-(f)vy-ov, 2 aor.) modified 
 ' to form the present stem ; inflection -/u ; ' thematic vowel,' in- 
 creasing the stem before inflection, -a- (appearing also as o in 
 (fievy-o-nev, e in (j)fvy-e-Tf, and 0, i, u in Latin, e. g. Jir-o=:<j)ep-a>, 
 Jer-i-muSj/er-u-nt) ^. 
 
 [Note that in the Conjugation of verbs we must distinguish (Tense- 
 difierent Verbal-stems called generally ' Tense-Stems,' each 
 the' common element of a number of forms of the same 
 verb. Thus in the scheme of Tvirra) we have the ' pure verbal 
 stem ' -Tvir- (seen in 2 aor. e-Tvn--ov) ; the ' present stem ' nmr- 
 common to all forms of present and imperfect tense ; the 
 ' perfect stem ' tctv^- ; the ' weak aorist stem ' -rvijra-, and 
 the ' strong aorist stem' -nm- identical with the 'pure verbal 
 stem.' The fuller consideration of these will fall under the 
 head of Verb-Inflection (chap, viii), and they are enumerated 
 now by way only of illustration.] 
 
 There are four main processes of word-formation from roots ; Word-for- 
 mation. 
 viz. : — 
 
 (i) Redwplication — in imitative names and perfect stems, 
 etc. 
 
 (2) Internal Change by 'raising' or 'intensifying' the root- 
 vowel. 
 
 ^ In the firat edition of this work the earlier view of Curtius that this 
 a (0, e, i, u) is a 'connecting vowel' was adopted ; but has now been 
 abandoned for reasons given below (eh. viii). 
 
loo Formation of Words. [chap. 
 
 (3) Addition of Suffiooes. 
 
 (4) Composition, i.e. the formation of two or more words into 
 one. 
 
 Of these processes (i) and (2) have been considered and illus- 
 trated under the head of ' Dynamic Change ' (chap. iv. pp. 51-55) : 
 (4) is generally treated of in the grammar of each language. 
 We are now, therefore, concerned mainly with (3) Addition of 
 Suffixes. 
 
 This term ' suffix ' is applied by writers on philology to 
 different elements in word-formation, which must be carefully 
 distinguished. Thus we speak of 
 
 1. 'Suffixes^ of Inflection, i.e. the 'Inflections' properly so 
 called ; the case-endings of nouns and person-endings of verbs. 
 These will be considered at length in chaps, vi-viii. 
 
 2. 'Formative Suffixes,' by the addition of which to ' roots ' 
 are formed ' bases ' or ' stems.' And as bases or stems are either 
 verbal or nominal (above, p. 93), so the formative suffixes may 
 be divided into ' verbal ' and ' nominal suffixes.' 
 
 a. The verbal suffixes are chiefly -ya (ja) and aya {-aja), 
 from which are formed the verb-stems of all the contracted (and 
 many other) verbs in Greek, and of the first, second, and fourth 
 ' conjugations ' in Latin [see below, in the Appendix to this 
 chapter]. 
 
 6. The nominal suffixes are more numerous : a list of the 
 more important is given on pp. 102, 103. Not unfrequently a 
 nominal stem is used to form a verb as well as a noun ; e. g. 0w- 
 Xdo-o-o) is formed from ^uXqk-, the nominal stem of ipvka^ {<f>v\aK-s), 
 acu-o from acu-, the nominal stem of acu-s (a needle). Such 
 verbs are called nominal (sometimes denominative) verbs. 
 
 3. Stem Suffixes, a class of verbal suffixes perhaps originally 
 formative like ya and aya ; but unlike these, found only in the 
 present and kindred tenses. Such are na, nu {jiap-va-fiai, 
 sper-no, SfU-vv-iit, etc.) ; ska (verbs in -ctko), -sco) ; ta (p,\cm-T-a>, 
 TtK-T-o), necto, etc.) ; and according to Schleicher the ' thematic ' 
 vowel a (bhar-ft-xni, <f)cp-a>[iu), etc.). Most common however as 
 a stem-suffix is ya (Ja), appearing as i in l8-l-a>, Sa-Ua, etc. ; as e 
 in doKea, yaiiea, and certain other verbs in eu which are distin- 
 
v.] Formation of Words. loi 
 
 guished from the regular formation with aya by having this 
 suffix confined to the present stem ; and passing by assimilation 
 (p. 104) into X\ {paKK(ii:=.^aKya), <r<T or rr ((^vXao-o-a)=0vXaK-y<i), 
 etc., see p. 75), or f (iK'Kl^ai=:e\m'&-ya>) ; and in the verbs in io 
 of the third conjugation in Latin {cap-i-o, fug-i-o, etc.), and 
 (possibly) after assimilation in pello, curro, etc. It seems diffi- 
 cult to believe that this ya is altogether distinct from the -ya 
 or aya which, as we have already seen, is the most common 
 formative verbal suffix; though of course such distinction is 
 possible. 
 
 Of the origin of all these suffixes nothing more is known than 
 that the verbal suffixes were probably for the most part ordinary 
 or ' predicative ' roots, the nominal suffixes for the most part 
 ' pronominal' or 'demonstrative' roots. 
 
APPENDIX TO CHAPTEE V. 
 A. List of Nominai Suffixes ^ 
 
 Derivative i. ya (^ja) (Greek -lo, Latin -io): &y-io-s, iimpa-=.ii6pja, Sa-a-a^ 
 
 oKJa, eadmius, coniugium, (root iug of iugum), ingeriiti/m. As 
 sign of feminine; (l>fpovaa = (j)epoVT-ja, ft4\mva=^n€\avja, 
 
 2. -va {vo), van=Fo, Fov: ala>v=alFi>v, aevum : m-vum (root 
 ar of ara/re), vacuus (vac in vac-m-e). 
 
 For : eldas^ elB-FoT-s. 
 
 3. -ma, -mo, -mon, -mat, -m,eno : Tiprj, 6vp6s, rKiipMv, et/ia=: 
 Fio-pjiT, Spiia^oTTpaT ; forma, animus, ser»rao(n)j par tic. mid. 
 and pass. SM/ievos, alumnus (aXo/ifvos) ; infin. eB-pfvai (Homer). 
 
 4. -cm, -ana, -na : Teprjv (-fvs), jpecten ; Spyavo-v, iKav&s, 
 Ti&ovTi. Infin. -vai, -ivai; \c\onrevM, aTrjvai, (pepeiv ^z (jicpevi = 
 
 (jje'pevm ; donum, somnus (sop-nus) : part, in -dus, -on-do, -en-do, 
 -un-do. 
 
 5. -ta, -to, -tat, in adject., subst., part, pass.,, and verbal 
 adject. : iroKirqi, koXtos, secta ; k\v-t6s, yvoo-xos, ama-tus ; veorris 
 (ve6T7]Ts), civitas (eivitat-s). 
 
 6. -tar, -ter, -tor, -tra, etc., ' in words expressive of relation- 
 ship and nomima agentis : Trarrjp, (ftpirmp, a-iorrip, tortop, larpos, 
 prjTpa : pater, victor (or with additional suffix for fern. vict(o)r- 
 ic-s victrix). Part. fut. stem -tu^o : and feminine nomina 
 actionis ; sapultura, usura (ut-tura). 
 
 7. -ti, -si, etc. in nomina actionis : liij-Ti-s (root /xa), (pd-n-s 
 (fiva-is ; messis {=^m,et-ti-s), vectis, potis, compos (compot-s), dos 
 
 ' Fuller particulars may be found in Schleicher, ' Compendium,' §§ 215- 
 231 (pp. 361-463 German third edition). 
 
Appendix to Chapter V. 103 
 
 (dots), mens (ment-s). Further formations ; -<Tia femin. Bv-ma, 
 boKifiacria : Latin -tio, -tia, initio, servitio, justitia. 
 
 8. -tu, Pparil-s, aa-TV ; -avvri (-Twr)), secondary Suffix in fivrjiio- 
 a-vvr], 8iKaio-(Tvvrj . In Latin mucli commoner ; e. g. verbal nouns 
 in -tus, whence supines in -U7n and -u, dictu, ama-tum, casum^ 
 cad-ium, etc. ; cp. appara-tus, soni-tu-s, etc. of purely sub- 
 stantival use. -iuo (-tva), a further formation in mortuus, 
 statua, etc. : -tu-ti, -tudon or -tu-din in servi-iut-is, dltitudinis. 
 
 9. -ant, -ent, -ont in partic. act. of pres. fut. and aor. i>v^ 
 i(T-6vT-, dovo-a=ze(r-ovT-ja; JcrTas^^tfrrair-s ; 6eis=6evT-s. So Latin 
 -ens^ent-s : e.g. in prae-sens, sens^es-ent-s(e(r-ovT-s) ; and the 
 further formation in praesentia corresponds to Greek fem. 
 
 -ovcra:=ovT-ja. ' 
 
 10. -es, -OS, -us in neuters, yivos, genus (genitive yev({a)-os, 
 gener-is), ijfevSfis (stem -es in neuter yjrtvSes and genitive \jfevSi(a-)os, 
 i^cuSoCs). Masculines in -or=os, sopor, honor, labor [honos. 
 labos). 
 
 11. -ha, -co, Grreek Sfj-nrj, and the common adjectival suffix 
 K0-, (jjvm-Kos, /c.r.X. Latin pau-cus, lo-cus, civi-cus, helli-cus, etc. 
 
 12. -ra, -la, Grreek ipvd-pb-s, Xa/i7r-pA-s, ax-po-s, K.r.\. ; dei-\6-c 
 (root 8t in Se-Si-fUV, Sc-8oi-(ca), a-iyrj-Xo-s ; <J)v-\ti, oplx-^l- -epo- is 
 a common variety of this suffix, (poPepos, Spoa-epos : but the e is 
 perhaps only the of stem (fiofia-, Spoao-. The element -po or -Xo 
 seems to enter into other suffixes, -vpo, -apr], -aXo, -cbXtj, -1X0. 
 Latin rvh-ro-, gna-ro-, ple-ro-, etc. Sella = sed-la (= eS-pa), 
 cande-la, loque-la, ala, velum, etc. The element -la also appears 
 in other suffixes, -ulo-, -ula-, -Hi-, etc. 
 
 B. Deeivativb Verbs in G-eeek and LATI^f, formed by 
 ADDITION OF THE SUFFIX -aya ifljo), or ya (Ja\ 
 
 -aa> ) ,. A • 
 
 \ :=aja> = aya-m% ; 
 
 I . -av '. -w = -6a) ^ -ao) 
 
 -are: -o^=-ao 
 e. g. Sanskrit dain4ya.ini, Greek hapaai, Latin domo (domao), 
 
 Gothic tamja, German zahme. 
 Many derivatives in -av, -are are connected with fem. sub- 
 stantive stems in -a ; e. g. Kopaa, Kop,S), Lat. como, with Kopij, 
 
I04 
 
 Appendix to Chapter V. 
 
 coma, Kojid-ja, comd-jo. Others with -o stems (originally -a) ; 
 e.g. avTiav (avTio-s), armare {armo-)firmare {Ji/rmo-). 
 2 . -eiv : 
 
 -ere: -eo ) 
 
 e. g. apxea, Lat. arceo (arkdjami). 
 -do) = oja = ay a/mi. 
 -ia> (or /fo>) ) , . 
 -io 7='-^-=' 
 
 = ayami. 
 
 
 ujomi. 
 
 3- -ow, 
 
 4. -UlV, 
 
 -ire, 
 
 5. -UflV, 
 
 -i«ere, -uo , 
 
 So in Sanskrit gAtu-yami (jraiit), Greek yjjpvQ) {=:garujomi), 
 stem yrjpv-. 
 
 In these the f arises from the effect of the j (y) 
 
 sound upon a preceding consonant, guttural or 
 
 dental J e.g. dp7rd^eiv:=dp7rdy-jciv (Apnay-fj), ^avfid^eiv 
 
 = davfidd-jeiv (6avpaT-jj 'jrie^eiv = iney-jeiv (wcTTtey- 
 
 fiai, imi^-driv), oljii>^ta>^olfwy-jfiv (olpay-rl), ATrifeij/ 
 
 ^D^jriS-jciu, xapl^eirOai, {j(dpiT-os), oXoXufciK {SKoKvy-ij). 
 
 -traeiv : (r(T=KJ, yj, xj> ''j, ^j- e.g. Baipria-creiv (stem BaprjK-), dK- 
 
 \do'(rciv (dXXay-ij), opvaireiv (op^X'v)) Kopitrtriai (jcopvB-), cpto-- 
 
 a-eiv (epcr-ijy, fpeT-ftos). See pp. 'J4—'j6. 
 
 -atp€iv^dp-jetv \ The _;' (y) sound heing thrown hack into 
 
 -eipeiv^ep-jeiv > the stem syllable and becoming the vowel 
 
 -vpeiv ^vp-jcui J sound of i (cp. pi\mva:=p.£Kav-ja, p. I02). 
 
 -aXXeiv \ . 
 
 Probably from X_; : but as no noun-stems end in 
 
 X {oK-s excepted), these are derivatives from stems 
 
 in -Xo, the stem-vowel o being lost. 
 
 6. -dfetv 
 
 8. 
 
 10. 
 
 j {y) sound thrown back as vowel into the stem 
 syllable, aa^-aipciv, etc., above. 
 
 [A large number of examples under each of the above heads 
 may be found in Leo Meyer's Vergleichende Grammatik, vol. ii. 
 pp. 1-78.] 
 
CHAPTEK VI. 
 
 Noun Inflection. 
 
 To the stem of an Indo-European noun are added (i) the 
 inflections of case ; (2) in the plural, the sign of number. 
 (The dual is a variety of the plural, which in Latin and 
 in most modern languages has fallen out of use altogether; 
 and where retained, as in Greek and Sanskrit, has a tendency 
 to disappear as a useless exuberance of expression. In 
 Hellenistic and Modern Greek it does not exist.) 
 
 The cases were originally eight : viz. Nominative, Accusative, Number of 
 Locative, Dative, Ablative, Genitive, Instrumental ; and outside 
 of these, the Vocative, which is no case properly so called, but 
 the uninflected noun-stem used as an interjection'- Sanskrit 
 alone, however, retains the full number of independent case 
 forms, and that only in the singular number : for in the plural 
 the vocative disappears (the nominative being used, as in Greek 
 or Latin), the dative and ablative unite, and the instrumental 
 has only one form (as against two in singular) ; while the 
 dual has only three distinct forms, one for nom. and ace, one 
 for instr. dat. and abl., and one for gen. and loc. In the 
 singular too gen. and abl., loc. and dat., are nearly related in 
 form. 
 
 In the kindred languages, the loss of distinct case-forms — or. Merging of 
 to speak more correctly, the merging of two or more originally Se-forms. 
 distinct case-forms into one — must have begun early in their 
 linguistic growth. The oldest accessible remains of the Greek 
 
 ' See however below, p. 117. 
 
io6 
 
 Notm Inflection. 
 
 [chap. 
 
 language show us the ablative merged in the genitive ; though 
 Latin, on the other hand, has retained the distinction of form. 
 The dative and locative, again, have become one in Greek, and 
 to a certain extent in Latin : while the instrumental has 
 vanished from both. In both languages, however, we shall find 
 remnants of both locative and instrumental forms, and Greek 
 has at least one conspicuous remnant of its lost ablative case 
 in the common adverbial termination -as. The confusion in 
 practice of the clear grammatical distinctions 'between different 
 cases naturally led to intermixture and confusion of forms ; 
 so that no formula will represent all the correspondences be- 
 tween the case terminations of the three languages in question; 
 but a general idea may be given thus : — 
 
 Sanskrit. 
 
 Greek 
 
 
 
 Latin 
 
 Nom. . 
 
 Nom. 
 
 . Nom 
 
 Ace. 
 
 Ace. 
 
 
 
 . Ace. 
 
 Dat. . 
 
 Dat. 
 
 
 
 . Dat. 
 
 Instr. . 
 
 „ 
 
 
 
 
 Loo. 
 
 » 
 
 
 
 . ( Abl. 
 
 Abl. . 
 Gen. . 
 
 JGen. 
 
 
 
 . Gen. 
 
 Voc. 
 
 Voo. 
 
 
 
 . Voc. 
 
 Gender. 
 
 Gender. The distinctions of gender, originating doubtless in the desire 
 
 to give different names for creatures in which there is con- 
 spicuous difference of sex, has been in most Indo-European 
 languages artificially extended far beyond the limits of natural 
 sex. ' The world of untraceably sexual or of unsexual objects 
 is not relegated to the indifferent "neuter;" great classes of 
 names are masculine or feminine partly by poetical analogy, 
 by an imaginary estimate of their distinctive qualities as like 
 those of one or the other sex in the higher animals, especially 
 man ; partly by grammatical analogy, by resemblance in forma- 
 tion to words of gender already established. At any rate, in 
 the common Indo-European period all 6r nearly all attributive 
 words were inflected in three somewhat varying modes, to 
 indicate generic distinctions ; and the names of things followed 
 
VI. J Noun Inflection. 107 
 
 one or other of these modes, and were masculine, or feminine, 
 or neuter^.' Yet, widespread as is their employment of generic Gender not 
 distmction, the Indo-European languages have no special pho- directly by 
 netic element for its expression ; but, as occasion arose, various elements. 
 Becondary means were employed. This seems to show that the 
 universal distinction of gender which we find in Greek and 
 Latin is neither original nor necessary, but a subsequent de- 
 velopment of language. 
 
 Modes of generic distinction : — 
 
 1. In Consonant-Stems and stems in -i-, -u-, or a diphthong 
 {narrjp, lirjTtjp, facilis, manus, vavs), the only distinction of gender 
 is by external means, i.e. by the gender of some other word in 
 grammatical agreement (0 ttot^p, t) priTrip, saeva mamus, etc.). 
 With d stems (including a- 0- stems) the raising of the vowel to 
 d (Greek ?;, Latin originally -a of fem. sing.) generally denotes 
 feminine gender. Occasionally however a is masc. {noKirrjs, 
 advend, the original quantity), and a {S, U) is fem. (080s, malus, 
 humus, etc.) : so that this means of generic distinction is not 
 of invariably certain application. 
 
 2. Certain case-suffixes are appropriated to a particular 
 gender ; or a case is not employed in a particular gender, but 
 its place is supplied by some other form. Thus in nom. sing. 
 the neuter has no final -s, either the accusative [novum, Shvov) 
 or the mere uninflected stem (dXijfles, dpi/, facile, facili) being 
 used. 
 
 3. Originally identical forms are distinguished, and the dis- 
 tinction adopted as a mark of gender, in-TrtiTT/s, dpfTrj : so with 
 the breaking up of the a sound into a 0, novo-d, nova-d 
 (originally navat). 
 
 4. Certain stem forms are appropriated to certain genders, 
 especially feminines, in -ja(-ya), -is, -ic, etc.; (pepovaaz^^epovrja, 
 d6Teipa=B6Tcpja, av\r]rp\s, vietrix^viet{o)r-ic-s, etc. 
 
 fThe distinction of gender is retained in the Teutonic Ian- Gender in 
 
 ■- " 1 1 T> 1 • • modem lan- 
 
 guages, e.g. modern German, and the Romance derivatives guages. 
 
 from Latin. English has abandoned the artificial part of the 
 
 • Ferrai's ' Comparative Grammar,' p. 200. See also Sayce, ' Principles 
 of Comp. Philology,' oh. yli. pp. 249-257, ist ed. 
 
io8 Notm Inflection. [chap. 
 
 Gender in system, retaining a difference vnform, only where sex is really 
 guages. an important distinction (e.g. m.wn, woman; buU, cow; and 
 the suflGx -ess in Princess, lioness, etc.), but it retains its 
 fundamental distinction in the pronouns he, she, it, or who and 
 what. Other languages (e.g. modern Persian) have lost even 
 that generic distinction : and in some languages of the Turanian 
 class (e.g. Turkish and Finnish) grammatical gender is said 
 never to have existed at all. There is of course in the necessity 
 of things no reason for choosing one particular accident of a 
 conception rather than another as a subject for grammatical 
 distinctions ; but, as a matter of fact, there is always a strong 
 natural personifying tendency at work in men's minds, leading 
 them to invest even inanimate things with the idea of sex. 
 Thus a ship to a sailor, a railway train to a porter, is always 
 ' she ;' and uneducated people often use the pronoun ' he' where 
 ordinary usage prescribes 'it.' These are examples of the 
 natural tendency to extend distinctions of gender taking effect 
 in a language which has generally repudiated such extension 
 to all objects as unnecessary: and it is to the unrestrained 
 working of such natural tendencies that we may ascribe the 
 great development of generic distinction at an early period in 
 the Indo-European languages, before, in fact, they had as yet 
 branched off from the primitive stock.] 
 
 JDeclension. 
 Principles of Nouns are divided into two main classes or 'declensions' 
 
 division into 
 
 Vowel and according to the final letter of the stem : viz. : — 
 
 Consonant ° . i.- i 
 
 Declension. I. Vowel- Declension (or A declension), includir^g stems which 
 end in -a, -e, -o (the three varieties of a the original vowel) ; 
 and thus comprising the ist (musa-), 2nd {domino-), and 5th 
 [facie-) 'declensions' of Latin Grammar; and the ist (TroXiTa- 
 liova-a-) and 2nd (javpo-) of Greek Grammar. 
 
 II. Consonant-Declension, including stems which end in a 
 consonant, or the semi-vowels -i, -to, or diphthongs av, ev, ov : 
 thus comprising the 3rd and 4th 'declensions' in Latin (judic-is, 
 navi-s, gradu-s), and the corresponding nouns in Greek [ipvKaic- 
 or, woKi-s, fiorpv-s, pa<ri\(ii-s, vaii-s, /3o0-i). A small number of 
 
VI. J Noun Inflection. 109 
 
 nouns with stems in o- or a>- follow the inflections of this 
 declension, e.g. irei6a>, n-etfld-os: rjpui-s, ^pa-os. The stem of 
 words in this declension is best recognised in Greek in gen. 
 sing., where all that remains after deducting the termination 
 -or IS the stem, e.g. Xemj/, \eovT-os; Svofia, ovofmr-os. This is 
 sometimes (but by no means always) the case in Latin, e. g. 
 comes, comit-is ; judex, judic-is. The final consonant will of 
 course generally be shown in this way, but the weakness of 
 Latin vowel sounds (p. 56) often obscures the true vowel 
 of the stem ; thus in auspex, auspic-is, the nom. auspec-s gives 
 the true form (spec-). Often neither retains it, e.g. remex, 
 remig-is, the true form being ag- ; auceps, aif,cupis (true form 
 cap-). Such varieties however fall under the head of Latin 
 Souud-Lor«. 
 
 There are certain differences between the inflections of the Differences 
 two classes thus arranged, which make it more convenient between the 
 to classify i, u, and diphthongal stems under the consonantal sions. 
 than under the vowel declension. Thus, in Greek : — 
 
 (ffl) In gen. sing, consonant declension has always -or (-as). 
 
 (6) In nom. plur. „ „ „ -es. 
 
 In Latin : — 
 
 (a) Gen. sing, and nom. plur. end in a long vowel or diphthong 
 in the vowel declension ; in -s in the consonant declension. 
 
 (6) Gen. plur. of vowel declension -rum ; consonant declen- 
 sion -um. 
 
 (c) Dat., abl. plur. of vowel declension -is; consonant declen- 
 sion -bus. 
 
 [In older Latin however some of these differences apparently 
 did not exist : for we find in nouns of the vowel declension -aes, 
 -as, -es as gen. sing, of a stems (see below, on Gen. Sing.), and 
 -um as term, of gen. plur. in both a and o stems ; while certain 
 words show -6ms in dat. and abl. plur. Archaic Latin thus 
 fiirnishes materials for approaching nearer to a uniform system 
 of inflection for all stems than do the earliest traceable stages 
 of the Greek language '- In Sanskrit there is but one general 
 scheme of terminations, the classes of declension (eight in 
 
 ' See Boby's ' Latin Grammar,' vol. i. Book II. ch. xii. 
 
no Noun Inflection, [chap. 
 
 number) signifying th« different modes of combining the final 
 letter of the stem or base with the termination : a system 
 which might, no doubt, have been carried out by Latin and 
 Greek grammarians, had there been an equally careful gram- 
 matical analysis at an equally early stage in the history of those 
 languages, and had the formation of nouns and verbs from roots 
 and ' crude bases ' or stems been traceable with the same clear- 
 ness as in Sanskrit.] 
 
 Nominative Singular : — 
 Nom. Sing., Formed in all nouns by suffixing -s to the stem. This -s is 
 Latin. generally regarded as representing a pronominal root -sa (de- 
 
 monstrative pronoun); sa=Greek o (cp. p. 66); sd (fem.)=^. 
 This demonstrative root or stem with -s of nom. sing, forms 
 sa-s, i.e. Greek os, which in Homer is demonstrative. In 
 Greek and Latin the -s of nom. sing, is retained in many 
 words, which therefore need no further explanation (e. g. Aeneas, 
 KpiTTjs ; dominus, 6e6s ; urbs, n-oXtr ; grains, fades, ^aa-iXeis). 
 From others it has disappeared, but its presence can generally 
 
 In vowel be traced ; e.g. masc. vowel stems in -a have lost it, but such 
 stems. ° . ' 
 
 double forms of masculine words as 'nrn&rrjs and hrnora (Hom.), 
 
 alxiir]Tris and alxiaiTo., are sufficient evidence for its having once 
 
 existed. (Compare also the archaic Latin forms paricidas, 
 
 hosticapas, and poeta, Apella, beside jtoitittis, 'ATreXXijt.) Bopp 
 
 (§ 136), Schleicher (§ 246), and others, assume its loss from 
 
 feminine stems in a-, as bona, dya6a, sivfl. : but there is no 
 
 satisfactory evidence that such stems ever took the -s of nom. 
 
 sing. Benfey (' Orient and Occident,' i. p. 298) maintains that 
 
 they did not. 
 
 In Greek Consonant Stems (Greek). 
 
 Cons, stems. . . , , 
 
 hruttwral and Labial stems : -9 with the stem vowel becomes 
 
 I or ■^. ipiXa^ (stem <t>vKaK-), 0Xo| (0Xoy-), Si/r (ojT-). 
 
 Dental Stems : t and 8 never remain before s, but disappear, 
 
 the preceding vowel being often lengthened in compenstetion, 
 
 e.g. "KafiTTCK (XofwraS-i), x^P^^ (x"/"''"0' '''^'""P^^ (renurfrfr-s). In 
 
 Sd/iap (SdnapT-) both t and s disappear. Stems in -vt sometimes 
 
 lose both consonants before s {Tir\jras=TvylfavT-s, Soiis^fiovT-i), 
 
 sometimes lose t and s retaining v {<f)epav=(j)epoiiT-s), 
 
VI.] Noun Inflection. iii 
 
 In -V stems sometimes the v, sometimes the -s is lost ; the Nom. Sing, 
 vowel being in each case lengthened, e.g. rakas (stem 
 ToKav-), <i>pr)v {(ppev-), ^6av (j(6ov-) : and sometimes both forms 
 are found, e.g. 6ls, 6iv ; StXc^ls, SeXi^/c (that in -s being the 
 older). 
 
 After -p stems, s is lost, jran^p (n-arep-s) : but AeoUc keeps 
 both consonants, — x^ps^ (^X^'p); p^i^o-ps- In paprvs {p,dpTvp-os) 
 the p disappears. The solitary -X stem (SKs) retains both \ 
 and s. 
 
 In -s stems the second -s denoting nom. sing, is lost and the 
 
 vowel lengthened, e.g. dXr;^r, stem dXij^tr. 
 
 Consonant Stems (Latin') : — In l**i" 
 
 ^ ' _ Cons, stems. 
 
 Guttural and Labial stems : s is added to the stem, e.g. vox 
 {voc-), lex (leg-), aueeps, urbs. 
 
 Dental stems : t and d disappear before -s and the preceding 
 vowel was originally lengthened in compensation ; but in 
 Classical Latin the tendency to shorten final syllables has again 
 shortened the vowel, except in monosyllables and after i- pre- 
 ceding. Tl\iMS pes (^ped-is), aries (ariet-is): hut miles (milit-is), 
 eques {equit-is), etc. Stems in -nt only reject t {amans, amant-is), 
 Latin being more tolerant than Greek of combinations of final 
 consonants ; but in old Latin and in the common dialect we 
 find infas, sapies, etc. (cp. the parallel forms quoties, quatiens). 
 
 In -s stems -s of nom. sing, is lost, and the preceding vowel 
 originally lengthened, but in Classical Latin generally short. 
 "We find however Ceres (Ceres-is), arbos (arbOr-is). In the 
 declension of such stems the final -s became r in oblique cases 
 (except vas), and this r often supplanted final -s of nom., e. g. 
 arbos, arbor ; honos, honor ; vomis, vomer ; robur cp. with 
 vetVjS (the -s being in all cases the older form). Stems in -n 
 if masc. or fem. lose -ns, as homo {homon-s) ; but in some words 
 n is retained {jpecten,flamen), and in sanguis (originally sanguis, 
 Lucr. iv. io^6)=^s(unguin-s, -s is retained and n lost. After -r 
 and -l stems -s is always lost, but the preceding vowel was 
 originally lengthened as in Greek : sal {sal-is), par (par-is), 
 actdr (actor-is). 
 
 In -i and -u stems s is generally kept (igni-s, gradu-s) : but 
 
iia 
 
 Noun Inflection. 
 
 [chap. 
 
 Sanskrit 
 Nom. sing, 
 
 where r or Z after another consonant precede i, the full termina- 
 tion -is is lost, and e inserted before r ; e. g. aceT-=acri-s, which 
 remains as fern. nom. ; vigil=vigili8. 
 
 [In Sanskrit nom. sing, -s is omitted after consonantal stems, 
 the vowel being sometimes lengthened in compensation, some- 
 times not. Thus va.k is nom. sing, from stem vak-, Latin vox=: 
 voe-s, and durman&s {Svaiievfis) nom. sing, of stem durmanas 
 {&v(Tiieves) ; but bhiran {(jjepav) is nom. sing, of stem bharant- 
 [(jifpovT-). Stems in ar (masc.) and &t (fem.) reject both r and 
 s, thus pita, (stem pitar)^7raT^p, d&tS, (stem dktkr)=^8oTrip. It 
 will be observed that in all these words the Greek forms of 
 nom. sing, are fuller than the corresponding Sanskrit.] 
 
 Nominative Plural : — 
 Nom. Hot.: Originally a reduplication of the sign for nom. sing., -sasa ; 
 then -SOS (which is actually found in Vedic Sanskrit as nom. 
 plur. termination in a- stems, e. g. i^va-sas from fisva-s) ; and 
 finally -as (Greek -fs of consonant declension), which is the 
 form in most Indo-European languages, and survives in one of 
 the few remaining English inflections, the -s of plural significa- 
 tion. 
 
 In Greek -es [=i-as) is added to consonant stems, as noifiiv-es, 
 lx6v-es, iidvTi-es. Sometimes the vowel of -i and -u stems is 
 raised ; e. g. irSKets, jrdXTjes, 7roXe€s:=7r<5X€y-er from iroXfi-, the 
 altered form of stem iroKi- (whence also the Ionic noKi-es without 
 raising the vowel) ; and jroXEis=woXcf-es from jroXev, the altered 
 form of stem woKv- : cp. raxhs, iyx^'Kfes. In these the v of stem 
 has been raised to fv, and the v of this diphthong then changed 
 to p, which of course disappears altogether (above, pp. 43, 68). 
 
 The nom. plur. of vowel stems -m, -ai, shows no trace of 
 final -s, though on the analogy of Latin (see below) we should 
 infer that it once existed. It has been suggested (Schleicher, 
 Comp. § 247) that the loss of -s began with nom. plur. of 
 pronominal stem ta- {to-) : i. e. to\, rai ; this stem ta- according 
 to the theory being increased by the sufiBx ya (ja) a common 
 derivative suffix (see above, p. 102), would form in nom. plur. 
 tay-as, which by loss of final syllable would become tai (rol or 
 rai) : and that this termination -ot -ai was gradually applied by 
 
 In Greek : 
 
Ti.J Noun Inflection. 113 
 
 analogy to all a- and 0- stems. This is ingenious ; but it rests Nom. Pimv 
 upon an assumption for which there is no evidence one way or 
 the other ; and in philological enquiries it is better to confine 
 ourselves to the facts of language, and to be content with 
 unsolved problems rather than risk hypotheses. 
 
 Latin Nom. Plural : — In Latin. 
 
 Consonantal stems; always in -es, the quantity of which is 
 supposed to be due to analogy from the i- stems. It is prob- 
 able, however, that the original termination was -es (corres- 
 ponding to Skt. as, Gk. es), which e. g. in quattuor has dropped 
 off altogether (cp. Ten-ap-es) : so in Umbrian/rai!er=/ra<(«)r-es, 
 Oscan censtur-=c6ns{t)or-es. Nom. plur. of i- stems always in 
 es : here es was probably added to stem, thus giving -ies, which 
 became es. Is or eis (all found on inscriptions and in MSS.). 
 [Another explanation is that the stem i- yas raised to ey (as 
 TToKeis^TTokey-es above) so that o«es=oV(Si?s=oi!«y-es.] 
 
 u- stems in -Us=u-es (cp. vfKv-es). 
 
 a- stems ; nom. plur. ae or in archaic Latin -ai ; but it , is 
 inferred from a comparison of the other Italian dialects that 
 the original form was -as ( = a-es) e. g. Umbrian urtas, totas 
 (=Latin ortae, totae ; Oscan aasas, serif tas (=-arae, scriptae), 
 Matrona (nom. plur.) found on an inscription is supposed to 
 point to this older form in -as with -s dropped ; but it might 
 equally be an error for matronal, and in inscriptions a wide 
 margin must always be allowed for merely casual errors of the 
 cutter \ If -as is the original form, how do we get ai, ae ? 
 The most plausible theory is that the i here represents an in- 
 crease of the stem by i, such as will hereafter be shown in the 
 pronominal declension {ha-i-c, Jiaec, etc. see below, chap. vii). 
 Thus equae=equai—equa-i-s (s being dropped as often in 
 Latin). Or it may be supposed that the -i- was added, upon 
 analogy of the pronominal declension, after the loss of final s. 
 
 In the 0- declension we get indications that o-es {-es added 
 to the stem o-) was the earliest form. The various forms 
 
 • Kitschl wishes to restore the form in -as in Plant. Trin. II. iv. 138 (to 
 avoid hiatus) : — 
 
 Nam fulguritae sunt alternas arbores. 
 
 I 
 
114 Noun Inflection. [chap. 
 
 JJom. Plur., actually found, which lead to this inference, may be thus 
 arranged ^ : — 
 
 a. Oldest forms : r. Fesc&ninoe, pilumnoe, poploe, (Carm. 
 Sal.) ; stem retained in full, and therefore probably the oldest, 
 -s only having dropped. 
 
 2. ploirurrie (Epit. L. Scipio, see Appendix I. i. 2) a contrac- 
 tion from oe, but connected with the later forms in ei, i. 
 
 h. Forms retaining -s {-es, -eis, -is) e. g. modies, ques (S. C. de 
 Bacch. see Appendix I. ii.), ds, libereis, magisbns, hisce (in Ter. 
 Eun. 269) These forms do not appear in inscriptions earlier 
 than 1 90 B. c, and remain for about a century. To explain 
 the presence in these later rorms of the final -s, which the 
 earlier forms had lost, Corssen supposes a transition (by 
 analogy) to the forms of the consonant (i-) declension : but it 
 seems at least as natural to suppose that in the early inscrip- 
 tions we see the result of a tendency to drop final consonants, 
 which was artificially corrected during the second century b. c. 
 (when we know that the literati of Rome took great pains to 
 establish a correct standard for their language), but finally 
 prevailed; pronunciation, as usual, obtaining the victory over 
 etymological considerations in fixing orthography. 
 
 c. The classical form in -i. 
 
 We therefore may trace the stages of change in these forms 
 thus : — 
 
 a- stems ; a-es, ds, a-i{s), ae. 
 
 0- stems ; o-es, e{s), e, i. 
 
 is. 
 
 [In Sanskrit, all masc. and fem. stems form nom. plur. in -as 
 before which i and u are raised ; vaA-as {voces), bMrant-as 
 {(jjepovT-es), sivas (siva-1-as), ^vay-as (from avi-s).] 
 
 Nominative Dual (Greek) : — 
 
 Nom. Dual, Schleicher assumes for this an original -sas, a lengthened 
 Greek and „ , 
 
 Sanskrit, form of nom. plur. (as I nom. dual neuter, 01 i nom. plur. 
 
 neuter j and hhydm dat. abl. instr. dual). This -sas would 
 ' See Wordsworth's ' Fragments,' etc. ; Introd. ix. 9. 
 
VI. J Noun Inflection. 115 
 
 next become -as ; but in all Indo-European languages it has 
 been further weakened : in Sanskrit to ka. (in feminine a- stems 
 to e) ; in Greek to e, which appears in the consonant declension, 
 but in the vowel declension coalesces with the stem vowel, wriro) 
 
 = «r7ro-€, x^pa^)(aipa-e. 
 
 In Latin duo (Sanskrit dvau) and ami>o (Sanskrit ubh&u, 
 
 Greek aixcpa) are the only dual forms. 
 
 Aecv^ative Singular : — Accus. Sing.: 
 
 In Greek ■ 
 General type ; -am for consonant, -m for vowel-stems. 
 
 In Greek, -m becomes -v by the euphonic laws of the lan- 
 guage : and with consonantal stems -av appears only as -a added 
 to the stem, KafiirdS-a, Tjpa-a. Vowel stems retain -v (tirno-v, 
 (pvyfj-v). Stems in i-, v- and diphthongs av-, ov- generally form 
 the accus. sing, on analogy of vowel-stems in -v ; irSKi-v, ^orpv-v, 
 fiov-v, vav-v. Stems in eu- however are generally treated as con- 
 sonant stems (y becoming f), thus ^a<Ti\e-a=0cuTiK€p-a (^ao-iXcv-) : 
 and the same is not unfrequently the case with other diphthongal 
 and I- and v- stems. Thus we have the Homeric vij-a {vqFa) 
 beside vav-v; evpia-=.(vpef-a {y- of stem raised to cu-, eF-) beside 
 eipii-v, and 6<j>pi-a : and in the other case-endings of jSoCs', j3oC 
 is treated as a consonant stem (|3of), /3o-6s=|3of-6r (Latin 
 bov-is). So too TriSKrj-a^iToXey-a (iroXi-s) beside nSKi-v. 
 
 The neuter accus. in consonantal stems is merely the stem 
 subject to euphonic laws of the Greek language: e.g. repas 
 (repaT-), p^Xi (/ifXiT-), (jyepov {(pepovr-), yXvKV- : in vowel stems it 
 ends in -v. 
 
 In Latin, -m is the invariable ending with masc. and fern, in Latin, 
 stems. The -em of consonantal declension is said not to re- 
 present I. E. -am, but i-m : i. e. the stem lengthened by -i, which 
 then became -e before m, in both stems thus lengthened and 
 original i- stems, with a few exceptions among the lattery- 
 It is no doubt desirable to regard these few accusative forms 
 in -im among the mass of forms in -em as survivals of a more 
 
 ' The following nouns fonn accus. in -im, and ablat. in -i : — 
 Always — buris, tussis, sitis, vis, Tiberis, etc. 
 Generally — febris, pelvis, pulvis, restis, seouris, turris. 
 Occasionally — clavis, navis, sementis, 
 
 I 2 
 
ll6 Noun Inflection. [chap. 
 
 Acous. Sing., primitive form ; and this is in harmony with the usual course 
 of vowel degeneration in Latin (above, p. 57) in which e is 
 the lowest point. It might, however, be maintained that -em 
 as seen in pedit-em, eqmt-em at first represented -Am {e being 
 a regular variety of original a) : and that this -em coalesced 
 with the final vowel of i- stems into -Im or -em {i-em), -im 
 being the earlier form; and that finally the analogy of this 
 -im or -em caused the -em of purely consonantal stems to be 
 regarded as a long syllable, upon the erroneous inference that 
 -em was exactly the same in all words which exhibited it. 
 This view is not less consistent with the observed facts of 
 languages, and obviates the difiiculty which cannot but be 
 felt in the theory of a different structure for one of three words 
 so obviously parallel as Sanskrit vdA-am, Old Bactrian v^k'-em, 
 Latin voc-em. This Old Bactrian accus. in -em of consonantal 
 stems seems to furnish a clear link between Sanskrit -am and 
 Latin -em, e.g. barent-em, cp. with Sanskrit toh^rant-am, Latin 
 ferent-em. 
 
 To the vowel stems in -a (-0) -m was added ; honum {bono-m), 
 musa-m. -m as we have seen (chap. iv. p. 73) was weakly 
 sounded in pronunciation, and is accordingly omitted on some 
 old inscriptions. 
 
 Aceusative Phi/ral: — 
 
 Accus. Plur.: General type, -ns, i.e. addition of s to termination of accus. 
 sing, m, which by assimilation to the dental sibilant s becomes n. 
 This -ns is retained only by Gothic, the euphonic laws of which 
 did not forbid such a combination at the end of a word, e.g. 
 gastins (stem gasti-), sununs {sunu-) : but there are teaces of it 
 in both Greek and Latin, and also in Sanskrit and Zend. 
 
 In Greek; Greek accus. plur. : formed by addition of s to ace. sing., but 
 -vs only retained in the Axgive and Cretan dialects, e.g. tovs, 
 :=Tovs, irpeiyevTavs=:irpea-^evTds. Elsewhere, in the vowel declen- 
 sion, V disappears, the vowel being usually raised in compensa- 
 tion, e.g. iinro-vs, iirnovs (Doric tTrn-fflr, like Latin -os) ; x^P'"'^, 
 xitpas. In Lesbian -ovs and -avt became -ois, -ats : thus KoKais, 
 = KaXa9, as in Pindar we have ^iK^(rcus^=<j)iK'j(ravs^^iKrja-avT-s, 
 Attic (j)iXrio-ag. In consonant stems -s follows -a of ace. sing. 
 
"VI.] Noun Infiection. 117 
 
 making -as : but in i- and u- stems there is .variety of form ; Acciis. Piur. 
 thus beside iroKi-as and 7rd\i/as we have 7roXrs=7rd\u/-r. TToXctf, 
 the ordinary accusative, is perhaps best taken as =TToKfy-as (see 
 above on nom. plur., p. 105) : but it might also represent 7rdXo/-s, 
 and be ■='koKTs. With neuters, a is added to the stem. 
 
 Latin accus. phi/r. of masc. and fern, stems always in -s. In Latin, 
 with long vowel preceding by compensation for the loss of 
 -m- ; thus -ds=-am-s, -ds=.-om-s; es {ls)=ems (ima), -us = 
 -wm-s. To neuter stems -a is added, corpora=corpos-a. 
 
 [In Sanskrit, traces of the termination -ns are found : but in 
 vowel-stems usually either n or s disappears and the vowel is 
 raised, e.g. dsva-s (equus), ace. plur. dsvan; dsva {equa), ace. 
 plur. dsvas. So Avi-n (masc), Avis (fern.) from stem avi-. 
 To masc. and fem. consonant stems, and monosyllabic vowel 
 stems, -as is added, vaJ-as, Asman-as, nasr-as (nau).] 
 
 The Accusative Bual in Greek (as also- in Sanskrit masc. Accus. Dual, 
 and fem.) is the same as nom. dual. In Latin d,uo, arnbo have 
 also a form duos, a/mhos, on analogy of plural, and in fem. only 
 this form {duas, ambas). 
 
 Vocative Si7igular : — 
 
 This, it has been already said (p. 105), seems to be in Indo- Voc. Sin^. 
 European languages no 'case,' but the mere stem used as an 
 interjection. 
 
 It has however been suggested, with some probability, that 
 the vocative is originally the nominative with the accent drawn 
 back so that the final syllable became shortened in pronuncia- 
 tion. The evidence for this is (i) the fact that in Sanskrit the 
 vocative is always accented on the ^rs4 syllable, when accented 
 at all, i.e. at the beginning of a sentence; (2) that in Greek a 
 certain number of words accented oxytone in the nominative 
 throw the accent back as far as it will go in the vocative (e. g. 
 aSeXcjibs, novrjpos, Trarfip, but SdeK(j)e, rrovripe, Trarep). It is also 
 noted that in Greek, if an oxytone noun becomes a proper name, 
 in a majority of cases the accent goes back (e.g. d/io/)yor,*A/iopyof): 
 and there is a passage in Aulus Gellius on the pronunciation of 
 Valeri as gen. or voc. sing, of Valerius. As vocative, he says, 
 it was accented Vdleri ; as genitive, Valeri. 
 
Hi 
 
 Noun Inflection. 
 
 [chap. 
 
 In Latin. 
 
 In Greek. In Greek guttural and labial stems, the nom. is used as 
 voc. (except ■yvj'ai=7ui'auc-): but in dental stems the mere stem 
 is used, subject to euphonic laws, e. g. nai (n-aiS-), ova (arax-), 
 yipov (yepovT-). irovs however (stem iro8-) and the nom. sing, 
 of participles in -as, -fts, -ovs, -a>v (stems in -vt) are used as 
 vocative. The voc. in -e of o- stems is the stem with o sunk 
 to e : 6eos however (as Beus) is generally used for voc. (though 
 in Matt, xxvii. 46 we have Bee fwv) ; so <j>lXos (Horn. Od. iii. 375) 
 and oiros. The voc. termination -01 of n-ciflii, alSa-s, etc., is 
 anomalous ; it appears however to stand to nom. sing, in -<a as 
 Sanskrit voc. of fern, a stems (e. g. asve) stands to the nomina- 
 tive : for e=-ai. 
 
 In Latin the nom. sing, is used for vocative, except in 
 masculine stems in 0-, where the stem with -0, changed to e 
 is used. In puer (for puer-us) the abbreviated from of nom. 
 is also voc. : but puere is found in Plautus. 
 
 Ths Vocative Dual and Plural in Sanskrit and Greek, and 
 the Vocative Plural in Latin are the same as the respective 
 nominatives. 
 
 Genitive Singular : — 
 
 There appear to have been two forms of Indo-European 
 pean Forms. sufRxes for the genitive case, viz. for a- stems (a-, 0-), sya; 
 for all others, -as or -s. These suffixes are probably pro- 
 nominal in their origin, and sya {sja) is perhaps compounded of 
 the two roots sa, ya (Ja) : but of this we can have no evidence'. 
 
 Greek Gen. Sing. : — 
 
 In consonantal stems -os [^-as) is added to the stem : mB-os, 
 
 yevovs (^=:yeveos=:ycve(T-os). -os is sometimes raised to -as (noKe-as, 
 
 ^aa-iKe-tas). In diphthongal stems v has generally passed into 
 F (consonantal) and thus disappeared, fioF-is (0oii-), l3aa-i\eF-a>s 
 
 Gten. Sing. 
 Indo-Euro 
 
 In Greek. 
 
 ' The old view that -sya of gen. sing, appears aJso as an adjectival 
 suffix in ST]fx6-(TtO'S, bo that the Homeric genitive b^iJioio=57jfi6t7iO'f the stem 
 of the adjective, though plausibly supported by the identity in Sanskrit 
 and other languages of genitive termination with adjectival suffixes (cp. 
 Max Miiller, Lectures I. iii), cannot, I think, hold against the question, 
 Why then do we never find Srjfioios instead of Srin6ffios ? aio in Si)n6aio-s 
 is the adjectival suffix no, the t being changed before i to s according to 
 the universal tendency of pronunciation botii in Greek and Latin. See 
 above, p. 78. 
 
VI.] Noun Inflection. 119 
 
 (^ao-tXet)-). In V- stems forms like ymvos lyom-), Sovpos (Sopv-) Gen.'Sins., 
 
 Grfifik 
 are trauspositions from yovv-bs, Sopv-6s : while yXuKc-or, ao-i-for, 
 
 etc., show that the stem vowel v has been raised to ev (eF) 
 
 and become diphthongal ; thus y\vKeos=:y\vKeF-os is analogous 
 
 to ^a(n\(-as=fia(ri\(F-os. Similarly irdXi-as and Homeric ttoXtz-os 
 
 =7r6\ey-os ; the stem vowel remaining unaltered in Ionic 
 
 TToXt-OS. 
 
 Fem. a- stems have -as or -s added to the stem vowel, a-ocplas, 
 (pvyrjs. Masc. and neuter stems in 0- originally formed gen. by 
 addition of -a-yo, whence the Epic gen. in -mo; a.ypow=^aypo-tTyo 
 by omission of <r ^. The Attic gen. in -m (Aeolic -m) arises by 
 contraction from -oo = -oo-to, with first o- and then i omitted. 
 
 From masc. stems in a- we find three forms in Homer, (i) -ao ; 
 (2) -f(B {'ArpeiSea), in which the quantity of the two syllables -do 
 is transposed, and u. weakened to c (-em then becomes one 
 syllable by synizesis and the accent remains unaltered in spite 
 of 0) in final syllable, as with TroXeo)?, etc.) ; (3) -<b by contrac- 
 tion from -ao, 'Epfieiu), ^opia (^'Epneta-s, (3opea-s). Aeolic has 
 
 -a (0 being lost), AlSd, KpoviSd. The earliest form in -ao is 
 probably =a-o-yo, ayo ; but Curtius, in his work on Greek 
 Etymology, derives -ao from aos=: a,yas, the gen. sing, termina- 
 tion of Sanski'it fem. stems in -a, which change the stem vowel 
 -a into ai {ay) before -as, e. g. Asvk {equa), gen. &svkyas ; but 
 a«va-s {equus), gen. 6sva-sya. The Attic gen. in -ov is a con- 
 traction from -ao, TroXiTOi/^TToXiVa-o. 
 Latin Genitive Singular : — 
 
 The suffix -as appears in Latin as -os, -us, -es, -is. in Latin: 
 
 (i) -OS, senatu-os in S. C. de Bacch. (Appendix I. ii.). stems. 
 
 (2) -us, on inscriptions up to the end of the seventh century, 
 A.U.C. (rarely after 100 B.C.), Cererws, Caesarus, hominus, 
 Venerus, etc. From this in u- stems arose the contraction us 
 ^uus (domuus inscr.) It also survives in alius, illius, etc. 
 
 (3) -es, on inscriptions before the Second Punic W&r, Salutes, 
 Apolones : and again in late Latin, Caesares, campestres, etc. 
 
 ' In the Thessalian dialect, the gen. sing, of o stems often ended in -oi, 
 about which there are two views: (i) that it = -oio, minus the final o 
 (Ahren.s, 'De Dialectis Aeolicis,' p. 221; ' De Dialecto Dorica,' p. 528 
 sqq.) ; (2) that it is am old locative used in a genitive signification. 
 
130 
 
 Noun Inflection. 
 
 [chap. 
 
 ft- stems. 
 
 Gen. Sing., (4) -is^ the ordinary gen. sing, of consonant stems proper. 
 The is of i- stems {ovis) was perhaps originally i8^=i-os : thus 
 ovis=:ovi-os, as we find a form alis-=alios {alius). The tendency 
 to shorten^final syllables, aided by the tendency to see analogy 
 between two similar terminations oms, nomints, would suf- 
 ficiently account for the subsequently uniform -is of both 
 consonant proper and i- stems, -is is obyiously parallel to 
 Greek -01, cp. genus, gener-is with yeVor, yivea-os {yeveos, 
 " yivovs). 
 
 The u- stems exhibit the greatest variety of gen. sing, inflec- 
 tion : thus we find (i) -uos as above; (2) -uus, as above (in 
 some cases however -uus may be due to a method of denoting 
 the length of a vowel by doubling it, introduced by the tragic 
 poet Accius, and prevalent on inscriptions from about 130-75' 
 B.C.; found also regularly in MSS. of Pliny the elder, where 
 -uus represents -us of gen. sing. nom. and ace. plur. of m- stems) ; 
 (3) -uis, retained in su-is, gru-is (which then, by analogy, are 
 declined like i- stems), and used by several writers up to temp. 
 Cicero, e.g. senatuis, domuis, etc., quoted by G-ellius ; quaestuis 
 (Ter. Hec. 'j^e,),fructuis, victuis, etc. (Varro) ; (4) -i, perhaps 
 an analogy of -0 stems (from similarity of nom. sing, -us), or 
 possibly from some confusion with the past part, in -tus, most 
 of the examples being from words where t precedes the stem 
 vowel U-, e.g. adventi, quaesti, ornati (all in Terence), senati, 
 furcti, etc. (see Eoby's Latin Grammar, vol. i. § 399) ; (5) the 
 ordinary termination -us, by contraction from -uus or -uis. 
 Voweistems. In 0- stems the gen. ends in -i or -ei (inscriptions from the 
 time of the Punic "Wars to Augustus). Three explanations of 
 this termination are suggested : — 
 
 (i) That it is a locative, which has supplanted the old 
 genitive. This would account for, and has probably been sug- 
 gested by, the apparently abnormal grammatical usages of 
 Romae, Tarenti in a locative sense ; these, however, may be 
 otherwise explained (see below, p. 1 1 9). 
 
 (2) That like Greek -ov, it arises from the termination -sya, 
 i. e. agri=agroi=agro{s)i{o). This gives at first sight a plausible ' 
 parallel between the Greek and Latin 0- declensions ; but the 
 
 Gen. in -i. 
 
VI.] Noun Inflection. lai 
 
 only real parallel to this supposed Latin abbreviation of -syo is Latin Gen. 
 the Thessalian gen. in -ot mentioned above (p. 1 1 2), and Bopp 
 truly points out that ' lupi and lupae from lupai rest on the 
 same principle ; and if lupi proceeds from Xvkow, whence can 
 lupai be derived, as the corresponding Grreek feminine nowhere 
 exhibits an -am or -t^io?' (Comp. Grammar, § 189). 
 
 (3) That the original termination was o-is (i. e. -as added 
 to the stem), the final -s being lost, and -oi contracted to -i. 
 This explanation is made more probable by traces of a final 
 -s in the other Italian dialects : e. g. Oscan suveis (sui), Pumpai- 
 aneis (Pompeiani) ; Umbrian puples {populi), katles {catuli), 
 etc., which lead us to infer an Italic genitive in -ois, whence 
 Oscan -eis, Umbrian -es, Latin -i. The analogy of fem. a- 
 stems will also bear out this conclusion. For them we have 
 in paterfamilias and the old genitives terrds (Naev.), vids (Enn.), 
 etc. distinct evidence of a termination -ds, the readiest solution 
 for which is that it is a contraction for -a-is [is^as added to 
 the stem), a termination found in one old inscription in Prose- 
 pnais^^Proserpinae, and on vulgar inscriptions (not before 
 seventh century A. u. c.) appearing as ' -aes or -ces (chiefly in 
 proper names of freed women and slaves Juliaes, Anniaes, 
 Vernaes, etc.). The other form in -di (Lucr. and Virg.) or 
 -ae may then be -traced to the same -ais by loss of final -s, 
 and corresponds exactly to agri^^agroi from agro-is. This 
 explanation, which reconciles the two forms -as and -ai (ae), 
 and harmonises the declension of both masc. and fem. a- stems 
 (0- and a-) appears upon the whole the simplest and most 
 satisfactory. 
 
 Of stems in -e four forms of gen. sing, are found, viz. -es Gen. Sing, of 
 (rabies, Lucr. iv. 1083), -ei, -e (fide, Hor. Od. iii. 7. 4; die,' 
 Virg. G. i. 208) and -i (e.g. dii, a variant for die, in Aen. i. 636; 
 see Eroby, Latin Grammar, i. § 357). Of these -e and -I are 
 contractions of -ei : -ei and -es are phonetic varieties of -ai 
 and -as of the a- stems, and the explanation above given covers 
 them. 
 
 Genitive Plural : — 
 
 An original Indo-European type -as-ams, i. e. -as (gen. sing.) indo-Buro- 
 
 pean type. 
 
122, 
 
 Noun, Inflection. 
 
 [chap. 
 
 Gen. Plur. + -am (pronominal element found in bhy-am etc. see p. 120) 
 + -s (plural sign), is imagined by ScHeicher and others : which 
 -asams, it is supposed, would gradually sink to -asam, -sam, 
 -am. Of these forms, -am = Greek -av, Latin -um ; -sam, = 
 Latin -rum of a- and 0- stems ; while the quantity of o 
 before -rum [equorum from stem equd), and of Sanskrit 
 tesham (horum,) from stem ta (hie) perhaps points to -asam, 
 whose initial vowel coalescing with the stem vowel would 
 make a long syllable. [In Sanskrit only the pronominal 
 declension retains this trace of a longer form -8a,ni or -as&m. : 
 with nouns, -km is added direct to consonant stems, e. g. 
 va^ -am {voc-um), while vowel stems are increased by n 
 before the addition of -ftm, e. g. fi,sva-n-ftm (d«va-s), dvl-n-a,m 
 (avi-).] 
 
 In Greek; Greek Genitive Plural, -av^dm, is added to the stem. The 
 o- of o- stems coalesces with it, XvKav = Xuko-mk : and the same 
 is apparently the case with a- stems, viz. x'^P''" = x'^P^-"'"- 
 The fern. gen. plur., however, is always accented with circum- 
 flex, the masc. only when the accent of nom. sing, is oxytone : 
 and it has been supposed that this difference points to an 
 original difference in formation, the d- stems having the suffix 
 -a-cov [-sam) ; so that x'"pS''=^X'"P°'''<"''' This is to some extent 
 borne out by the Homeric form a-av of such gen. plur., and 
 the comparison of e. g. ra-av gen. plur. fem. with Sanskrit 
 ta,-sftm (harum) from stem ta. a- would of course naturally fall 
 out between two vowels (p. 66). rd-mv, is-ta-rum and t^-s^m 
 would thus be parallel forms. 
 
 In Latin. Latin Genitive Plural. Formed by adding-wm or -om (found 
 
 in u- stems, and in 0- stems after ii or v) to consonantal o-, i-, 
 or u- stems : e. g. fuhnin-wm, avi-um, magistratuom,frii,ctuum ; 
 and -uum, sometimes contracted into -um, passum (Lucilius, 
 Martial), currum (Virg. Aen. vi. 653). Many consonantal 
 stems are increased by -i on analogy of the -i stems, e. g. 
 merc-i-um,, penat-i-um, amant-i-um (also amant-um, which is 
 not a contraction of, but an earlier form than that in -ium) : 
 but this addition is very rare with stems ending in -n, -r, -s 
 (except vir-i-um, complur-i-nm). Some consonantal stems 
 
'^i.] Noun Infiection. 123 
 
 follow the analogy of -m stems, e. g. alitu-um,' (Lucr. and Virg.) Gen. Piur. 
 beside alit-um. 
 
 0- stems (masc. and neut.) add either -um (^-om) or -orum 
 {-asam) to the stem. The first is not, as sometimes regarded, 
 a contraction of the longer form : it is in feet probably the 
 older, being the only one known to the Umbrian and Oscan 
 dialects, occurring exclusively on early coins of fifth century 
 A. u. c, and most frequently on inscriptions of an early date 
 {Bomanom, sovom=zsuorum, divom, etc.). The other form in 
 -drum gradually superseded it, and occurs commonly on in- 
 scriptions of the second century b. c. and later : and in and 
 after Cicero's time ', the form in -um was found only in 
 certain words : e. g. nwnmuTn, dena/rium, etc. ; duum, dueen- 
 tum and other numerals, especially distributive ; deum, divum, 
 virum and compounds, Italum, etc. ; nostrum and vestrum 
 (see below, p. 136). a- stems form gen. plur. in -drum; but 
 -vm, is form.ed (i) from masc. patronymics in -des {Aeneadum, 
 etc.), (2) compounds of gigno and colo [terrigenum, caelicolum) — 
 both in dactylic poetry only ; (3) from the fem. stems amphora, 
 drachma (but these are probably borrowed from Greek), -e 
 stems have the form in -rum, (dierum, etc.). 
 
 The forms boverum, nucerum, regerum, lapiderum, noticed 
 by Varro, seem (if genuine) to point to the occurrence of the , 
 longer form in consonantal stems with e = « as a connecting 
 vowel ; i being sufiixed to the stem as with the other form in 
 -um where the termination -ium is formed from a purely con- 
 sonantal stem. Another explanation supposes an addition to 
 the stem of -er, because in some words an -r is found in gen. 
 sing, (aoipenseris, cuewm&ris, etc.), and is therefore perhaps not 
 peculiar to the plural number in the words in question. 
 
 The Genitive Dual agrees in form with the Dative Dual 
 (p. 123). 
 
 A hlative Singular : — 
 
 The Ablative has been retained as a distinct form in Declen- Abiat. Sing., 
 sion only by Old Persian (Zend) and Latin. Sanskrit preserves guages re- 
 
 ' See Cicero, 'Orator,' xlvi. § 155 ; and compare Roby's 'Latin Gram- 
 mar,' § 365 (toI. i. p. 1 24). 
 
124 Noun Inflection. [chap. 
 
 AbiQt. Sing, it in masc. and neut. a- stems, Greek in adverbs in -tor, where 
 s ^ t of Sanskrit ablative ; this -t probably representing an 
 original -d retained in Old Latin and in Zend. 
 
 Thus o;ii(or=:o/itoT=saniat, abl. of sama, ' similar :' n-Ss (Ionic 
 k£s)=I. E. hvat, abl. of kva. The Sanskrit a- stems (masc. and 
 neutr.) preserve the final -t (siv^t, dsv&t) : in all others the 
 ablative is identical in form with the genitive in -as ; such 
 merging being prevented in the a- stems by the retention of 
 the longer genitive form in -sya. 
 Latin Ablative Singular : — 
 
 In Latin. Here the original -d (as in Zend d — written t by Schleicher) 
 
 has been retained ; but is only found in Old Latin and Oscan, 
 being lost in Classical Latin and Umbrian. Thus we find on 
 inscriptions senatud, praidad/ gnaivod (Ep. Scip. Appendix 
 I. i. l) ; sententiad, couentionid, and the adverbs sicprad, extrad, 
 facilumed (S. C. de Bacch., Appendix I. ii.). This form 
 facilwned, with e. g. Oscan a/mprufi-d (^=:improbe), suggests that 
 the ordinary adverbial termination in -e is an ablative in -ed, 
 from adjectives in -us, -a, -um, and thus distinguished in form 
 from the masc. and fem. ablatives in -od, -ad (which are also 
 sometimes adverbial, as in cito(d), supra(d), contra{d), etc.). 
 The original quantity of the adverbial ablative in ~e{d) is 
 generally retained, though shortened in some words in constant 
 use, e.g. bene, male. 
 
 In 0-, a-, e-, and u- stems, the long vowel of the ablative -o, 
 -a, -e, -u seems originally to have been followed by the charac- 
 teristic -d, which however fell off at an early period. The latest 
 inscription on which it occurs is the S. C. de Bacch. (i86 b. c), 
 a formal legal document with much in its orthography that was 
 archaic at the time ; and it is by no means found constantly 
 even in the earliest inscriptions. Thus on that of Scipio 
 Barbatus (see Appendix I. i. i) we find gnaivod, but patre ; 
 while on the other Scipionic inscriptions it hardly occurs at all. 
 Eitschl, indeed, holds that it was in use in the time of Plautus, 
 whom he assumes to have used it or not at pleasure : and 
 accordingly he restoi-es to the text of Plautus forms like med, 
 ted, sed {me, te, se) almost ad libitum for metrical conveni- 
 
^i-j Nonn Inflection, 125 
 
 ence ', Corssen however maintains that the final -d of abl. sing. Abiat. Sing, 
 was no longer heard or spoken, and therefore not likely to 
 have been written, in the time of Plautus and Ennius; ad- 
 mitting at the same time that Plautus might have availed him- 
 self occasionally mebri gratia of an archaic form no longer in 
 use, just as Virgil in later times used the archaic genitive aquai, 
 or infinitive yan'er^. 
 
 la consonant and i- stems we find both -i and -e as abl. 
 terminations. In classical Latin, most adjectives in -is have -i 
 (thereby securing a distinction from the neut. sing, in -e); ' 
 most substantives and participles -e. Some substantives how- 
 ever regularly have -i (see above, p. 108, note); in others which 
 usually have -e, -i is also found, especially in Lucretius (see 
 Munro on i. 978). In late and vulgar Latin all ablatives in 
 -i are weakened to -e. The history of the forms is as follows. 
 The original form was probably -id (I. E. -&i), seen e. g. in marid 
 (Columna Eostrata, b. c. 260), couentionid (S. C. de Bacch.), 
 and traceable in ante-hac (the non-elision of which is perhaps 
 due to its original form antid-Jmc). This -id became -ed and 
 then -e, which quantity is found on the Epit. of Scipio Bar- 
 batus (Appendix I. i. i.) in a Saturnian verse, Gnaivod \ patre 
 I progna | tus || . . . ., and Plant. Capt. 807 (trochaic), Turn 
 pistores scrqfipasci qui alunt furfure sues. From 150 b. c. 
 onwards -e becomes most common, -ei and -i are also found : 
 e. g. on Ep. Scip. 4 (Appendix I. i.), virtutei, ablat., and -i in 
 consonantal stems, luci Plant. Aul, 741; Ter. Ad. 841; 
 Lucr. iv. 235 ; Cic. Phil, xii. 25. The tendency of pronuncia- 
 tion to obscure and weaken all final syllables brought all these 
 forms at last down to the weakest form -e (see above, p. 57). 
 
 ' Eitschl, ' Neue Plautin.' Excurs. i. 106, 
 
 " ' Ueber Ausspraohe,' etc. II. pp. 1005-1008. The objections here 
 urged by Corssen appear almost conclusive against Kitschl's view : — e.g. 
 his citation (i) of many instances from Plautus where the final vowel of 
 abl. sing, coalesces by ' aynaloepha ' with a following vowel, with no such 
 traces of the influence of a final -d, as are found for example in Homer of 
 the lost ' digamma'; and (2) of examples from Ennius of ablat. in -S (voce 
 videtur, Corde meo, etc. in hexameters) without a trace of length by posi- 
 tion. The shortening of this -e, it may farther be observed, is itself 
 a subsequent process, presupposing the entire disappearance of -d from the 
 original ending -ed (see below). 
 
12,6 
 
 Noun Inflection. 
 
 [chap. 
 
 Loc. Sing., 
 Indo-Euro- 
 pean type. 
 
 In Greek;; 
 
 In Latin. 
 
 The Ablative Plural agrees in form with the Dative Plural 
 (see p. 129). 
 
 Locative Singular : — 
 
 The general type is -i, for Indo-European noun-stems; but 
 pronominal stems have -in, which is the older form. (The 
 locative termination of fern, stems in Sanskrit -am seems to 
 point to the oldest form of the suffix.) The locative is retained 
 as an independent case in Sanskrit, Zend, Slavonic, Lithuanian ; 
 in Greek, Latin, Teutonic, and Keltic, it has coalesced with the 
 dative and genitive (its functions being also shared with gen. 
 and abl. in Greek and Latin). 
 
 In Greek, the dative singular in -i is properly a locative form 
 TToS-i, yipovT-i, etc. ; and the locative meaning is retained in 
 such forms as MapaOavi, SaKafuvi, wktI, k.tX. The dative of 0- 
 and a- stems is a true dative (see below, p. 127): but side by 
 side with it we find locative forms Such as oixot (oiko- + 1) x°/"»' 
 {x"!'^- + 0- This locative -ot of o- stems becomes in Aeolic vi — 
 TvlSe, ii€(Tvi J and in Doric -ei, e.g. »r« (wol), rrjvel, Tovrei, reiSe: 
 this form in -« being also found in Attic d/iaxel, iravoticei, ekei 
 (=e-(co-i from stem ko-). ayx} is perhaps locative (=ayx^') from 
 a stem ayxp-, whence ayxov : and aU\ might be locative of a stem 
 aifo-= Sanskrit Sva, Latin aevo- : though .the Doric form aih 
 points rather to a stem in -s, the locative of which is aU{a)i. 
 /xoi, <To\ (Doric Tol) are probably locatives. 
 
 In Latin there are but few traces of a distinct locative case ; 
 the locative, both in form and ^functions, having become merged 
 in either the ablative or dative (? genitive) case. In consonant 
 declension forms like rv/n, vesperl, heri {=hes-i from hes, Greek 
 x6h, cp. hesternus) may be locatives, but are not distinguishable 
 from dat. or abl. in I. 
 
 From 0- stems humi, belli, foci, Corinthi, etc. are perhaps 
 locatives, but are assimilated in form to the dative (as in illi, 
 soli) or the genitive. Postri-die, quotidie, etc., and in Old 
 Latin die quinti, die crastini, seem to point to a locative form 
 merged in the dative ; and so Homae, militiae (Soma-i, mili- 
 tia-i) are perhaps originally locatives formed by adding i to 
 the stem', but are now undistinguishable from the gen. or dat. 
 
'^i-J Noun Inflection. ' 137 
 
 Ferendie, jam and the pronominal adverbs in -im (Appendix Loo. Sing. 
 II, A.) are supposed to indicate the still older locative termina- 
 tion -in=-am. 
 
 [The locative terminations in Sanskrit are -i (consonant and In Sanskrit. 
 diphthong stems), -a,u (masc. stems in i- and u-, the stem vowel 
 disappearing), -am (fem. stems in k- i- ti-), -e ( = a-|-i masc. 
 and neut. stems in a-), and -in (only in pronominal declension).] 
 Locative Plural : — 
 
 From the forms of this case in the Asiatic branch of Indo- Loo. Piur. 
 European languages (Sanskrit -su, -shu, Zend shva, shu, -shu, 
 -hva, -ha, -hu and Old Persian -swvd) an original type sva-sa 
 (sva pronominal, and sa plural sign) is postulated by some 
 philologists (Schleicher, Comp. § 256). However this may be, 
 the Sanskrit -su evidently corresponds to the Greek -tri or -ara-c 
 (aFi) of the dative plural, which is thus, like the dat. sing, in 
 N -(, really locative, -a-i or -<r(n are sometimes added direct to 
 consonant stems, sometimes by a ' connecting vowel ' e : e. g. 
 TToa&i (=:7rd8-o-i), but also 7rd8-f-(r(ri (Epic.) j kv(t\ and Kvv-c-irtTi, 
 (/cue-) ; TTciKl-e-o-cn, TrdXicri and noKea-i from n6\i-. In o- and a- 
 stems the forms ittttoio-i, x^pa'ci may (as is suggested by some) 
 result from a lengthening of the stem by t [liriro-t-a-t, ^mpa-i-o-t) : 
 but it is simpler to regard them as formed by the addition of 
 the plural sign -o-t to the locative sing. (twTroi-a-i, ;(a)pai-(ri). In a 
 few feminine forms like Bvpcun, 'Adrivrjm the termination -a-i 
 appears to be added to the stem ; unless we may suppose these 
 forms to have been written with an ' i subscriptum,' like the 
 Epic forms -30-1, -i;s, which are from lengthening of a in -aicn, 
 -ais. 
 
 In Latin, the locative plural is merged in the form of dat. 
 and abl. plur. The dat. abl. in -is, however, of 0- and a- stems 
 is very possibly a locative form (see below, p. 129). 
 
 No Locative Dual is found in Greek or Latin : but Sanskrit 
 has a form ending in -6a. 
 
 Dative Singular : — 
 
 General type -ai, Sanskrit -e, Greek -9 (=0- 01), a (=a-|-aOi Dat. Sing., 
 Latin -i. Different views are held as to the origin of this ter- typ°. 
 mination, viz. ; — • 
 
ia8 
 
 Noun Inflection. 
 
 [chap;' 
 
 Dat. Sing., 
 
 General 
 
 type. 
 
 In Greek; 
 
 la Latin. 
 
 (i) That it is a strengthening of the locative -i. 
 
 (2) That it represents the preposition dbhi, bh being lost, as 
 in Sanskrit sivais, instrumental plur. of siva, and as ^ is 
 , lost in Doric iiuv and Epic rctv (= ma-bhyam, tu-bhyam). 
 These pronominal datives (cp. Latin tibi, sibi, etc.) give bh : and 
 as pronouns generally preserve more archaic forms than nouns, 
 they perhaps point to the original form of the dative. The use 
 of the preposition ahhi, ' towards,' to form the dative would be 
 analogous to the Latin idiom of ad carnificem dabo, etc. found in 
 Plautus, and to the use in modern languages of the prepositions 
 to, d,, zu, etc. to express the dative. 
 
 In Greek, a- stems (a-, rj-, o-) alone have the true dative termi- 
 nation, o(Kij)=oiKo--|-oe, 6ea.=6ea- + m: in all others the locative 
 in 1 is used. Infinitives in -/ievai, -evm, -mi, -ai are probably 
 datives of consonantal stems (see below, chap. viii). 
 
 In Latin, the dative in -* of consonantal and i- and u- stems 
 perhaps represents Indo-European -ai. Inscriptions of fifth 
 century A. u. c. give -e, which is analogous to Sanskrit -e=-ai: 
 e. g. Junone, matre, salute, Diove (Jovi) : and this form appears 
 to have been retained in some technical phrases, e. g. solvendo 
 aere alieno, jure dicundo in Livy and Suetonius. Umbrian has 
 -e : pa<re= Sanskrit pitr§ {patri). Later inscriptions (sixth 
 century A. v. c. to time of Augustus) give -ei : e. g. quoiet (Ep. 
 Scip. 4, Appendix I. i.) Apolenei, Biovii, Hercolei, etc. Oscan 
 has -ei; paterei, Biuvei, etc. Finally -ei became -i, which does not 
 appear on inscriptions before the date of the Grracchi. Corssen 
 considers -ei to be the original dative suffix and -i the locative : 
 Schleicher (Comp. § 254) regards -i, -e, -ei as varieties of the 
 same form, but locative. It would seem perhaps more reason- 
 able to regard the three forms as identical : and the history of 
 Latin orthography will supply many analogies to this variety of 
 spelling (e. g. ni, ne, nei : cp. Eitschl's view of the relations of e, 
 ei, and I quoted in Eoby's Latin Grammar, vol. i. § 268). 
 
 In a-, e-, 0- stems the oldest forms appear to be -ai, -ei, -oi 
 (quoi, populoi, traceable in huic^Jioic). With such stems the 
 first letter of the termination -ai united itself to the stem vowel, 
 and the results -01, -di, -ei seem to have been sometimes pro- 
 
"^i] Noun Inflection, 139 
 
 nounced as dissyllable ; as is still the case with e- stems, where Bat. Sing. : 
 the full form is retained \ ^^^ 
 
 In 0- stems the final -i was dropped {jpojpulo^populd-i^^ 
 po]pulo-oi :) just as in Greek (^mTta^iTiTa-i,="nr7!o-oi, etc.), where 
 however the traces of i remain in ' i subscriptum.' In a- stems 
 i survives in Classical Latin -ae=-ai: but, dative forms like 
 Matuta, Tusco-lana, etc. are parallel to Greek 6e5.=6em=.6€a-aX, 
 and Latin populo, in the loss of -i. In e- stems also a similar 
 form in -e is found : e. g. die, Plaut. ; re, Trin. 635, 657 : Jide, 
 Anl. 659, Amph. 391 ; facie, Lucilius, etc. : and a monosyllabic 
 pronunciation metri gratia of -ei gives the intermediate stage. 
 Compare the remark of Aulus Gellius (ix. 14), 'In casu dandi 
 qui purissime locuti sunt, non " faciei " uti nunc dicitur sed 
 "facie" dixertint.' 
 
 Dative Plural: — 
 
 General type hhyams, i. e. Vhyam, of dat. sing, (as seen in !)»*• P>"r-. 
 Sanskrit pronominal declension, p. 143) + plural sign -s. The pean type. 
 Old Prussian -mans of dat. plur. is the natural representative, 
 by the laws of phonetic change, of Indo-European hhyams, and 
 therefore confirms the inference that this is the primitive form 
 -mus in Lithuanian (mumus, jumus=-nobis, vobis) points to the 
 same form : for if the original had been Sanskrit bhyas, Lith. 
 would have -mas ; but u is accounted for by the nasal m. 
 
 This case form appears in all Indo-European languages except 
 Greek, which employs locative plur., as in sing, (see p. 126). 
 In Sanskrit -hhyams becomes -bhyas (cp. ace. plur. -as^am,-s, 
 p. 116). 
 
 In Latin, -hhyas became -bios or -hius, then -hos, -bus (for In Latin, 
 loss of i cp. min-us=mi'nius and see also on p. 59) : and a 
 parallel form appears in no-his, earlier noheis. This -bus is the Dat. in -iw. 
 regular termination for consonant, i-, and u- stems, and is also 
 found in amho-bus, duo-hus (0- stems) ; deahus, filiahus, liberta- 
 bus on inscriptions, and amha-bus, duahus, classical (a- stems) ; 
 diebus, rebus, classical (e- stems), -i is usually added before it 
 
 ' The dative termination -ei of the ordinary fifth declension seems to 
 have been variously scanned as e-j, ei, and el; see for examples Eoby, 
 'Latin Grammar,' § 306 (i. p. 122). 
 
 E 
 
130 Noun Inflection. [chap. 
 
 Dat. Piur. to consonant stems {nomin-i-bus, etc.) ; but in lo-bus, hu-hu8=. 
 hov-hus we have possibly the remnant of an earlier formation by 
 adding -6ms direct to the stem. In i- stems i is found as « in 
 Old Latin, e. g. temjpestatebus, navebos : and in u- stems, u some- 
 times becomes i, as fruetihus. 
 
 Dat. in-i». The dat. (also abl.) plur. of 0- and a- stems (with the 
 exceptions above given) ends always in -is, of which form there 
 are two explanations : — (i) that it is dative, arising from -fios 
 =.-bhyas, which then becomes -hios (cp. mi-hei beside ti-hei) and 
 then by contraction -is (see Schleicher, Comp. § 261, and on 
 y=6A above, p. 69). This however is very hypothetical ; and 
 it seems simpler to believe (2) that -is is a locative termination ; 
 so that musts, dominis^musais, dominois=nmsaisi, dominoisi, 
 and correspond exactly to Greek x<"P<"<''', aypoiai (see above, 
 p. 127). That -ois, -ais were the original terminations of the 
 dat. plur. is shown by the old forms oloes {illis), privicloes 
 {priviculis) noticed by Festus, and by the other Italian dialects. 
 Thus an old inscription (possibly of Latin origin 1) gives suois, 
 cnatois {mis, gnatis). Oscan has Neidanuis, legatuis, diumpais 
 (Nolanis, legatis, lymphis); and in Umbrian the dat. plur. 
 of 0- and a- stems ends in -eis, -es, -is (later -eir, -er, -ir), 
 and of i- stems in -ds -es (perhaps on analogy of a- and 0- 
 stems) '- 
 
 Dat. Dual. Dative Dual: — 
 
 Indo-European -hhydms, lengthened from -hhy&ms. Sanskrit 
 here drops the -s, and has -bhyam. 
 
 In Greek -hhyams became something like -tjiuov, which became 
 -ipiv and finally -iv, as in 0- stems i7nro-iv^ijnro-(j)iv, x^P"'^"^ 
 Xapa-<piv. All other stems follow the analogy of 0- stems, and 
 thus -o-cf)iv, -oiv is the usual termination throughout, e. g. ytw'oo/, 
 yevoiv ^ y€Ue(T-o-(j>iv, iraTcpoiv = narep-o-fjiiv, evpi-oiv := eipeF-o-<j>iv 
 (stem evpv- with stem vowel raised). The Homeric forms to'uv, 
 fi\c(^apouv, etc. (from stems to- /3Xc<^a/)o-) appear to have an i 
 added to the stem ; so that ToCiv^To-i-^iv. The same form is 
 found in some consonantal stems, e. g. ttoSouV = woS-o-i-<j>iv, 
 
 ' See Ferrar's ' Comparative Grammar,' p. 269. 
 
Ti.] Noun Inflection. 
 
 131 
 
 Ssipfivouv=2eiprjv-o-i-(j)iv, where 0-, and afterwards (-, have been 
 added to the original stem, each under the influence of 
 analogy. 
 
 There is no trace of -hhydms in Latin or any Italian dialect. 
 
 Instrumental Sinqula/r: — Instru- 
 
 „, mental Sing. 
 
 inere appear to have been two Indo-European forms, (i) -(2, indo-Euro- 
 (2) -IM : and it is suggested (Schleicher, Comp. § 258) that 
 these originally corresponded to the twofold meaning of {a) 
 comitative ('I went with him'), (6) 'instrumental' proper ('I 
 cut it with a knife'), which are united in the Latin ablative 
 case, and in our preposition ' with.' 
 
 (i) - a is found iji Sanskrit {vtk-h) : and in Greek possibly 
 in the adverbial forms aim. (Doric Ayta), Six" (^'X"), Taxa, ^t 
 (II. i. 144, xiv. 499), dWaxrj, irdvTt] (Doric rravTo), B^:=dya=yd 
 from pronominal stem ya, whence the locat. jam (see for 
 ' parasitic d,' p. 80). 
 
 (2) -bhi, which does not appear in Sanskrit, is in Greek -04, 
 a termination common in Homer, and not to be confused with 
 the supposed earlier form of the dative dual -lu mentioned 
 above. It is used as (a) comitative (0/1' rjol (jimvoiievrjcjiiv), (6) 
 instrumental proper {§(pi ^k'ph Od. xxi. 313, cp. II. xvi. 734); 
 but more often in a locative or ablative signification, by the 
 easy transition from the notion of ' circumstances under which ' 
 or ' by which ' to ' place at which ' or ' from at which : ' e. g. 
 eV ia-xapo<j>iv ; or with aTTo, c|, fK ttoito^w, ' from on the sea,' 
 (whence -<l>iv has sometimes been wrongly interpreted as a 
 genitive termination). 
 
 Latin offers no trace of either -d or -bhi. 
 
 Instrumental Plural : — 
 
 Indo-European -bhis, i.e. bhi + s, of plural. Sanskrit has -bhis. 
 except in a- stems, where bh disappears (asvais): the Vedas 
 however show dsve-bhis. 
 
 In Greek the final s is lost after v (see p. 68) and the form is 
 therefore identical with the singular -(j)iv, e.g. mOcjiiv, II. ii. 794 i 
 KOTvKr]Sov6(jii.v, Od. V. 433, Beocfiiv, etc. 
 
 Comparison of Adjectives : — 
 
 The declension of adjectives has been sufficiently explained ^omparison 
 K 2 tivea. 
 
13a Noim Inflection. [chap. 
 
 Comparisoii under that of substantives : and there only remains the question 
 tives. of the formation of ' degrees of comparison.' This is really part 
 
 of the composition of words, i. e. the formation of stems from 
 roots : for comparative and superlative are formed by addition 
 to stem of positive of particular suffixes, in no way different 
 from other suffixes, and not confined to adjectives. But the 
 wide and general use of pairticular suffixes for this purpose, 
 and the order usually observed in grammars, make it convenient 
 to consider them at this stage. 
 
 Comparative degree. The supposed type of this stem in 
 Indo-European is formed by a suffix -yans {=yan-ta) or -tara. 
 These may be derived either from (i) verbal, or (2) pronominal 
 roots. Those who derive from verbal roots connect -yan with 
 Jndo-European ya, 'to go,' whence Sanskrit yS,, Greek Uvai; 
 'tara with Indo-European tar, 'to cross over,' whence Latin 
 trans, English through : both roots thus signifying progression, 
 and heightening the idea of the positive. 
 
 But it seems better, without trying to attach so definite a 
 meaning to the suffixes in question, to regard them as derived 
 from pronominal roots and akin, to certain other pronominal 
 suffixes traceable in Indo-European languages, -yans e.g. con- 
 nected with the common suffixes -ant (part. act. in Sanskrit, 
 Greek, Latin) : -vant (' provided with anything,' Greek fevr in 
 lx6v6-pevT, xapifVT-, etc., -«9, -ecrcra, -ev) j and -manf (mana-, mari') 
 in Afj-iiav, irol-iir]v, ani-mo, al-mo, certa-men, car-men : partic. 
 -jjxvo-, plur. 2 per. -mini, etc. 
 
 -tara-=-ta-ra, the latter of which elements sometimes expresses 
 Compar. the idea of comparative as in Latin sup-er-us, etc. 
 
 suffix j/am. ' ^ *■ 
 
 (i) -yan {^-yarns') or -ians. 
 
 Sanskrit -lyas (base of comparative). 
 
 nom. sing. lyan(B) masc, lyaa neut. 
 ace. ly&nsam. 
 
 in^str. lyas-&. 
 
 Cheek -lov {-tav nom. — tov-r, s being lost and the vowel 
 lengthened), before which final -o, -v, -po, of stem, are drqpped : 
 e.g. <j)[\-iov (^(Xo-), fj8-iov (^8u-). 6a<r(rov=:Tax-iov (raxo-), jjaa- 
 
VI.] Noun Inflection. 133 
 
 aov^imK-iov {jiaKpo-), alax-lov {auTX-po), iieiCov:=fi£ycov {fieyas), foot Comparison 
 
 i\ of Adiec- 
 
 1^7- )■ tives. 
 
 J. . ._ ( -ior, -or, 
 
 Latin -los, ■! . 
 
 \ -lus, -us. 
 
 -ids is the oldest form, n being lost as in accus. plur., but 
 retained in adverbial -iens. In adding -ios to vowel stems, the 
 stem vowel is omitted ; ^db-ior (prob-o-), sapient-ior. 
 
 In Classical Latin -ios split into -ior, -ios ; and the two forms 
 then served to mark distinction of gender (melior, melius). 
 The distinction between them was not originally very marked ; 
 and remains of Old Latin give e. g. melios, masc, and prior, 
 posterior, neut. ^- 
 
 -ior=-ios. o long in Old Latin (as in oblique cases), so 
 Plant. Amphit. 548 : — 
 
 ' Atque quanto ndx fuisti l6ngior haec prdxuma.' 
 Capt. 782 (cMctior), and in neut., tts, Menoeohmi, ^26 : — 
 ' Proin til ne quo abeas I6ngius ab aedibus.' 
 
 (2) -tara, Sanskrit tara, Greek -repo-s. Ctompar. 
 
 ^ ' suffix -tara. 
 
 In Latin it appears in e. g. pos-teri, ce-teri, al-ter, neu-ter, 
 u-ter, dex-ter, citra, ultra, frustra ; and possibly adverbs in -ter 
 and igi-tu/r, etc. (see however Appendix II, A.). 
 
 In pure comparatives, only in composition with the other form 
 -ios, e. g. ci-ter-ior, de-ter-ior, etc., or sin-is-tero, min-is-tero-, etc. 
 (cp. Greek XaX-iore/jo-r). 
 
 \mag-ister, min-ister^^ihe greater, the less person, cp. English 
 'mayor,' 'major,' 'minor.'] 
 
 Comparison of A djectives. — Svperlative. 
 
 The elements of superlative formation in Indo-European 
 languages are -ta and -ma, either separately or combined, or 
 either of them doubled, or in combination with the comparative 
 stem ; as under the following heads : — 
 
 ' On the production of these forms by assimilation, see above, pp. 75, 6 ; 
 and compare Schleicher, ' Comp.' § 148 d, e (pp. 224-226), and Peile, p. 228. 
 
 ^ Priscian quotes ' senatus consultum priori ' bellum Punicum posterior,^ 
 and says, ' Vetustissimi etiam neutriim in -or finiebant, et erat eadem 
 terminatio communis trium gencrum.' In the appropriation of -ior to the 
 masculine gender we perhaps see the result of analogy with substantives in 
 -or, honor, labor, etc., which are usually masculine. 
 
134 Noun Inflection. [chap. 
 
 sufixes''™ ^- "*"' J ill numerals irparo-s, ckto-s, sexto-. 
 2 -m,a ; wpofio-s, prima-, summo-, etc, 
 
 3. ta-ta ; the ordinary Greek superlative in -raro-s. 
 
 4. ta-ma ; in Latin optumo-, dextumo- (or without the superla- 
 tive iiea,,flnitimo-, maritimo-). With these Corssen classes the 
 superlatives fadllimMS, aoerrvmus, veterrimus, which forms he 
 ho\ds=facil-timo-, aeer-timo-,veter-timo-; -timo after l,r becom- 
 ing first -simo-, and then by assimilation (p. 74) -lima-, -rimo-. 
 Thus proonmus := p^opic-timius (from an adjectival stem pro- 
 pica-) ; thence prapicsumus and by contraction proximus. 
 MaccumMS = mag-tum,us, mag-sumus. Another explanation 
 however (see Roby's Latin Grammar, Preface, p. Ixi) makes these 
 forms=-ios or-«s of comp. + -i + -mus (-mas); e.g.facillimus=- 
 facil-is-i-mus under strong contraction. The difficulty here is 
 in the insertion of i between is and mus. Both theories are 
 possible : Corssen's however follows a wider analogy, and 
 (except in one or two cases) rests upon an easier contraction 
 (see below, number 8, on -issimus). 
 
 5. ma-ta; Greek jri/iaTo-s, e^hoimTo-s. 
 
 6. yans-ta ; Greek superlative in ta-ro-, fieytaro-s, etc. 
 
 7. yans-ma=ias-ma=:is-ma=:i-ma in minima-, plurimo-, etc., 
 where i is all that remains of the comparative suffix. 
 
 8. yans-ta-ma=is-t%i/mo, found in two words, sallistumum 
 (tripudium), Cic. de Div. ii. 34. 7 2, explained as ' perfect ' from 
 sonMs-=tatus ; and sinistumus, superlative of sinister, as deoctu- 
 mus of deleter. On the analogy of these two forms, Corssen 
 explains the regular Latin superlative in -issimus=i-is-tum'U:S= 
 -ias-tumus (see above, number 4, onfacillimus, etc.). Here again 
 another explanation is offered, that -issvmor=-is-imo ', the double 
 s being due partly to the desire to indicate the length of the 
 preceding syllable, partly to an attempt to preserve the sound 
 of s sharp. It seems doubtful, however, whether s is ever 
 ' sharpened ' to ss for merely phonetic reasons, except where a 
 syllable is lost before it (as in locassim, etc.) ; and of this 
 moreover there is no example in noun-forms. 
 
 9. tara-ma, in extremo-, postremo-. 
 
 ' See Roby's Latin ' Grammar,' vol. i. Preface, p. bd. 
 
yi.] 
 
 Noun Inflection. 
 
 -^iS 
 
 Paradigms of Koun Inflection in Sanskrit *, Greek, and Latin. Paradigms 
 
 (From Bopp, Schleicher, and Ferrar.) Lto^ ^"" 
 
 I (a). Consonant Stems : — Consonant 
 
 Stems. 
 
 
 Sanskrit. 
 
 Greek. 
 
 Latin. 
 
 (i) stem 
 
 vai- (fern.) 
 
 iir- (fern.) 
 
 voc- (fem.) {voc-i) 
 
 Singular : — 
 
 
 
 
 Noiu. 
 
 vak 
 
 'iTT-S, 
 
 voo-s (tox) 
 
 Accus. 
 
 vai'-am 
 
 0-n-o. 
 
 v6o-em 
 
 Gen. 
 
 vaft'-fe 
 
 6n-6s 
 
 v6o-is 
 
 Abl. 
 
 
 
 voc-e(d) 
 
 Loc. 
 
 vai'-I 
 
 bit-i (dat.) 
 
 
 Dat. 
 
 vai'-e (=ai) 
 
 
 voc-i 
 
 Instr. 
 
 vaifc'-a 
 
 
 
 Voc. 
 
 vak 
 
 
 
 Dual:— 
 
 
 
 
 Nom. Ace. 
 
 vai'-a, -au 
 
 OTT-e 
 
 
 Gen. Loc. 
 
 vaft'-ds 
 
 
 
 Dat. Abl. 1 
 Instr. J 
 
 
 hir-o-iv 
 
 
 ♦vag-bbyam 
 
 
 Plural :— 
 
 
 
 
 Nom. 
 
 v^i'-as 
 
 SiT-es 
 
 voo-es (voci-es) 
 
 Ace. 
 
 vaF-as 
 
 Sir-as 
 
 vqe-es 
 
 Gen. 
 
 vaii'-am 
 
 i-ir-av 
 
 voc-um 
 
 Loc. 
 
 vak-sbti 
 
 m-al (dat.) 
 
 
 Dat. Abl. 
 
 *vag-bhyd,3 
 
 
 voc-i-bus 
 
 Instr. 
 
 *vag-bhi3 
 
 (-*'") 
 
 
 ' In the orthography of Sanskrit words in this and the following Para- 
 digms, I have in the main followed Schleicher. 
 
 * fe of stem assimilated to the media Ih (see p. 77). This Tc is the palatal 
 modification (cA) of guttural k, hioh appears only in nom. vak. The sign 
 k is preferable to cK here, as showing the easy transition from the nomina- 
 tive to the oblique cases (see p. 33, note i). 
 
136 
 
 Noun Inflection. 
 
 [chap. 
 
 Inflection of 
 
 Consonantr 
 
 Stems. 
 
 
 Sanskrit. 
 
 Greek. 
 
 Latin. 
 
 
 (2) Stem 
 
 hharant (masc. n.) 
 
 *(j)lpovT- (m. n.) 
 
 fermt (m. f. n.), 
 
 
 Singular : — 
 
 
 
 ferent'i 
 
 
 Kom. 
 
 bMran, bharat (n.) 
 
 (pipoiv (-OVT-S) 
 
 feren(t)-s 
 
 
 AocuB. 
 
 bh^rant-am, 
 bh4rat (n.) 
 
 tpipovT-a 
 
 ferent-era 
 
 
 Gen. 
 
 bb^at-as 
 
 <pipOVT-Ot 
 
 ferent-is 
 
 
 Abl. 
 
 
 
 ferent-e(d) 
 
 
 Loo. 
 
 bharat-i 
 
 <pipOVT-i 
 
 
 
 Dat. 
 
 bhdrat-e 
 
 
 ferent-i 
 
 
 Instr. 
 
 bh4rat-a 
 
 
 . 
 
 
 Voc. 
 
 bMran.bhaiat (n.) 
 
 
 
 
 Dual:— 
 
 
 
 
 
 Nom. Aoc. 
 
 bharant-a, -au (m.) 
 bharant-1 (n.) 
 
 <t>ipOVT-f 
 
 
 
 Gen. Loo. 
 
 bMrat-OB 
 
 
 
 
 Dat. Abl. 1 
 Instr. J 
 
 *bb&rad-bhyam 
 
 <j>ep6vT-o-iv 
 
 
 
 Plural:— 
 
 
 
 
 
 Nom. 
 
 bharant-as, 
 bMrant-i (n.) 
 
 <pipovT~es 
 -a 
 
 ferent-es (-ies) 
 
 
 Ace. 
 
 bh&rat-as (m.), 
 bhirant-i (n.) 
 
 ffpovr-as 
 -a 
 
 ferent-es 
 
 
 Gen. 
 
 bbarat-am 
 
 (pep6vTcuv 
 
 ferent-ium (-um) 
 
 
 Loo. 
 
 bMrat-su 
 
 <pipoiiT-ai (-oviri) 
 
 
 
 Dat. Abl. 
 
 *bh4rad-bhya9 
 
 
 ferenti-bus 
 
 
 Instr. 
 
 *bharad-bhis 
 
 (-*">') 
 
 
 
 (3) Stem 
 
 mdnas- (n.) 
 
 nlvos, liives (n.) 
 
 genua, genes (n.) 
 
 
 
 dwr-mmias (m. f.) 
 
 Sva-jixvis (m. f.) 
 
 vetii,s, vetes (m. f.) 
 
 
 Singular : — 
 
 
 
 (s becomes r) 
 
 
 Nom. 
 
 m4naa 
 
 nivos 
 
 genus, arbos (f.) 
 
 
 
 dur-mana8(m.f.) 
 
 ivaniv^i 
 
 vetus 
 
 
 Aoo. 
 
 mdnas 
 
 jiivos 
 
 genus 
 
 
 
 duimanas-am 
 
 Svaufvla-a, -ia, -ij 
 
 veter-em (m. f.) 
 
 
 
 (m. f.) 
 
 
 
 Greek forms a feminine by addition of suffix -ja (ya) to this stem, 
 tpipovT-ja, ipipovaa, which is declined as a fem. a- stem. The same for- 
 mation exists in Latin as a fem. noun : 'patiea{t)a, patient-ia. 
 
 * Assimilation of t to bh. 
 
VI-] 
 
 ,Noun Inflection. 
 
 ^37 
 
 
 Sanskrit. 
 
 Greek. 
 
 Latin. 
 
 Gen. 
 
 mSnas-as 
 
 lifvka-os, -COS, -ovs 
 
 gener-is 
 
 Abl. 
 
 
 
 gener-e(d) 
 
 Loc. 
 
 m&nas-i 
 
 fi6V€<T-i, tievu 
 
 (run ?) 
 
 Dat. 
 
 manas-e (=ai) 
 
 
 gener-i 
 
 Instr. 
 
 iu4nas-a 
 
 (ax«0'-^' = bhi) 
 
 
 Voc. 
 
 manas 
 
 nivos 
 SvaiJ.evis (m. f.) 
 
 genua 
 
 Dual :— 
 
 
 
 
 Nom. Aoc. 
 
 mdnas-i (n.) 
 
 (liivfff-f) niyrj 
 
 
 / 
 
 durmanas-a, -au 
 
 Svaiievicr-e, 
 
 
 
 (m. f.) 
 
 Svciiivij 
 
 
 Gen. Loo. 
 
 manas-os 
 
 
 
 Dat. Abl. 1 
 Instr. J 
 
 mano-bhyam 
 
 jifviao-iv, -epiv, 
 -oTv 
 
 
 Plural :— 
 
 
 
 
 Nom. 
 
 mdnams-i (n.) 
 
 lihta-a, -ea, -ij 
 
 gener-a (n.) 
 
 
 durmanas-as 
 
 Svaiiet>e<r-es(ia.{. 
 
 veter-es 
 
 
 (m. f.) 
 
 -t(S, -us 
 
 (m. f.) 
 
 Aeo. 
 
 mAnams-i 
 
 nivea-a, iiivr] 
 
 gener-a (n.) 
 
 
 dur-manas-as 
 
 5vfffjLcyeff~as (m . f .) 
 
 veter-es (m. f.) 
 
 
 (m. f.) 
 
 -cas, -eTs 
 
 
 Gen. 
 
 m4nas-am 
 
 }jLev4iy-aVj ~iwv, -aiv 
 
 gener-um 
 
 Loc. 
 
 manas-su 
 
 nivea-ai, nlveai 
 
 \ 
 
 Dat. Abl. 
 
 mAno-bhyas 
 
 
 gener-ibus 
 
 Instr. 
 
 mdno-bhis 
 
 (liivea-<pi) 
 
 
 N. B. — Maso. and fem. forms are only given where they differ &om those 
 of the neuter stem. 
 
138 Noun Inflection. 
 
 I (8). Stems in i- : — 
 
 [chap. 
 
 
 Sanskrit. 
 
 Greek. 
 
 latin. 
 
 Stem 
 
 dvi- (m. f.) 
 
 iro\i- (fern.) 
 
 oei 
 
 
 van (n.) 
 
 iSp.. (adj.) 
 
 mart- (n.) 
 
 Singular : — 
 
 
 
 
 Nom. 
 
 dvi-B van (n.) 
 
 v6\i-s iSpi (n.) 
 
 ovi-s mare (n.) 
 
 Aocus. 
 
 dvi-m v^ri 
 
 it6\t-v ISpi 
 
 ove-m mare 
 
 Gen. 
 
 aves v5ri-n-aB 
 
 iti\i-os, ir(i\€-<os 
 
 ovis 
 
 
 6vy^ (f.) 
 
 iriS\jj-os = v6\fy-os 
 (-««, adv.) 
 
 
 Abl. 
 
 
 
 ove(d) mari-(d) 
 
 Loc. 
 
 Avau viri-n-i 
 
 ir(5X6-i', irii\6j 
 
 
 
 iivyam (f.) 
 
 ir<5A.7)-i- 
 
 
 Dat. 
 
 &vay-e v£ri-n-e 
 ^vy-al (f.) 
 
 
 ovi 
 
 Instr. 
 
 &vi-n-» v&i-n-a 
 ^vy-a (f.) 
 
 
 
 Dual :— 
 
 
 
 
 Nom. Ace. 
 
 kn v^ri-n-i 
 
 TTlSXl-E, 1l6\fi 
 
 
 Gen. Loc. 
 
 Syy-os v^ri-n.5B 
 
 
 
 Dat. Abl. 1 
 Instr. 
 
 4vi-bhyam 
 
 {itoKi-o-iv) ■noKfoTv 
 
 
 Plural:— 
 
 
 
 
 Nom, 
 
 4vay-aa viri-n-i 
 
 v6Kri-et = ir6Key-es 
 
 oves mari-a 
 
 
 
 T6\i-ts, n6Kfis 
 
 (ovi-es) 
 
 
 
 ISpt-a (n.) 
 
 
 Ace. 
 
 4vi-n (m.) vSri-n-i 
 
 ir6\tj-as tSpi-a 
 
 ovea mari-a 
 
 
 avi-B (f.) 
 
 nS\eis 
 
 
 Gen. 
 
 avT-n-am 
 
 vo\i-ojy, n6\(0)v 
 
 ovi-um 
 
 Loc. 
 
 Avi-shu 
 
 v6\iat, voKi-ai 
 ■noKl-i-aai 
 
 
 Dat. Abl. 
 
 avi-bhyas 
 
 
 ovi-bus 
 
 Instr. 
 
 dvi-bhia 
 
 
 
 N.B. — Neuter forms are only given where they differ from those of 
 masc. and fern, stems. 
 
VI.] Noun Inflection. 
 
 I (c). Diphthongal Stems : — 
 
 139 
 
 
 Sanskrit. 
 
 Greek. 
 
 Latin 1. 
 
 Stem 
 
 I. nau- 
 
 
 I. vav' 
 
 
 I. (navi-) 
 
 
 2. gau- 
 
 
 2. 0m- 
 
 
 J. hou- (boi;-i) 
 
 Singular : — 
 
 
 
 
 
 
 Nom. 
 
 nau-B 
 
 gau-s 
 
 coC-s 
 
 ;SoC-s 
 
 navi-B bo-s (bous) 
 
 Ace, 
 
 n^T-am 
 
 ga-m 
 
 vav-v 
 
 0oS-v 
 
 nav-em bov-em 
 
 Gen. 
 
 nay-as 
 
 go-s 
 
 vriF-6s 
 
 Vi-ilS 
 
 0of-6i 
 
 navi-3 bov-is 
 
 AbL 
 
 nav-iJs 
 
 gos 
 
 
 
 naTe(d) bove(d) 
 
 Loo. 
 
 nav-i 
 
 gav-i 
 
 vrip-i 
 
 0OF-1 
 
 
 Dat. 
 
 nav-e 
 
 gdv-e 
 
 
 
 nav-i bov-i 
 
 Instr. 
 
 nav-a 
 
 gav-a 
 
 vav-(pi 
 
 
 
 Voc. 
 
 (as nom.) 
 
 
 vav 
 
 00V 
 
 (as nom.) 
 
 Dual :— 
 
 
 
 
 
 
 Nom. Aco. 
 
 uav-a, -5u 
 
 g^T-a, -au 
 
 "Sf-f 
 
 06f-i 
 
 
 Gen. Loo. 
 
 nav-6s 
 
 gd,v-5s 
 
 
 
 
 Dat. Abl. \ 
 Instr. / 
 
 nau-bhySm 
 
 go-bhyam 
 
 vt,po-tv 
 vf-oiir 
 
 06Fo^v 
 
 
 Plural:— 
 
 
 
 
 
 
 Nom. 
 
 n^v-as 
 
 g^v-as 
 
 vfif-K 
 
 06f.(s 
 
 naves boves (ies) 
 
 Aco. 
 
 nSv-aa 
 
 g^T-as 
 
 tijf-as 
 
 0of-as 
 
 naves boves 
 
 
 
 gas 
 
 vav-s 
 
 0OV-S 
 
 
 Gen. 
 
 nav-am 
 
 gdv-am 
 
 vrjp-wv 
 vf-av 
 
 0of-Sjv 
 
 navi-um bo-um ^ 
 bov-om 
 
 Loo. 
 
 nau-shii 
 
 go-shu 
 
 vriv-iri 
 rav-ai 
 viiff-ain 
 
 0ov-al 
 06fi-ttai 
 
 
 Dat. Abl. 
 
 nau-bhj&B 
 
 gd-bhyas 
 
 
 
 navi-buB bo-bus 
 bii-bus 
 
 Instr. 
 
 nau-bhis 
 
 gd-bhis 
 
 vav-<piv 
 
 
 
 * The disappearance of diphthongs in Latin (see p. 56) leaves but little 
 trace of diphthongal stems, navis is throughout an i- stem: but the 
 declension of hos retains traces of a stem 60M- or iao- corresponding to 
 Greek 0ov-, 0of-. 
 
146 Noun Inflection, 
 
 II (a). Vowel Stems (masc. and Beut.) : — 
 
 [chap. 
 
 
 Sanskrit. 
 
 Greek. 
 
 Latin. 
 
 stem 
 
 fisva- (m.) 
 
 tinro- (m.) 
 
 equo- (m.) 
 
 
 yug4 (n.) 
 
 fuyo- (n.) 
 
 jugp- (n.) 
 
 Singular : — 
 
 
 
 
 Nom. 
 
 isva-B 
 
 ITTITO'S 
 
 equo-s (equus) 
 
 
 yug^m 
 
 Cvyb-v 
 
 jugu-m (-om) 
 
 Aco. 
 
 &va-m 
 
 Imro-v 
 
 equo-m 
 
 
 yugi-m 
 
 (vy6-v 
 
 jugu-m 
 
 Gen. 
 
 ^sva-sya 
 
 iwfro-(^a)io ' 
 
 equo-is 
 
 Abl. 
 
 itsva-t 
 
 iTTTTO-o, tnirov 
 
 equo-i, equi 
 equo(d) 
 
 Loir. 
 
 asre (asva-i) 
 
 (oi«0-l, OlKOt) 
 
 (domi=domo-i?) 
 
 Dat. 
 
 ^va-y-a 
 
 ITTirO-OI, iTTTry 
 
 equoi=equ6-oi 
 equo 
 
 Instr. 
 
 iCsven-a 
 (Vedic ^va) 
 
 VOVT6-lpl 
 
 
 Voo. 
 
 Asva. 
 
 iTTtre ( = i7nro-) 
 
 eque (equo) 
 
 
 
 ivy6.v 
 
 jugu-m 
 
 Dual:— 
 
 
 
 
 Nom. Aco. 
 
 isva,, &SVS.V. 
 
 iB-iro-c, iirirou 
 
 
 Gen. Loo. 
 
 isva-y-os 
 
 
 
 Dat. Abl. 1 
 Instr. J 
 
 ^va-bhyam 
 
 tmto-iv 
 
 
 Plural :— 
 
 . 
 
 
 
 Nom. 
 
 asva-a 
 
 tiriro-i 
 
 (equo-es, equeia) 
 
 
 (Vedio dert-aas) 
 
 (vya (n.) 
 
 equi 
 
 
 yug^n-i (n.) 
 
 
 jug-a (m.) 
 
 
 (Vedic yagi) 
 
 
 
 Aco. 
 
 isvsa (=a»van-B) 
 
 tvnovs = TTTirov-s 
 
 equos=equom-s 
 
 
 yugSn-i 
 
 (vyi 
 
 juga 
 
 Gen. 
 
 ^a-n-am 
 
 tirnS-av, iirmiv 
 
 equo-rum 
 equCm = equo-om 
 
 Loc. 
 
 iCsve-shu 
 
 tirwm-m, firiroi-a 
 
 (equo-is) equis 
 
 Dat. Abl. 
 
 &ve-bhyas 
 
 
 
 Inetr. 
 
 asva-is 
 
 
 
 
 (Vedio^ve-bhis) 
 
 {ee6-<piv) 
 
 
Ti.] Noun Inflection. 
 
 II (6). Vowel Stems (feminine a-). 
 
 141 
 
 
 Sanskrit. 
 
 Greek. 
 
 Latin. 
 
 Stem 
 
 asva 
 
 X&pa- 
 
 equor 
 
 Singular : — 
 
 
 
 
 Nom. 
 
 &sw% 
 
 Xiipa 
 
 eqna 
 
 Accus. 
 
 ^vS-m 
 
 X^pa-v 
 
 equa-m 
 
 Gen. 
 
 &va-y-as 
 
 Xiipa-s 
 
 (equa-is, equas) 
 equai, equae 
 
 Abl. 
 
 
 
 equa(d) 
 
 Loo. 
 
 ^va-y-am 
 
 Xa/M^ 
 
 (Eomae= Eoma-i ?) 
 
 Dat. 
 
 ^va-y-ai 
 
 
 equal = equa-ai 
 
 
 Ved.^sva-i (a-ai) 
 
 Xiipa {X^pa-at) 
 
 equae 
 
 Instr. 
 
 &va-y-a 
 
 
 
 Dual :— 
 
 Ved. ^sva 
 
 Piri-<pt 
 
 
 Nom. Aco. 
 
 ^ve 
 
 Xiipa. 
 
 
 Gren, Loo. 
 
 asva-y-5s 
 
 
 
 Dat. Abl. \ 
 Instr. J 
 
 Jisva-bhyara 
 
 X^pa-iv 
 
 
 Plural:— 
 
 
 
 
 Nom. 
 
 iswans 
 
 xSipai 
 
 (equa-es, equas ; 
 
 
 (Ved. a«va-sas) 
 
 
 equals, equai) 
 equae 
 
 Aco. 
 
 isvar-a 
 
 Xiipas (-av-s) 
 
 equas (-am-s) 
 
 Gen. 
 
 ikva-n-am 
 
 
 
 
 Ved. &yam 
 
 XQipaiv (a-ojv) 
 
 equa-rum 
 
 Loo. 
 
 disva-su 
 
 Xiipcu-ai, xcupar-s 
 
 (equa-ls) equis 
 
 Dat. Abl. 
 
 ilsva-bhyas 
 
 
 equa-bus 
 
 Instr. 
 
 ^va-bhis 
 
 {.<t>iv) 
 
 
CHAPTEE VII. 
 
 Inflection op Pronouns. 
 
 The Pronouns exhibit certain irregularities of inflection, 
 
 which make it necessary to consider them separately from 
 
 nouns. In many cases they have undergone such changes that 
 
 the forms admit of only conjectural explanation : and the variety 
 
 of pronominal roots employed makes it difficult, if not impossible, 
 
 to reduce them to any uniform scheme. The Pronouns of the 
 
 ist and 2nd person, and the reflexive pronoun (Indo-European 
 
 ma-, tva-, sva-) have no distinction of gender : a fact which is 
 
 accounted for by their antiquity, if (as appears likely) they are 
 
 the oldest extant elements in language, developed previously to 
 
 the introduction of distinction of gender. The presence or 
 
 absence of this distinction divides the pronouns roughly into two 
 
 main heads, viz. (i) Pronouns without Gender (as above), and 
 
 (2) Pronouns with Gender. 
 
 Pronouns (i) Pronouns without Gender (ist and 2nd Personal, and 
 without -D \ ■ \ 
 Gender. Kenexive). 
 
 The original of these three pronouns, ma, tva, sva, are trace- 
 able in the oblique cases, and in Verb Inflections of person {-mi, 
 -si, -ii, see below, ch. viii); but all speculation as to the deriva- 
 tion of meaning of these elements is fruitless. The declension 
 of these three pronouns has many points of similarity, and they 
 might without difficulty be considered together : but it seems 
 best upon the whole to take them separately. 
 
 1st Personal Pronoun {ma), 
 
 Nom. Sing. Here we are met at once by a different form ; 
 viz. Sanskrit ah&m, Greek iyitv (Doric), eyavya,,eya> (Attic), iaiv, 
 
Inflection of Pronouns. 143 
 
 lavya, laya, lavei (Boeotian), iyavrj (Laconian and Tarentine) ; 1st Personal 
 Latin egd, later ego^- Ah&m and eyav probably arise from 
 a common form agJiam, 
 
 Accus. Sing. Sanskrit ma-m, ma ; Greek yA or ink (e ' pros- 
 thetic' or auxiliary, see p. 83), ifixi Doric; Latin me. Quin- 
 tilian (i. 5. 20) speaks of mehe : and in Old Latin med, ted, sed 
 occur, probably formed on analogy of the ablative in -d. The 
 quantity of me as compared with /ic is variously explained as 
 arising from confusion with the ablat. me (Corssen ''), as a con- 
 sequence of its being monosyllable (Schleicher, § 265), or as 
 a compensation for the loss of -m, i.e. ?rae=me-m, mi-m (stem 
 mi- as in mi-hi). 
 
 Gen. Sing. Sanskrit m.&ma (stem reduplicated, case ending 
 lost); Greek e^"" (Epic)=ej[i6-<7yo (as -ou> of nouns, p. 119), 
 eiieia (Doric), then by loss of i (_/, y) e/i€o, and by contraction 
 ifiov, iiov (Attic), e/ieC, fuv (Doric). efi.i-8ev (Homer and Eurip. 
 Hel. 177) is formed by the suffix -BiV. so iiiBev quoted by 
 Ahrens from Sophron (circ. 450 B.C.) The fornis f/^c'oj, eVoCy, 
 f/xEuj (Doric), iiielas, ifms (Syracusan), are usually explained 
 as addition of gen. sign -t to the old genitive. In Latin mei 
 is probably a locative, or borrowed from the possessive meus. 
 An old genitive mis is said to have been used by Ennius. 
 
 Ablat. Sing. Sanskrit ma-t, Latin me-d (as ie-d, se-d), a form 
 restored by Eitschl to many passages in Plautus, e.g. Trin. 258, 
 1080; Amph. 812; Most. 365. 
 
 Locat. Sing. Sanskrit mayi ; Greek dat. ^oi (/to- + 1) ; and 
 perhaps Latin gen. sing. msi. 
 
 Dat. Sing. Sanskrit m^-hyam; Greek ifuv (Doric) =e/ie-^i(' 
 ={e)ma-hhyam, cp. p. 129; a form i/iivr] (Tarentine) is also 
 quoted. Latin mi-hei, mihl (afterwards mihi) is perhaps for 
 mi-fei, f representing an original bh, which becomes b in 
 tibei, sihei. 
 
 Tnstr. Sing. No trace in Greek or Latin. 
 
 Nom, Plv/r. The Indo-European stem of this case was perhaps 
 
 ' See Wordsworth, ' Fragments,' etc. Introd. xii. 4. 
 
 •^ 'Kritisohe Beitrage zur Lateinischen Fonnenlehre,' p. 528. 
 
144 Inflection of Pronouns, [chap. 
 
 ist Personal formed by addition of the pronominal element -sma (sa-ma) to 
 
 Pronoun. , •, . . 
 
 the demonstrative stems ma-, a-, va-, i.e. ma-sma-, a-sma, 
 va-sma. The first would account for Lithuanian mes ; the 
 second for (Vedic) Sanskrit asme ; the third for Sanskrit vay^m, 
 Gothic vds, English we. Greek rjfiels^, Sfi/ics (Aeolic), A/ies 
 (Doric), arise from stem asma, asmi^^anju- (by assimilation, 
 p. 74) or rjiib- by loss of s and compensatory lengthening of a 
 to T) (cp. icr-iii, eljiij. 
 
 Latiu nos {ends, Carm. Arval.) seems connected with stem no, 
 which occurs in Greek vaC, Sanskrit dual ndu, and accus. gen. 
 dat. plur. nds. It may be that nos is an accus. used as nom., 
 and originally n8s (Sanskrit nas), but lengthened from analogy 
 of the common accus. plur. in -os (equos). Bopp, however, con- 
 siders that nos- is the stem, found e.g. in nos-ter, and connects 
 both it and Sanskrit nas with sma, whence he derives -Wiet in 
 egomet, etc., and i'fwmo'=ismo {i-smcC). 
 
 Acous. Plur. Sanskrit asinan=asmaii-s ; Greek ifiieas, ayLfic 
 (Aeolic), from same stem as nom. plur. : Latin nos as nom. 
 plur. 
 
 Gen. Plur. Sanskrit asm^kam (an adjective in ace. sing, 
 neut.), nas : Greek afifiiav (Aeolic), ij^iemv (Ionic), rifiSiv, fffieiav 
 (Epic) from stem dufie-, rjfu-; Latin nostrum=.nostro-um, gen. 
 plur. of possessive stem nostra-. NosVro-rum is also found in 
 Plautus. 
 
 Ahl. Plw. Sanskrit asm&-t ; Latin «o-6is (as dat.). 
 
 Loc. Plur. Sanskrit asm^-su ; Greek (Aeolic) ajini-iriv. 
 
 Dat. Plur. Sanskrit asmfi-bhyam or nas ; Greek tjiuv, aixfuv, 
 where iv=i-(j>iv (see above, p. 130); Latin no-bis (=wos-6«s, if 
 nos- be stem). 
 
 Dual. Greek nom. ace. vS>i, koj, wSe (Boeotian), gen. dat. vrnXv, 
 vav, are forms from a stem j/(o-= Sanskrit nau, which is used 
 (without inflection) for nom. gen. and dat. dual. In form this 
 nau is a regular nom. ace. dual from stem na-, as ^.svau from 
 A«va-, 
 
 2nd Personal Pronoun ifva). 
 
 • An Ionic form fiiiUs, sometimeB found in MSS. of Herodotus, seems 
 to have had no existence. 
 
VII.] Inflection of Pronouns. 145 
 
 Nom. Sing. Sanskrit tvam (perhaps =<a + Da + ma) : Greek 2nd Personal 
 
 ... . , „ . „ , , , Pronoun. 
 
 (TV, TV (Doric), ■with Boeotian forms tow or rw-ij, where v pro- 
 bably = -m of tvam ; Latin tu. In tv, tu, the u represents v 
 or F. 
 
 Ace. Sing. Sanskrit tva-m or tva; Greek o-e, re (Doric) 
 {=:(rFe, rFe), nv (Boeotian); Latin te=^tve-m, stem tvi- (for 
 quantity see above on ist pers. pron.) ; TJmbrian has tiom, 
 which either =:tuom, :=tvam, or (Corssen) =:tmo-m from stem 
 tvi- lengthened by a (cp. e-^t^m=i-o-m from stem i-). 
 
 Gen. Sing. Sanskrit t^va; Greek reoio (Epic), =«fo-o-yo= 
 tava-sya (see p. ii8)j <Teio=tva-sya, then o-eo, o-eO (Ionic), 
 and aov ; Doric reos, reovs, revs, tIos, tIws, in which -s is added 
 to the old form (cp. on ist pers. pron.). 
 
 Abl. Sing. Sanskrit tva-t; Latin te (old Latin ted). 
 
 Bat. Sing. Sanskrit tu-bhyam ; Latin ti-hi ; Greek Tetv 
 (Epic). 
 
 Nom. Plur. The Sanskrit forms yushme (Vedic nom.) and 
 yushma-n (accus.) point to tva-sma as the Indo-European form. 
 The Greek and Latin forms {v/iels, li/iiies, ifie, vos) are parallel to' 
 those of ist pers. pron., and so throughout the plural of 2nd 
 pers. 
 
 Dual. The Greek forms o-</)Si, o-</)a> (nom. ace), and a(pS>iv, 
 <r^au (gen. dat.), retain in </> the v sound of tva: (T<{>=tv. 
 Latin has retained the v in tui and the possessive tuus 
 (=toass). 
 
 Reflexive Pronoun {sva). 
 
 The stem sva- appears in Sanskrit only in compounds, e.g.Beflexive 
 Bva-yam (self), sva-tas (by oneself), etc.^ : but it is used to form 
 the possessive sva-s= Latin sMits= Greek irFt)!, which appears 
 (by loss of F and change of a-, see p. 66) as oy the possessive 
 pronoun in Homer. There is one distinction of gender in this 
 pronoun, viz. Greek nom. ace. plur. neut. <T^4a. 
 
 Nom. Sing., wanting in Greek and Latin. 
 
 Aceus. Sing. Greek f, Aeolic F€=(rFe, Epic ies=a-eFe (see 
 pp. 66, 68). <rF€=sva, with inflection lost. The forms julw 
 (Epic), vlv (Doric), are perhaps reduplicated accusatives of stem 
 
 • See Curtius' ' Elueidations,' p. 85. 
 L 
 
146 Inflection of Pronouns. [chap. 
 
 Reflexive i- (i/ut/i) : cp. Latin sese. Latin se^sve-=s'oi-m : Oscan siom 
 either=S'ua-m or svi-o-m (see above on and pers.). Old Latin 
 sed ; see on mei, ted accus. 
 
 Gen. Sing, eib (Epic)=o-fe-(ryo, eb (Ionic), o5 (Attic), eS, iov 
 (Doric), eoOs (Boeotian), (see above on ifiovs, rcoSy). In all these 
 forms the aspirate =0- (p. 66) and F has disappeared. (?) Latin 
 sui ; see on mei, tui, p. 143. 
 
 Ahl. Sing, se, original ied=sei-d from stem si {in si-bi) 
 ^ sui. 
 
 Loc^ Sing, of, Aeolic f 01= o-fo-Z; Latin smi. 
 
 Dat. Sing. Boeotian Hv, Doric tv (cii')=€-0iV; Latin si-bi, 
 Oscan si-fei, Umbrian si-be and se-so (Tab. Eugub.^), which is 
 perhaps a reduplicated locative =se-io-i(?), or a genitive form 
 ^sva-sya .• but these are only conjectures to explain a very 
 obscure form. 
 
 In the plural Greek retains the stem but little altered (o-^t- 
 =.sva), and has a very complete set of forms; while in Latin 
 the plural forms are identical with the singular. Thus in 
 Greek, — 
 
 Nam. Plwr. tr^6is=o-^t-es. 
 
 Accus. Pl/wr. (r^as, cr^i-as (Ionic), a-c^d-as (stem raised), Doric 
 <r(pi (inflection lost), Aeolic S-o-^e (a ' prosthetic' or ' auxiliary,' 
 see above, p. 83). 
 
 Gen. Plur. a-cpav, (r(j>4-av (Ionic), a-<peiav (Aeolic). 
 
 Loc. Plur. a-<pi-in. 
 
 In Greek Dual the stem a^a- = sva. <r^S-e then has the 
 dual nom. ace. inflection as well as the stem vowel lengthened ; 
 ir(j>aiv (gen. dat.) = o-(j>S>-(jiiv. 
 
 ' Wordsworth's ' Fragments,' Introd. xii. 9. 
 
Til.] Inflection of Pronouns. 147 
 
 I. Peonominal Declension without Gender. 
 (i.) Pronoun of the 1st Person ima-") : — 
 
 
 Sanskrit. 
 
 Greek. 
 
 Latin. 
 
 Singular ; — 
 
 
 
 
 Nom. 
 
 aMm 
 
 i^(hv, kyiij 
 
 ego 
 
 Accus. 
 
 ma-m, ma 
 
 i-f^i, l^i 
 
 me 
 
 Gen. 
 
 m4ma 
 
 efjtov-s 
 
 (mei ?) 
 
 Abl. 
 
 mart 
 
 
 me-d 
 
 Loc. 
 
 m4-yi 
 
 IjlO-X, iio-l 
 
 mei (unless gen.) 
 
 Dat. 
 
 ma-hyam 
 
 kfdv (ifit-^tv) 
 
 mi-hei, mihi 
 
 Instr. 
 
 m4-ya 
 
 
 
 Dual :— 
 
 
 
 
 Nom. 
 Aoo. 
 
 avam I 
 avani, nau J 
 
 vSii, via 
 
 
 Gen. Loo. 
 
 av4-yoB 
 
 
 
 Dat. Abl. 1 
 Instr. J 
 
 a-va-bhyam 
 
 vSi-iv, v^v 
 
 
 (nau) 
 
 
 
 Plural:— 
 
 
 
 
 Nom. 
 
 vay4m 
 
 ctft/jLes (affiu-) 
 
 nos (? nos) 
 
 
 a8me(Ved.). 
 
 
 (enos, Carm. Arv.) 
 
 Aoo. 
 
 asm an 
 
 a/i/xe 
 
 
 
 nas 
 
 ■qjiias, ■fifjias 
 
 nos 
 
 Gen. 
 
 asm5k-am (adj.) 
 
 a^iLjiioiv 
 
 nostrum (nostro-um 
 
 
 nas 
 
 ■qiui-tiiv, ^/ii-uy 
 
 nostri 
 
 Abl. 
 
 asmfC-t 
 
 
 no-bis (dat.) 
 
 Loc. 
 
 asmi-su 
 
 dii/ie-aiv 
 
 
 Dat. 
 
 asma-bhyam 
 
 ajijuv (i-<piv) 
 
 no-bis- 
 
 
 nas 
 
 ^lUV 
 
 
 Instr. 
 
 asma-bhis 
 
 
 
 L 2 
 
J 48 Inflection of Pronouns. 
 
 (ii.) Pronoun of the 2nd. Person (tva-) : — 
 
 [chap. 
 
 
 Sanskrit. 
 
 Greek. 
 
 Latin. 
 
 Singular : — 
 
 
 
 
 Nom. 
 
 tva-m 
 
 t{f, av 
 
 tu 
 
 Accus. 
 
 tvii-m, tva 
 
 Tj,<rJ(=Tfe) 
 
 te=tve-m 
 
 Gen. 
 
 t&va 
 
 alio, ceo 
 
 GOV, ffiV 
 
 TeoS-t 
 
 (tui?) 
 
 Abl. 
 
 tva-t 
 
 
 te-d (=tei-d) 
 
 Loc. 
 
 tv4-yi 
 
 ffo-l (tva-1) 
 
 tui (gen.) . 
 
 Dat. 
 
 tii-bhyam 
 
 reiv (i-^iy) 
 
 ti-bei, tibi 
 
 Instr. 
 
 tva-ya 
 
 
 
 Dual :— 
 
 
 
 
 Nom. 
 
 yuvim "1 
 yuvilm, v5m J 
 
 
 
 Ace. 
 
 ff^S-i, Oipii 
 
 
 Gen. Loo. 
 
 yuva-yo3 
 
 
 
 Dat. bl. 1 
 Instr. J 
 
 yuv4-bhyam 
 
 a<pw-iv (^-i-ipiv) 
 
 
 vam 
 
 <T<p^v 
 
 
 Plural:— 
 
 
 
 
 Nom. 
 
 yu-y4m 
 
 vn/ifs 
 
 vos 
 
 
 yushme (Ved.) 
 
 i/iies, b/juTs 
 
 
 Ace. 
 
 yushma-n 
 
 V/ilie 
 
 V03 
 
 
 vas 
 
 vfiiaSf ifxiis 
 
 
 Gen. 
 
 (yuBhm5k:a-m,adi.) 
 
 v/i/xeaiv 
 
 vostrum 
 
 
 vas 
 
 {ifxeiojVf v/ieotv 
 
 vostri 
 
 Abl. 
 
 yushm4-t 
 
 
 vo-bis (dat.) 
 
 Loc. 
 
 yushma-su ■ 
 
 
 
 Dat. 
 
 -yushma-bhyara 
 
 iHHiv (i-(fic) 
 
 vo-bis 
 
 
 vas 
 
 VjUV 
 
 
 Instr. 
 
 yushmd-bhis 
 
 
 
T^ii.] Inflection of Pronouns. 
 
 (iii.) Refleodve Pronoun (sva-) : — 
 
 149 
 
 
 Greek. 
 
 Latin. 
 
 Singular : — ■ 
 
 
 
 Nom. 
 
 
 
 Accus. 
 
 k, pe, l€ = ffcfe 
 
 se ( = svi-m) - 
 
 
 /uy, yiv 
 
 siom (Oac.) 
 
 Gen. 
 
 €0, oS, cS, lot) 
 loO-s 
 
 (suil) 
 
 AM. 
 
 
 ae-d, se 
 
 Loc. 
 
 lor, of 
 
 sul 
 
 Dat. 
 
 liv, tv = €-(piv 
 
 sibei, sibi 
 Oac. si-fei 
 
 Dual :— 
 
 
 
 Nom. Ago. 
 
 atpoi-k 
 
 
 Dat. Abl. Instr. 
 
 a<pai-iv 
 
 
 Plural:— 
 
 
 
 Nom. 
 Accus. 
 
 ^ , ^ ^ , ' \. (T(bi-tt (neut.) 
 
 -| 
 
 
 d-a(j>e 
 
 ' as in sing. 
 
 Gen. 
 
 aipicev, ffipeioiv 
 
 
 
 ff^ajv 
 
 ^ 
 
 Loc. 
 
 ij<l>i-(ii 
 
 
 Dat. 
 
 a(pi{y) (i-ijiiv) 
 
 as in sing. 
 
 (2) Pronouns with Grender. 
 
 The declension of these is rather complex, especially in Latin,, Pronouns 
 where a great variety of pronominal stems is found ; and an ^ " ™' 
 examination of all their forms belongs to the special grammar 
 of each language. The following tables give the declension of 
 the Indo-European demonstrative stem to- in Sanskrit, Greek, 
 and Latin; which, compared with the declension of nouns already 
 given (above, ch. vi.), will serve for the general illustration of 
 this class of pronouns. 
 
T50 Inflection of Pronouns. 
 
 Pronominal Stem to-. 
 I. Masc. and Neut. : — 
 
 [chap. 
 
 
 Sanskrit. 
 
 Greek. 
 
 Latin. 
 
 Stem 
 
 ta- 
 
 TO- 
 
 is-to- (i + aa + ta) 
 
 Singular : — 
 
 
 
 
 Nom. 
 
 sa, ta-d 
 
 (5(s), Tb 
 
 istu-s, is-te, istu-d 
 
 Aoc. 
 
 ta-m, ta-d 
 
 rb-v, TO 
 
 istu-m, iatu-d' 
 
 Gen. 
 
 tii-sya 
 
 TQ~tOf TOV 
 
 istiua (isto-i-os) 
 
 Abl. 
 
 td-sma-t 
 
 [Ta)S = TI»-T] 
 
 isto-d 
 
 Loo. 
 
 t4-sm-in 
 
 [or=d..] 1 
 
 T$ = TO-Ol 
 
 . . J =isto-i (loc.) as humi, quoi 
 ' l=iato-ei (dat.) as quoiei 
 
 Dat. 
 
 tSrsmSi 
 
 Instr. 
 
 ten-a 
 
 
 
 Dual :— 
 
 
 
 
 Nom. Aoo. 
 
 t3u (tft), tS 
 
 rii 
 
 
 Dat. Abl. 
 
 ta-bhySm 
 
 TO-Tv 
 
 
 Gen. Loo. 
 
 ta-y-6s 
 
 
 
 Plural:— 
 
 
 
 
 Nom. 
 
 te, ta-n-i 
 
 Tol, ot, Tcl 
 
 isti, ista, ista-e-c (eis, his, qnes) 
 
 Aoo. 
 
 ta-n, ta-n-i 
 
 tJvs Toiis, rd 
 
 isto-a, lata 
 
 Gen. 
 
 te-sham 
 
 rSiv 
 
 iato-rum 
 
 Loo. 
 
 te-Bhu 
 
 ToT-ai, Tois 
 
 istia (queis) 
 
 Dat. Abl. 
 
 te-bhyaa 
 
 
 (qui-bua, hi-bua, hoi-bus) 
 
 Instr. 
 
 tais 
 
 
 
 11. Feminine ; 
 
 
 Sanskrit. 
 
 Greek. 
 
 Latin. 
 
 Stem 
 
 ta- 
 
 TO- 
 
 
 ta-is-ta 
 
 Singular : — 
 
 
 
 
 
 Nom. 
 
 sa 
 
 4 
 
 
 ia-ta, qua-i (quae) 
 
 Aoo. 
 
 ta-m 
 
 rfi-v 
 
 
 is-ta-m 
 
 Gen. 
 
 td-sy-as 
 
 Tij; 
 
 
 is-tius 
 
 Abl. 
 
 
 
 
 is-ta(-d) 
 
 Loc. 
 Dat. 
 
 t4-ay-5m 
 t&-ay-ai 
 
 Tji 
 
 } 
 
 is-ti (as above) 
 
 Instr. 
 
 tayS 
 
 (.m 
 
 
 
vir.] 
 
 Inflection of Pronotms. 
 
 151 
 
 
 Sanskrit. 
 
 Greek. 
 
 liatin. 
 
 Dual :— 
 
 
 
 
 Nom. Aco. 
 
 te 
 
 ra 
 
 
 Dat. Abl. 
 
 t^-bhyam 
 
 ra-tv 
 
 
 Gen. Loo. 
 
 t4-y-6s 
 
 
 
 Plural:— 
 
 
 
 
 Nom. 
 
 ta-s 
 
 TO? 
 
 is-tae 
 
 Aoc. 
 
 ta-s 
 
 TCi-S 
 
 is-ta-8 
 
 Gen. 
 
 tg-sam 
 
 Td-ay^ TcDc 
 
 is-ta-Tum 
 
 Loc. 
 
 ta-su 
 
 Tp-<ri, Tai-s 
 
 is-tis 
 
 Dat. Abl. 
 
 ta-bhyas 
 
 
 
 Instr. 
 
 ta-bbis 
 
 
 
 N.B. Gen. sing. -ius ) show increase of ( isto-i-us 
 
 Nom. fern, and neut. istaec, quae J stem by i: thus \ ista-i-ce, qua-i. 
 
 Comparing these forms witt those of the nommal declension Pronoum ^^ 
 it will be seen that the nom. sing, termination -s is omitted 
 ii-om Sanskrit sa (masc), being in fact (see p. no) a mere 
 repetition of sa. Thus beside Greek os (Epic as a demonstra- 
 tive pronoun) =:sa-s, we have the later form 6 = sa. 
 
 In Latin the -s is lost in ille, ipse, iste (where final stem 
 vowel sinks to e), qui {=quo-i-s), and hie {=ho-i-ce); but is 
 retained in the old forms ipsus, isttis, alius, and regularly in 
 quis ( = 2m5-s), is, alius, etc. 
 
 The nom. ace. termination of neut. sing, in -d is characteristic Temmation 
 of this class of pronouns. Sanskrit and Latin have retained Sing, 
 the d, which Greek has lost : Zend has d in tad : and Gothic 
 thata, whence our that. Bopp, Schleicher, Curtius and others 
 give tat as the Sanskrit neut. sing. ; Bopp explaining ta-t as 
 =ta + ta, a repetition of the pronominal element (like sas= 
 sa+sa), and illustrating the change to Latin d by the old 
 ablatives gnaivod, etc. ; cp. with Sanskrit abl. in t. But here 
 also (see above, p. 124) -d is probably the earlier form of the 
 termination. And if tat were the true form in Sanskrit, pho- 
 netic analogy would (it is said) require in Latin istut, in Gothic 
 tha-tha. It is probable therefore that Latin -d is the original 
 
152 Inflection of Pronouns. ' [chap. 
 
 Pronoims form, and not a modificatiou of *. Greek neut. sing. nom. 
 ' having thus lost the final t (8) ends in : but in other respects 
 the pronominal is the same as the nominal declension of 0- and 
 a- stems. 
 
 In Latin also the a of stem ta is represented by (masc. and 
 neut.) and a (fem.), quis, is, aU-s being exceptions. But on 
 the other hand there are certain peculiarities which distinguish 
 the Latin pronominal declension more markedly than the Greek 
 from that of noun stems in a-, 0-, e.g. — 
 
 Peculiarities (a.) Gen. sing, formed by addition of -os or -us (Indo- 
 
 of Latin _^ ' ^ ^ •' , , , , 
 
 Pronoun. European -as, see p. ii8)=-isot consonantal declension to the 
 declension. , , ,, -i , . ,1 . . . . ■, . ■, . 
 
 stem lengthened by t : thus isttios=is-fo-i-os, hujus^no-t-os, 
 
 cujiis=: quo-i-os. 
 
 (6.) Locatyve singular in -i used as dat., e.g. is-ti-=is-to-i 
 (see p. 126). The form quoiei, however, is apparently a true 
 dative, from the stem increased by i {quo-i-ei), the locative form 
 quo-i (cut) being the more usual in classical times. 
 
 (c.) It has both locative plural (ts-Jts, see p. 127) and dative 
 plural {qwi-hus). 
 
 (d.) The neuter termination -d (see above). 
 
 (e.) Increase of the stem by i. This i is probably parallel to 
 the Greek suffix 1, found with pronouns especially, but after the 
 case-suffix, e. g. ovtoitX, Toh-\ outoii, etc. Its place in Latin is 
 between the stem and the case-suffix, and it is not carried 
 through all the forms. It occurs always in gen. sing. i-us= 
 o-i-os; often in nom. fem. sing, {qua-i, Jia-i-e, etc.) and in neut. 
 plurals which are similar in form ; in masc. and fem. plurals in 
 -I and -oe-=o-i, a-i, as in nominal declension of 0- and a- stems 
 (p. 1 1 3), and in such dat. forms as quo-i-ei, e-i-ei. 
 
 The declension of Latin pronouns is treated very fully in 
 Wordsworth's 'Fragments and Specimens of -Early Latin' 
 (Introd. ch. xiii. pp. 91-112). I give here the main results 
 of Mr. Wordsworth's discussion, referring the reader to his 
 work for details. 
 
 The chief pronominal stems in Latin are 7io-, to-, so-, co- {quo-), 
 olio-, i- or eo-. These are employed in the formation of pro- 
 nouns in various ways, viz. : — 
 
"fTii.] Inflection of Pronouns, 153 
 
 (1) Simply, as quo-d, ollu-S, is. Formation 
 
 (2) Eeduplicated or compounded, as in is-to, i-p-so, quis-^onotTro- 
 
 „,,,•„ nouns in 
 
 ^^^- Latin. 
 
 (3) Increased by i, as qui (quo-i), eid (see above). 
 
 (4) Increased by i and compounded, as ko-i-ee {Mc), is-to-i-c 
 (istie). 
 
 These are divisible into three classes, according to simplicity 
 of declension : viz. : — 
 
 (i. ) OU0-, isto-, ipso-, alio-, etc. 
 
 (ii.) Eo-, quo-, i- or eo-. 
 
 (lii.) So-, to-, etc., defective and enclitic stems. 
 
 The following peculiar or archaic forms may be noted under 
 each of these classes : — 
 
 (i.) Ollus (=ille) found in Ennius and old inscriptions, and Peculiar or 
 
 STCllSilC 
 
 surviving in olim, which preserves the oldest form with one I. forms. 
 In Lucretius and Virgil it is an intentional archaism. 
 
 IstMs, vpsus are found in Plautus, and even in Terence (Hec. 
 455). The element p in i-p-so is the same which appears (as p 
 or pe) in rea-p-se, quis-p-iam; nem-pe, pro-pe. 
 
 Alls, olid are late contractions of alius, aliud : alis is fem. in 
 CatuU. Ixvi. 28 : olid is frequent in Lucr., who does not use alis. 
 
 The locative is-to4, isti (see above, p. 126) is strictly locative 
 in the adverbs illi-o, isti-c. A few examples of its use as 
 genitive (perhaps from analogy of noun stems in 0-) are given : 
 e.g. Ter. And.- iii. 5. 2, nulli consili; Plaut. Trin. ii. 2. 37, 
 coloris ulli capiendi; ib. v. 38, isii modi. In each of these 
 cases an ordinary genitive in -i is close at hand to suggest the 
 analogy. Terence has alterae and Plautus isiae for dat. fem. 
 
 A locative formation in -im or -in- (cp. Sanskrit ta-sm-in) 
 occurs in the adverbs olim, illin-c, istinc, Mn-o, etc., cp. long-in- 
 quus, prop-in-quus. 
 
 The plural is declined like ordinary 0- stems.. 
 
 (ii.) The stems ho- and quo- are further increased by i ; ho- 
 having generally the enclitic c, or ce appended (a remnant of 
 the pronominal stem co-). The increase takes place in sing, 
 nom. hlc (ho-i-c), gen. huius (ho-i-os), loc. hlc (ho-i-c), and plur. 
 
154 Inflection of Pronouns. [chap. 
 
 Declension nom. hi (Jio-i), ha-i ; dat. and abl. hihus (ho-i-bus, for ex- 
 of Pronouns , ^.^'', , . ^ ' 
 
 in Latin: amples see Wordsworth, p. iC?). 
 peculiar or / i ■ \ i -i 
 
 archaic ine Stem quo- (relative), when used as an interrogative pro- 
 
 noun, has a special inflection for nom. with the case suffix s, the 
 stem vowel o being weakened to i {qui-s, qui-d). The same 
 form is used indefinitely in ali-quis, si quis, ne quia ; but then 
 nom. fem. sing, and neut. plur. are not increased by i : e. g. 
 aliqua, si qua. 
 
 In the declension of ts, the stem i is sometimes increased by i 
 to ei (nom. e-i-s, gen. e-i-us, dat. e4-ei, nom. plur. i-i, e-i-s), 
 sometimes turned into an o- or a- stem {eo-m, ea-m, etc.). 
 
 The following peculiar forms may be noticed : 
 
 Nom. Sing. : ei'S (raised form of I'-s) is found on some 
 inscriptions; and perhaps adeo represents a raised form of 
 neut. id (ad, eod) : but this is at best doubtful. 
 
 ffic {ho-i-o) is sometimes shortened to Mc (Aen. iv. 22), but 
 not often : hoc {hod-ce) never. 
 
 Aec. Sing. : im, em in quotations from old laws^ point to 
 a time before the stem i was raised to eo-. 
 
 Htic (adv.) is originally Aoc, ' to this place,' as in Aen. viii. 423, 
 and Cicero Epp. 
 
 Hone (Ep. Scip. Appendix I. i. 2 ) and quo-m (S. C. de Bacch.) 
 are old Latin forms. Quom is the adverb quum or cum.; cp. 
 quon-dam : and quaan, quanquam,, quod are all adverbial ac- 
 cusatives. (See Appendix II. A.) 
 
 Gen. Sing. : the suffixed i generally becomes consonantal ; 
 and in old poets Ifmdus, cuius, eius are often monosyllables. 
 
 Local. Sing. : heio or hie (adv.) is locative =ho-i-c, and so 
 perhaps are qui, qui-ne, and qui-jppe (sometimes explained as 
 ablat.). The form quo-i is found in Plautus^ in the phrase 
 quoimodi, apparently gen. (cp. istimodi above, p. 153) ; and cui- 
 modi or cuicuimodi are found in Cicero '. 
 
 ' E.g. XII. Tab. i. i, 'Si in jua vocat, ito; ni it, antestamino ; igitur 
 em capitoj' and viii. 12, 'Si nox furtum faxsit, si im oooisit, jure caesus 
 eato.' The existing remains of the XII Tables are given in Wordsworth's 
 ' Fragments,' pp. 254-265. 
 
 ' For references see Wordsworth, Introd. xiii. 30 (p. 103). 
 
 ' Pro Rose. Araer. 95, ' Vereor enim emcuimodi es;' Att. iii. 22 ad fin., 
 ' cmommodi agam.' 
 
vii.] Inflection of Pronovms. 155 
 
 Dat. Sing. : ei-ei on inscriptions ; ei-i or e-i in Plautus, and Declension 
 
 T •• y / -1 ■ ,-,. ^ ... ... ofPrononns 
 
 Jjucr. II. 1 130 (cibus omms diditwr ei) : quot-ei on inscriptions, in Latin. 
 
 iVom. Flur. : the forms eis, hisee, ques have been alluded to 
 under the noun declension (p. 114) : for examples see Words- 
 worth, Fragments, Introduction ix. 9 and xiii. 34. In the 
 feminine haec (for the more usual hoe) is not uncommon : e. g. 
 Virg. G. iii. 305, Aen. vii. 175 (Eom. MS.), 852 (Pal.); 
 Lucr. iii. 601 and vi. 456; Catullus, Ixiv. 320. 
 
 The adverbial iorms postilld, propiered, postea, anted; posthdc 
 antidhdc, qudpropter ; and the numerals trigintd, etc., show 
 an earlier neut. plural of the pronouns in a without increase by 
 i, as in. ha-i-c (haec) ; and are evidence for the original quantity 
 of neut. plur. d, seen in Sanskrit, and traceable in Latin poetry : 
 e.g. Virg. Aen. iii. 464 gravid, Ter. Ad. 612 debilid. (See 
 Corssen, vol. ii. p. 460; Wordsworth, Introduction ix. 10, and 
 xiii. 35 ; and cp. Wagner, Introd. to Terence, p. 14.) A form 
 ead in S. C. de Bacch. is disputed, Bopp thinking it accus., 
 Eitschl ablat. 
 
 Log. Plur. : eis, quels or quts, heis or Ms are all in common 
 use as dative and abl. as well as the dative forms in -bus, which 
 are more usual with quo-. Examples of hibus=ho-i-bus, i-bus 
 and l-bus are given by Wordsworth, p. .107. 
 
 (iii.) The defective stems so-, to- and others are chiefly trace- Defective 
 
 ^ ' ' ■' stems. 
 
 able in adverbial forms. 
 
 So- (Sapskrit sa-, Greek o, English slie, German sie) is seen in 
 i-p-so, and the old accusatives sum, sam, sos, sas in Ennius and 
 the XII Tables (Wordsworth, p. 108). Sic {sei-ce) and si (set) 
 may be locatives of the same stem: but the analogy of Oscan 
 savi and Umbrian sve (=: Latin si) rather points to the pro- 
 nominal element sva. Sei, si (Italian se) is the same word as 
 si-c, originally a pronominal adverb =' there,' 'in that way,' ' in 
 case that,' and so 'thus' and 'if;' cp. the use of so='if' in 
 English, e. g. Tennyson's ' Guinevere :' — 
 
 ' It may be, so thou purify thyself, 
 
 And ^0 thou lean on our fair father Christ, 
 
 Hereafter in the land where all is pure 
 
 We two may meet.' 
 
15^ Inflection of Pronouns, [chap. 
 
 l^nonnsin Ta-, to- (Greek ro- in Q.V-T6-S, ov'To-s and oblique cases of 
 
 Defective article : declined throushout in is-to, see the Table on p. i Ko) 
 
 stems. . . -Ill r u / 
 
 survive m a number of adverbs : e. g. ta-m, ta-ntus, etc. 
 
 (retaining the vowel a) ; turn, tun-c, i-tem, au-tem {a sunk to u 
 
 and e). U-ti (w-td, ut) is a locative from this stem; and att-f, 
 
 a-t, e-t are perhaps locatives similarly shortened. I-ta (so in 
 
 Naevius' Epitaph, 'Itdqvs p6stquam'est Orci trdditfls thesaliro') 
 
 is ablative=' this wise :' so alm-ta, ' otherwise.' 
 
 From da-, do-, a similar stem to ta-, appear to be formed 
 numerous adverbs and terminations, e. g. -dam, -do, -dwm, -dem, 
 -de, as in quon-dam, quan^do, do-nec, acfe-dum, etc. ; tan-dem, 
 qui-diem, etc. ; in-de, un-de, etc. Into all these forms the idea of 
 time enters (not necessarily dv/ration of time as distinct from 
 point of time in turn, etc.; for e.g. -(ZM»i='now' in age-dwm, 
 etc., as well as ' while ') * : hence it is possible that they may 
 all be referred to the root di'e-, the origin of dies, de-us and 
 many Indo-European words for the conception of 'brightness' 
 or ' day ' leading to that of ' God ' ^, in Sanskrit Dyaus-pitar 
 (J)ies-piter, ' sky-father '), whence Zevs, Jup-pifer, Dius Fidius 
 (Zeiry marios), etc. ; inter-dius, inter-diu ; prope-diem, pri-die, 
 etc. There is certainly in these latter words and the adverbs 
 of which we are speaking a close parallel between the noun 
 stem div- and the supposed pronominal stem da-, do- : and the 
 two stems are identified by Corssen. Other philologists, how- 
 ever, regard the identification as improbable ; and it certainly 
 cannot be taken for granted. 
 
 De (prep.) is abl. from stem da-, like se-d, se, from sa. la-m is 
 explained by Corssen (i. p. 213) as =^dia-m,, 'this day' (die-m), 
 like Greek 6ij, ^'8?;, Srjv. Others refer it to a stemjV, ya- (? Ger- 
 man _/«; our yea). In quis-p-iam its temporal sense is lost : but 
 et-iam, quoniam retain it in their original use. 
 
 A stem na- {no-) or an- is supposed to account for na-m, quis- 
 
 • On the various uses of 'dwm, see- Ramsay's ' Mbstellaria,' Excursus, 
 ii. p. 184. 
 
 " On these words, see especially Max Muller's ' Lectures,' Series II. 
 Lect. X. pp. 425-461, 1st ed.; and Peile, ' Introduction,'' ch. v. p. 122 
 (3rd ed.). 
 
VII.] Inflection of Pronotms. 157 
 
 na/m, etc. The full form is found in these and various weakened Pronouns in 
 forms in nu-m, nun-c, vvv ; nem-pe, e-nim, and Greek vh, which Defective 
 last is of course pronominal (see above, p. 145). These are all 
 accus. forms : nae (na-i), ne, Greek va-l, are locative. 
 
 The stem an- is preserved in Greek ava, hv, Latin an and in, 
 endo. Its force is ' that,' ' the other.' Sanskrit has ana and 
 lithuanian ana-s^iMe. Curtius compares av-a, and Latin an- 
 hdare, 'to draw up breath.' The negative prefixes av-, a-, Latin in, 
 Sanskrit an-, a-, German and English un-, are perhaps connected 
 with the same stem.; hv, an in hypothetical sentences and 
 questions are also akin. In, endo, Greek iv-X and ds=^iv-s or 
 iv-\-s are local in meaning : the two uses of in with ace. and 
 abl. being parallel to the two Greek forms. 
 
 The enclitic terminations ce or que in hi-o, ne-e, ne-que, at-que, 
 etc. must arise from a stem co-, perhaps a variety of quo- { = 
 ' who,' ' which,' ' any '), with the demonstrative meaning ' there : ' 
 -pe in i-p-se, quis-p-iam, rea-p-se, etc., and in qui-ppe, nem-pe, 
 pro-jpe is possibly a dialectic variety of ee ; Oscan and Umbriaa 
 substituting p for k (by ' Labialism,' see chap. iv. p. 50). 
 
CHAPTEE VIII. 
 
 Vehb Inflection. 
 
 Distinction The Verb (prjiia, verhum, the ' word ' par exceUence) in Greek 
 
 between 
 
 Verb and or Latin exhibits a much greater variety of inflection than the 
 
 Noun. Time, mood, person, number, and voice are all ex- 
 pressed, and in some forms all at once, by inflectional additions 
 to the root or simplest form expressing the idea ; this verbal 
 root being in no way difierent from a nominal root, so far as 
 any power of expressing action etc. is concerned. Verbs are 
 thus only nouns with a pronominal affix. The abstract idea 
 of e. g. action, motion, sensation, etc., can be expressed equally 
 by a nominal or a verbal root ; but when expressed by a verbal 
 root it is further brought into relation as a verb with other 
 words in a sentence, (i) by Person endings, attaching it to 
 a definite subject or subjects (the distinction of Number being 
 expressed, as in Noun Inflection); (2) by Modal elements, 
 defining the aspect under which the action is regarded, as 
 a fact or a supposition; (3) by Tense elements, ascribing it to 
 a particular relation in time. 
 Verb forms A verb form, then, is distinguished from a noun form mainly 
 piei. by the greater number of different elements combined in it. 
 
 In any case-form of a noun we find one invariable element, the 
 stem, and one variable element, the case-ending, as in iraih-hs, 
 jrat8-(, naiS-ae, etc. ; or at most a vowel besides, inserted to 
 connect the case-ending with the stem, as in TraiS-e-o-o-t. But 
 
Verb Inflection. 159 
 
 there are very few verb forms of so simple a character. In 
 e.g. l-fiev we have only stem and inflection; in ay-o-ficv, stem, 
 connecting vowel, and inflection ; but in iwfiev, ayoire, we have 
 an additional element denoting moody in dydyoiTe a further 
 addition (to the verb stem) denoting time; in ijyayov another 
 addition, viz. the augment. And a comparison of any of the 
 more complex verb forms, in a ' synthetic ' language like Greek 
 or Latin, with their equivalents in an ' analytic' language like 
 our own, is the best illustration of the general difference 
 between an analytic and a synthetic language, and the specially 
 complicated nature of the verb forms in the latter. The analy- 
 sis of two such forms as i<f>iKrj(ravTo and regeremus shows (a) in 
 
 i-<i)CK{]-a-a-vT-o, 'they loved themselves,' i. augment; 2. verb- 
 stem ; 3. addition to form tense-stem ; 4. ' thematic ' vowel 
 added to the stem before inflection ; 5. reUc of a pronoun, 
 denoting 3rd per. plur. ; 6. voice-letter, indicating middle or 
 
 passive: (0) m reg-er-e-iwa-s, i. stem; 2. sign of past time; 
 3. sign of mood; 4. sign of ist personal pronoun; 5. plural 
 Bign, denoting others with the person of speaking. If s be 
 changed to r (regeremur) it denotes that the speaker and others 
 with him are passive instead of active. 
 
 The example last given {reg-er-e-mus) shows the order in Order of 
 which verb inflections are usually attached to the stem — viz. tions. 
 tense, mood, person, number, and voice. Modifications for tense 
 and mood come between the stem denoting the idea of action, 
 and the personal pronoun denoting the agent; inflections of 
 number and voice, which affect the position of the agent by 
 showing him to be either one of a number or passive instead 
 of active, are appended after the personal pronoun. 
 
 The analysis of verb forms is thus more complicated than 
 that of noun forms from the greater number of elements to 
 be distinguished. There is also a further difference, that whereas Further dif- 
 in explaining the different forms of every single noun we can tween Verb 
 begin with one invariable element, i. e. with one stem only, in Tense- 
 the explanation of verb forms we must assume several fixed 
 points, from each of which we start as from a separate stem 
 
l6o Verb Infieetion. [chap. 
 
 Verb Mec- in the explanation of certain groups of forms. These special 
 stems. stems, or subordinate centres of classification formed from the 
 verb-stem common to all, are known as Tense Stems; the 
 elements of inflection by which time is denoted being of a less 
 separable and general character than those of person or mood, 
 and being in fact suffixes for the formation of subordinate stems, 
 each of which is the permanent element or stem in a series of 
 verb forms. Thus in a Latin verb the perfect and supine often 
 show a different stem from that which appears in the present 
 tense : and in a Greek verb such elements as ayay- in dydy-anev, 
 
 dydy-oiTe, dyay-(iv, dyay-ecrBai ; d^- in a^ofiev, S^oi/iev, a^eiv, etc. ; 
 \va-a- in eKvva, \viTa-i-ii€v, \v(ra-s, \v<Ta-(r6ai ; or XeXu- in XeXu-xa, 
 XeXu-jaat, c-XcXv-ro, have each a comparative permanence among 
 a series of verb forms, and must be regarded as stems, though 
 not the verb-stem, in each case. All scientific analysis, there- 
 fore, of verb forms in Greek and Latin must take into account 
 these special stems formed from the verb-stem (e. g. dyay- and 
 d|- from dy- the stem of aya ; Xvo-a- and XfXv- from Xv- the stem 
 of Xuw j Txmr- pres., TfruTT- perf., and rui/ra- ,aor., from tutt- the 
 stem of Ttin-To), appearing in e-Tim-ov ^) : and thus the question 
 of stem, formation must accompany that of inflection in the case 
 of verbs to a much greater degree than in the case of nouns. 
 ' To state the case briefly, it may be said that in the noun, 
 formation — ^that is, formation of the word, or more correctly 
 of the stem — and inflection in the narrower sense are distinct ; 
 but in the verb they combine, and encroach each upon the 
 other. He alone is completely master of the verb forms who 
 from the verb-stem common to all can first form all the special 
 stems, and secondly can inflect the stems when correctlyformed ".' 
 It is not however necessary to follow the rigidly scientific order of 
 (i) formation, (2) inflection, or learn always to forin uninflected 
 stems before we know how to inflect them when formed : but 
 the formation of stems must in the case of verbs be discussed 
 
 • Tt)ir- is here called the ' pure verbal stem,' as distinguished from tuttt-, 
 the 'present stem,' Trnfia- the weak aorist stem, etc. The 'strong' or 2nd 
 aorist usually exhibits the pure verbal stem. 
 
 * CurtiuB, 'Elucidations, p. 93. 
 
VIII.] Verb Inflection. i6i 
 
 at lea&i pari passu with their inflection. It is not enougii to 
 
 understand how to inflect tutttco, -eus, -ft, rirv^a, -as, -e, eTvyjra, 
 
 -as, -e, etc., without understanding how these different forms are 
 
 connected together in one verb. 
 
 In discussing the elements of Verb Inflection, it is usual to Elements 
 
 of Verb 
 take them in the reverse order of their attachment to the verb- inflection: 
 
 stem, beginning from the end of the word with (i) the most 
 universal and characteristic inflections, the person-endings, with 
 their forms for active and middle (passive) voice and their in- 
 flections oi number ; (2) signs oimood/ (3) <ew««-inflections, 
 i. e. the formation of ' tense-stems.' 
 
 I. Person-endvngs, i.e. pronominal sufiixes of ist, 2nd, and Person Bnd- 
 3rd person in singular, dual, and plural number. There are 
 
 two series for (i) active, (2) passive 'voice ;' and in each series 
 there appear a fuller and a weaker form, which are distinguished 
 as (a) primwry {-mi, -si, -ti, etc., used with present, future, 
 and perfect tenses), (6) secondary (-?», Greek -v, -s, -t, used with 
 augmented tenses of Indicative), e. g. — 
 
 Primary, riBrj-jxi, nnrTo-nai, TvTtTo-vTcu.. 
 
 Secondary, i-T'iBrj-v, i-TV7rr6-iJLr]v, c-rvTrTO-mo ^. 
 
 II. Mood Signs, to distinguish the Conjunctive and Optative Mood signs. 
 ' Moods ' from the Indicative. Of the other so-called ' Moods,' 
 
 the ' Infinitive ' is a verbal noun, while the ' Imperative ' is 
 distinguished from the indicative by a weaker form of person- 
 ending, standing in the relation not of nominative but of 
 vocative, and is thus a sort of verbal interjection. 
 
 [The force of Moods is a question of Syntax : but it may be 
 noted here that there can be, strictly speaking, only two 5 moods' 
 (jnodi actionis, modes or aspects under which the action is 
 regarded), viz. (i) direct assertion that it is taking, has taken, 
 or wUl take place — 'Indicative' Mood: (2) the idea or suppo- 
 sition of its taking place in past, present, or future time — ' Sub- 
 junctive' or Indirect Mood. This latter includes two distinct 
 series of forms, called respectively Subjunctive and Optative 
 ' Mood,' and denoted in Indo-European by distinct sufiixes ; 
 but the grammatical relation of these two so-called ' Moods ' 
 ^ Curtiua' 'Greek Grammar,' § 226. 
 M 
 
1 6a Verb Inflection. [chaf. 
 
 Elements of is rather analogous to that of Primary and Historic Tenses 
 
 Verblnfleo- . , ^ _ . ,_ , , . , _ . 
 
 tion. m the Indicative Mood, and in the Latin verb' but one 'Sub- 
 
 junctive Mood ' is recognised. The position of the sign of 
 mood (see above, p. 159) is appropriate to its functions, as 
 modifying the relation between the subject (person-ending) and 
 action (verbal-stem).] 
 
 Tense HI. Tense Stems may be thus enumerated : — 
 
 stems ; . , , , 
 
 1. Perfect stem, originally formed by reduplication. 
 
 2. Simple or Strong Aorist (2nd aor.), generally ex- 
 hibiting the verbal-stem in its simplest form K 
 
 3. Present stem, from vfhich with the augment is formed 
 the linperfect in Greek. 
 
 4. Weak or Compound Aorist. 
 g. Future stem. 
 
 6. Strong Passive Aorist (2 aor. pass.). 
 
 7. Weak Passive Aorist (i aor. pass.).. 
 
 By 'Strong' tense-stems we mean those which are formed 
 from the verbal stem^ by reduplication or increase. 'Weak' 
 or ' Compound ' stems are formed by combination of the verbal 
 stem, generally with some formation from the root cbs (is), 
 ' to be.' Under this head fall also such subordinate formations 
 as the Pluperfect and ' Futurum Exactum' in Greek and IJatin, 
 from the Perfect Stem; or the special formation of the Imper- 
 fect in Latin. 
 
 Two other elements enter into Verb Inflection, an explana- 
 tion of which may be given here — the Augment and the 
 Thematic (or Connecting) Vowel'. 
 
 IV. TJie Augment. 
 
 The Aug- Language seems originally to have employed, as a means for 
 
 expressing past time, the Augment — in Sanskrit a, in Greek e 
 (a-bhar-a-n, f-tpcp-o^v), prefixed to aorist, imperfect, and plu- 
 perfect tenses in both those languages. It is always accompanied 
 by the secondary person-endings : but Gurtius (' Das Verbum,' 
 
 ^ As few Greek verbs develop both the strong and weak forms of the 
 aorist, there are practically five groups of tenses in each Greek verb, viz. 
 Present, Aorist, Future, Perfect, and Aorist Passive. 
 
 ' See Curtius' 'Greek Grammar,' § 246. 
 
 merit ; 
 
VIII.] Verb Inflection. 163 
 
 p. 104) suggests that it was originally the sole expression ofTheAug- 
 past time, and by increasing the word at the beginning gave 
 occasion for shortening the terminations. It was probably at 
 first a separable prefix :. for (i) in older Sanskrit it is separable 
 and (as in Homer) omitted at pleasure.; (2) this separable 
 character would help- to account for its total disappearance 
 in Latin, and in Greek from all moods but indicative. Of 
 its origin there have been various theories : — 
 
 (i.) That it is a variety or abbreviation of Reduplication ; Theories of 
 the vowel of both being e, and the two apparently coinciding in 
 such perfect forms as e-yvaxa, £--\jfa\Ka. But this resemblance 
 seems to be purely accidental and the pluperfect tense, with 
 both augment- and reduplication (i-ye-ypa^uv), points to. their 
 being distinct forms. 
 
 (ii.) That it=' c? privativum,' because past time is a denial of 
 present time — ' Past = Not Present.' This theory scarcely needs 
 refutation., 
 
 (iii.) The view generally adopted (e.g. by Curtius', Schleicher, 
 etc.) is that it is a demonstrative pronoun-stem referring to 
 past time,, like the German da, damals. Its original form in 
 Greek (as- in Sanskrit) was a, of which traces remain in. the 
 Greek dialects (e. g. air^fo-Se, a^paxf, SSeipe). This before a 
 consonant became e (' Syllabic Augment ') : . before a vowel it 
 took the form of that vowel and combined with it to form one 
 long syllable ('Temporal Augment'). Thus, e.g. the Doric &yov 
 (imperf. of aya)=z.a-ay-ov : and &PTO corresponds to Sanskrit 
 S.rta, which no doubt arises from a-arta. This contraction took 
 place before the separation of the a sound into a, e, (p. 36): 
 and .after the root wr- became fixed in Greek as op-, Spro stood 
 to opvvfi.1 as arta to ar-nau-mi in Sanskrit. With initial i and 
 V we should have expected a diphthong, i.e. ai, au{el, ei): but 
 verb-stems beginning with i or v seem to have followed the 
 analogy of other initial vowels ; and no doubt the augment 
 came to be regarded as nothing more thana lengthening of the 
 vowel. 
 
 ' 'Elucidations,' p. no; 'Das Verbum,' pp. 104 sqq. 
 
164 
 
 Verb Inflection. 
 
 [chap. 
 
 The Aug. 
 meat. 
 
 Apparent 
 irregulari- 
 ties in the 
 Augment 
 explained. 
 
 The separable character of the augment in Sanskrit and 
 Homeric Greek is no proof that it is unessential. ' Language ' 
 (as Curtius well remarks) ' not unfrequently lays aside individual 
 symbols of meaning, when by means of them forms have been 
 coined so distinctly marked, that the original elements are no 
 longer absolutely necessary.' In Attic Greek it is never 
 omitted except in xp^" impf. of xP"? > in a few instances at the 
 beginning of lines in the speeches of ayyeXoi in the Tragedians * ; 
 and occasionally in pluperfect tense (but mainly in the Hellen- 
 istic Greek of the New Testament). 
 
 The position of the augment in verbs compounded with 
 ^prepositions (Curtius, Greek Grammar, § 238) is due to the 
 looseness of connection between verb and preposition. Where 
 however the parts of a compound verb are not so separable, the 
 .augment is placed first : e. g. awtoSd^ijo-a from oiKoSo^em. 
 
 Certain apparent irregularities in the form of the augment 
 (Greek Grammar, ,§§.2 36, 7) may be explained by the loss of a 
 consonant : — 
 
 (a) ct instead of rj before idi^to, eXxm, inofiai, cpya^o/xat, ex.'", 
 
 epira, itmaai, edco,' etc. With the exception of eda>, the origin of 
 which is doubtful, it can be shown that all these verbs began 
 originally with a consonant, and therefore had originally the 
 . syllabic augment f, which after the loss of the initial consonant 
 . naturally coalesced with the following e into tt : e. g. Fepyd^ofuu 
 {'work,' see p. 68), imp. epepyaCo/njV. ((r)ep7t<o (Latin serpo), 
 f-(Te(mov, e-epirov : feKltraio {volvo), e-Ff\t(r(rov ^. 
 
 (6) Syllabic augment before a vowel in eaSov (avbdva), iaBavv 
 
 {i>6ta>), imvovfuriv {itveofiai), etc' Thus &vhavM^=<Tra.vhdvto (Latin 
 
 suavis, cp. the Homeric form €0a8e) ; aiveofiai=if<i>viO)i.ai (Latin 
 vendo). The loss of the consonant was perhaps in the first 
 instance compensated for by lengthening the preceding vowel, 
 
 ' In such exajuples as Soph. Oed. Col. i6o3, .tox« TrSpevcav, and i6o8, 
 irarpos ircaovaat kkmov, we probably have instances of 'prodelision ' of the 
 initial vowel after a filial vowel sound of the preceding word. 
 
 ' Curtius (' Das Yerbum,' I. pp. 131-126) examines fourteen words, in 
 seven of which he traces the disappearance of f, in five that of a. 
 
 ' t-dyriv, idX'qv, t-dh-av, i-&vaaae (Alcaeus), i-i'ittoiv, k-iaaaTO, l-rjicf, 
 i-iipom, dyi^jiyov, are other examples under this-head. Most of the words 
 referred to are discussed by Curtius in his 'Prineiples of Greek Etymology.' 
 
VIII.] Verl Inflection. 165 
 
 i. e. the augment itself, whence such forms as Epic ^£iSiii=E-/rfi'Si7 Elements of 
 (root vid-) : but afterwards the reverse process took place and tion. 
 the following vowel was lengthened, whence such forms as 
 i-fjvSavov, i-avoxoei (Homer), l-itpav (root F^P't cp. Latin ver-eor),. 
 eoKav, apparently with a ' double augment.' [Two exactly 
 similar processes of compensation for the loss of f (u) are seen, 
 in the forms /Sao-iXij-of, ^aa-iKe-as, both representing ^aa-iKeF-os 
 (stem jSao-iXfD-), see above, p. 118.] 
 
 (c) Doubling p after augment is generally owing to the fact 
 that a consonant has fallen out before it ; which consonant can 
 sometimes be discovered by comparison with the kindred lan- 
 guages, e. g. in tppeov:=i-(ip€f-ov, Sanskrit a-srav-a-m, from root 
 <rpv-, Sanskrit sru-, and in fppenov = t-fpe7!-oviiova a root fpeir-. 
 the p of which is seen in KoKar-vpoi^. 
 
 Y. The 'Thematic Vowel.' In the ordinary conjugation ofThethema. 
 Greek verbs, the person-endings are not added directly to the 
 stem, as in the conjugation, of 'verbs in -/xt' (e.g. in-jitv, Si-Sore): 
 but between the pure verbal or tense stem and the inflection 
 there intervenes (especially in Present and Strong Aorist) a 
 vowel which appears as o,, <o. or f — e.g. tu?it'<» = TunT-<a-/ii, 
 nnrT-o-iiev, 'rivT-o-v-n (Doric for tutttouo-i), ervTr-e-s — and per- 
 haps also as a in the Perfect Tense {yeyov-a, . yey6v-a-fiiv). Im 
 Latin it appears as 0, i, u — leg-o ( = leg-o-mi), leg-i-miis, leg- 
 u-nt; as a in er-a-m, ( = a,s-a-m, Greek rja) ; and perhaps as e, 
 i in Noun declension peJ-e-m, homih-i-bws (see however p. 116). 
 
 The nature of this vowel has been much disputed. Bopp Various 
 
 « \ 1 -, . '11. theories of 
 
 (Comp. Gram. § 495) regarded it as a pronommal element its origin. 
 ' through which the action or quality, which is expressed in the 
 root in ahstracto, becomes something concrete — e.g. the ex- 
 pression of the idea " to love " becomes the expression of the 
 person " who loves." ' This however is the function of the 
 person-ending ; and besides, all analogy of language shows that 
 ■ concrete' conceptions are prior to 'abstractions.' Others (like 
 Pott) take refuge in metaphor and call it the sinew (Nerv) of 
 the verb : but this explains little, and leaves us to wonder why 
 the verbs in -jii and the second principal conjugation in San- 
 skrit lack this ' sinew ' altogether, without their vitality being 
 
i66 Ferb Inflection. [chap. 
 
 impaired. More satisfactory and more consistent with facts 
 Theory that was the view assumed by Curtius in his Greek Grammar and 
 nectin^ discussed more fully in 'Tempora und Modi' (pp. 39-52), that 
 this vowel is not a suffix of any actual meaning, but a purely 
 phonetio element. A definite meaning, he there urges, can only 
 be ascribed to it on the supposition that it originally belonged 
 to all verbs : whereasthe history of language teaches us that in 
 the oldest verbs, both of Sanskrit and Greek, it is wanting ; nor 
 have we any ^ound for assuming that they ever had it. 
 
 On these grounds it was concluded that the vowel in question 
 is a ' connecting vowel,' a device of language whose primary 
 occasion is to render easier the pronunciation of two con- 
 tiguous sounds — as e.g. in the forms /3/3ejM-f-r;;r, veCJjeKrjyep-e-rris, 
 yev-e'-Ti/s, gen-i-tor, Sanskrit g^n-i-ta ; (()ep'e-Tpov beside cpeprpov ; 
 Sanskrit tup-i-tas beside tvitto-s ; in such noun forms -as homin- 
 i-hus ; and in 'verb forms like Xey-o-/ii6i', 7r€t5-o-;ttE5a. In all these 
 forms the pronunciation is no doubt rendered easier by the 
 insertion of a vowel ; while forms like Tinrr-fiev, nW-ade could 
 hardly be pronounced at all without it : and thus the principle 
 of euphony seems to coincide with the effort to attain clearness ; 
 for the direct addition of person-endings to consonant-stems 
 could not have been consistently carried out without the elision 
 or modification of 'important consonants (e. g. the m or s of first 
 or second personal pronoun), and consequent obscurity of mean- 
 ing where clearness was all-important. 
 
 This theory of a ' connecting vowel ' has much at first sight 
 to recommend it : for it accounts, by a sufficiently natural 
 explanation, for a large number of the phenomena to which it is 
 applied in Greek and Latin. It is not, however, borne out by 
 the phenomena of verb-conjugation 'in Sanskrit': and it has now 
 
 ' E . g. in verbs of the ' Tud ' class of verbal bases ending in ^, a, the 
 accent always falls on the a added to the root ; thus, base H^, tud, to strike, 
 3 sing. H^fifi tud-S-ti. This stress is against the vowel being an unmean- 
 ing phonetio adjunct : and so indeed is the employment of ^, a, the strongest 
 and fullest of all the vowel sounds. 
 
 In one sense, of course, it is a ' connecting ' for perhaps we should rather 
 say 'intermediate') vowel, as coming between the original stem or root' 
 and the terminations. 
 
Till.] Verb Inflection. 167 
 
 been abandoned by Curtius himself, who, in his recently com- 
 pleted work ' Das Verbum der Griechischen Sprache,' explains 
 the vowel an question. as a 'thematic vowel,' i.e. a suffix to or 
 increase of the stem or 'theme' previous to the reception of 
 inflections. Instead of the person-endings being attached 
 directly, for example, to the root ag, a nominal stem aga is 
 formed from this root by addition of the pronominal suffix a 
 (an addition which, in the early stages of word-formation, has 
 the same 'attributive' force as an adjective or pronoun with 
 a suj)stantive has in a more developed stage of language, e. g. 
 ag-a, Greek ay-6{-s) : root ag : : ovtos 6 avijp : dvfjp) ■ and this 
 stem ago, is ■ combined as a verbal stem with ta, (ti), the sign of 
 3 sing. (aga-ti:=ayei, Latin agi-i), just as in its capacity of 
 nominal stem it is combined with the demonstrative element 
 s{a), the sign of nom. sing {aga-s=^ay6-s). So ag-mas, a con- 
 ceivable I plur. form, would stand to aga-mas (ayo-nes, agi-mus), 
 much as the Latin nominal-stem ag-men might stand to a con- 
 ceivable form agi-men on the analogy of regi-men. These a- 
 stems had in course of time so overgrown the earlier stratum, 
 80 to speak, of forms which attached the person-endings directly 
 to the root, that they became the rule of conjugation, the others 
 remaining as a group of more or less exceptional and anomalous 
 forms (in the conjugation of verbs in -fit). 
 
 With this increase of verbal stems by a might perhaps be 
 classed the addition of the suffixes na, nu (e.g. in trKtS-va-fiev, 
 SeU-iw-fiev), which a comparison of the cognate languages shows 
 to have been a very early feature of Indo-European word- 
 formation : and Keduplication and Vowel-Intensification might 
 also come under the general head of modification of the root in 
 the formation of words. But all of these are more conveniently 
 considered elsewhere, under the head of the special phenomena 
 produced by them. 
 
 Classification of Verbs. 
 A. Latin : — 
 
 The traditional classification into four ' Conjugations,' accord- Classifica- 
 ing to the vowel preceding -re in the infinitive Mood, establishes, in Latin. 
 
1 68 Verl Inflection. [chap. 
 
 Ciassifloa- by a happy instinct rather than upon any scientific principles, a 
 
 Verbs. tolerably complete classification upon the same principle as that 
 
 already applied to nouns (p. io8), viz. according to the final or 
 
 ' characteristic ' letters of the stem. The Latin Grammar gives 
 
 us under its four heads v — 
 
 1. Vowel stems in a- (ama-). 
 
 2. Vowel stems in e- (mone-). 
 
 3. Consonant-stems and stems in ?- iU semivowel {rego, 
 capi-o, trihu-o). 
 
 4. Vowel stems in i- {audi-). 
 
 This division may be retained, but simplified into the broad 
 distinction already drawn in case of nouns between (a) consonant- 
 stems (including i-,.u- semivowel), i.e. the '3rd conjugation,' 
 and (6) vowel- stems, i.e. the ist, 2nd, and 4th 'conjugations.' 
 The fluctuating forms of tenses may be considered separately 
 under the different 'tense-^tems.' 
 B. Greek:— 
 In Greek. The traditional classification of ©reek verbs into verbs in -a, 
 contracted verbs (in -m), and verbs in -/tt, though in itself far 
 from scientific, may be to a certain extent utilised as the basis 
 of philological analysis. The -a of i sing., represents, as we 
 have seen, the thematic vowel, which is a distinctive feature of 
 the ordinary conjugation of verbs : . while the verbs in -^t stand 
 apart as a form of conjugation, rarer, and for several reasons 
 presumably older, than that in ordinary use. In these verbs, 
 for example, (i) the pronouns which form the person-endings 
 are less obliterated, and are added directly to the stem (fV-^ieV, 
 8iSo-te) without the intervention of a thematic vowel ; (2) this 
 form predominates in Sanskrit, and is more frequent in the 
 oldest dialects of Greek ; (3) the verbs in -ju contain the most 
 elementary roots and denote the simplest ideas 'to be,' 'to 
 give,' etc. Putting these then aside as one form of conjugation, 
 we may set on the other hand the conjugation of all other verbs 
 (treating the fluctuating forms of tenses under the head of 
 Tense-foymation), dividing these according to the final letter 
 of the Present-stem (exclusive of the thematic vowel). Thus we 
 have 
 
'^iii'] Verb Inflection. 169 
 
 I. Verbs in Q. ■ Classiflca- 
 
 A. Vowel-stems. Verbs in 
 
 1. Uncontracted, \v-a>. 
 
 2. Contracted, Tijud-o), rrote-a, Rov\6-w. 
 
 B. Consonant-stems. 
 
 1. Guttural, ttXek-o). 
 
 2. Dental, i/ceuS-o-fiat, 7reid-a>, KO/itf-<o. 
 
 3. Labial, Tre/nTr-a), Xfi7r-«o.. 
 
 4. Liquid, Sep-m, dyyeXX-o). 
 
 II. Verbs in MI- 
 
 1. Inflections of the Present joined directly to stem: 
 
 <jiri-fil.. 
 
 2. Present stem formed by adding w to the pure stem ; 
 SelK-vv-fii. (These verbs belong to this conjugation 
 only in respect of the inflection of the Present stem). 
 
 This 2nd principal Conjugation (verbs in -p) differs from 
 the ist (verbs in -to) only in the inflection of the Present and 
 Strong Aorist stems : and here the basis of distinction is the 
 presence or absence of the ' thematic vowel,' a distinction which 
 appears in the same manner in Sanskrit, e. g. As-mi, bhar-a,-mi, 
 compared with eV-/*!, (f)ep-a(^iii). The inflections of persons 
 themselves are not distinct iii the two classes, except so far as 
 the -/it verbs retain a fuller and less impaired type of the person- 
 endings — especially in their 'primary form' (see above, p. 161) 
 — and offer greater facilities for tracing the origin of these 
 inflections. 
 
 I. Person-endings [A ciive). 
 
 I Sing. : Sanskrit -mi, Greek -fu, from ist pers. pronominal Person-end- 
 stem ma (weakened as in mi-hi, and German mir). Secondary 1 Sing, 
 form : Sanskrit -m, Greek -v, of imperf. and 2 aor. The full 
 termination remains in pres. indie, act. of ' verbs in -fit ;' in cer- 
 tain Homeric conjunctive forms, i6eKa>-fu, eiTrafii, etc. ; and in 
 optative forms, Tiirroi-fii, tv^oi-im. In Latin it appears as -m 
 in two present tenses indie, sum and inquam, and in the termina- 
 tions of I sing. imp. and plup. indie, and all subjunctive tenses 
 throughout ; and in fut. indie, of consonant and I- verbs. It 
 remains also in English am, German bin. 
 
170 Verb Inflection. [chap. 
 
 Person- I Plural : Sanskrit -mas, Greek -/les (Doric), Latin -mus. 
 
 1 Piur. " Two explanations (or rather, guesses at explanation) are given 
 
 for this form, (i) that it=m + as of plural nom. (as in 7ro8-fr, 
 pad-as): (2) that it =^ ma-si, i.e. ist-f-znd pers. pron., so 
 that ' we' = 'I + thou.' A form -masi is actually found in Vedic 
 Sanskrit. The ordinary -fiev of Attic Greek arises from -fus 
 by loss of 9, and subsequent addition of v etjjekKva-nKov. 
 
 I Bual ; Sanskrit -vas, a variation of nom. j)Iur. -mas: cp. 
 vayAm, nom. pi. of ist pers. pronoun (p. 144). In Greek the 
 nom. plur. of active forms serves as nom. dual. Lithuanian 
 retains -va, e.g. e«^'ua=(as)-vas, nom. dual of Sanskrit as-mi 
 {sum). 
 
 2 Sing. 2 Sing, The 2 pers. pronominal element tva (see p. 145), or 
 
 by loss of V sound ta, appears in Indo-European inflection with 
 both consonant and vowel weakened, i. e. t by aspiration to th 
 or dh, or by weakening to s (p. 78) and a weakened .to i. 
 The series of possible forms, then, of this suflSx is -ta, -tha, -thi 
 (dhi), -si, -s. Of these the last two are most generally found, 
 as primary and secondary forms respectively; thus Sanskrit 
 has in pres. indie. - si, in imperfect -s ; Greek -a-t (as in ar-o-i, 
 Ionic for ti) and -9 {e<pepe-s). The ordinary termination in -ek 
 is variously explained. The primary form is adinitted to be, 
 e. g. <fiepe-(Ti, whence, by simple loss of final i, comes the Doric 
 form in -es, e. g. (nipwSes (Theocr. i. 3), afieXyes (iv. 3). Some 
 hold (f)fpns to arise from raising -es in compensation for the loss 
 of (, the changes being cj)epe-(n, ipfpes, (jtspeis : others (as Curtius) 
 make (pepei-ai the middle step, and regard the second syllable as 
 raised not in ' compensation,' but by stress of pronunciation : 
 while others (as Schleicher) regard the t in ^epets as thrown 
 back from cj>epecn (as e.g. in pf\mva=pi\avya, ^aivai^(j)av-ya>). 
 On the whole, perhaps, this last is preferable. In Latin es= 
 es-si (ea-crl), es {edo)-=ed-si, legis^lege-si. The original quantity 
 seems to have been legls : and as we know that ?, ei are often 
 interchanged, we may assume Zeg'eis= Greek \cyets, and account- 
 able for whatever interpretation we give to the latter. It is 
 worth noticing that a Boeotian Greek form Xeyis (with 3 sing. 
 Xeyi) is found. 
 
Till.] Yerh Inflection. 171 
 
 The imperative 2 pars, suffix -61 (= Sanskrit -dhi) is an older Person- 
 form, commoner in Epic dialect (rerXa-^t, ma-Bi, Spm-Oi, etc.),cta^es 
 but surviving in Attic forms, like yvS>-6t, to-^i=?ir-A from stemS^^™?^,. 
 «-, or FiS-di from stem FiS (8 assimilated, p. 74), i-flt, <tt9j-6i, 
 and I aor. pass, rpairri-di, Tvcftdri-n (by dissimilation, p. 79). 
 In ordinary Greek this termination is changed in various 
 ways : — 
 
 (a.) The vowel is dropped, and 6 changed into the sibilant 
 whence 80s, Sis, o-x^s = 86-61, a-xe-di. 
 
 (6.) 6 is dropped, and the preceding vowel lengthened in com- 
 pensation, whence to-ri/ (larra-di), 8l8ov {8180-61), SecKvv {8dKvv-6i.). 
 
 (c.) In 2 sing, imper. -act. of ordinary conjugation, the ter- 
 mination is lost altogether, and final -e of rvnTt, etc., is the 
 connecting vowel. In i aor. imper. -v is added, and the a 
 which is characteristic of -the weak aorist stem sinks to 0- 
 
 (\C«ro-j'). 
 
 In Latin the termination -dhi of imperative has disappeared 
 altogether, leaving the bare stem (or stem + connecting vowel), 
 as i, ama, es, curve. 
 
 The oldest and least corrupt form of this inflection {-ta, -tha) 
 is retained in the perfect-stem, e.g. Sanskrit dadi-tlia= Latin 
 dedisti; Sanskrit vet-tha (stem vid-)= Greek ol<T6a (stem F18) 
 \Fi.8 raised to F018- as Sanskrit vid to 'ved : o'S-ra becomes by 
 dissimilation (p. 79) otb-ra, and this by influence of the spirant 
 ois-6a\ ; cp. ^a-6a (stem es=^f-ra). In these two Greek examples 
 the a- preceding 3a is satisfactorily accounted for as part of the 
 stem. But there are a number of other instances (chiefly in 
 the Epic dialect) of -a^6a as -2 sing, termination, where no 2 Sing, form 
 such account of a- is possible. These are thus enumerated 
 by Curtius (' Das Verbum,' p. 50) : — 
 
 (a) Twelve Homeric conjunctives : efieXijo-fia (H. i. 554, and 
 fifteen other passages), emrja-da (xx. 250, Od. xi. 224, xxii. 373), 
 
 ^6iKrja-6a (Od. xii. 221), ^ov\iVJ]a-6a (II. i.'99), 8r)6ivri(Tda (Od. xii. 
 121), cv8t](rda (viii, 44S), txw^'^ (II- six. 180), tqirBa (x. 67), 
 ivadrjiT6a (xxiv. ggl), ffape^cXdoTjo-^a. (xisiii. 344), mrjo-da (xxiv. 260), 
 a-iTev8t]a-6a (Od. iv. 591). 
 
 (b) Five Epic, four Aeolic, one Doric Present Indie, and one 
 
173 Ferh Inflection. [chap. 
 
 endings Entire Indic. : rWijo-^a (Od. ix. 404), <i>7iaea (II. xxi. 186), 
 SiSoio-^a (Bekker SlBcoa-Sa, II. xix. 270), etcrda (x. 450); exaaSa, 
 ^iKjjirBa (Sappho), ideXeia-Ba (Theocr. xxix. 4), iroBoprja-da (vi. 8), 
 XPW^a (Megarian, in Arist. Ach. 7.78), a-xria-furda (Hymn Cer. 
 366). 
 
 (c) Imperfect Indic. : (^o-flo), ((prjo-Oa (II. i. 397), ^ua-6a (Plato, 
 Euth. 4, Tim. twenty-six; in compounds), y8r,a-da (v. 1. ^Seiada, 
 Od. xix. 93, and Attic). 
 
 (d) Optative: ^akoia-Ba (II. xv. 571), KKaioiaBa (xxiv. 619), 
 npo(j)iyoia-da (Od. xxii. 3.2 5),. Eii;(r5a (Theognis). 
 
 Various explanations have been given of these forms : — 
 
 (i) Bopp suggested that they were due to a false analogy 
 from ol<r6a, rjo-Ba, causing -a-Qa to be regarded as the termina- 
 tion. But we should, expect to. find the effects of such analogy 
 either in one or two isolated cases, or carried to a much greater 
 extent, as some hold to be the case with the Teutonic termina- 
 tion -St (e.g. German hist, Jiast, gib-st, cp. English doest). 
 
 This St- is held by some philologists (e. g. Schleicher, Comp. 
 § 272) to arise by false analogy (such as that supposed by Bopp 
 for -o-6a) from the case of dental stems, where t, d became s 
 before -ti, e.g. Gothic vais-t (stem w<)= Greek Foio-6a; this si- 
 being first applied to all perfect stems, and then extended to 
 all 2 pers. sing, as in German and English. With this view 
 Bopp's view of -a-da as given above would of course harmonise ; 
 the difference being that the analogy is more consistently and 
 naturally carried out in the Teutonic than in the Greek 
 forms. 
 
 {2) The view of the older grammarians, followed by Schleicher 
 (§ 272) and others, is that these forms iii -ada are a later 
 formation by the addition of -to to the customary form ending 
 in s, e.g. cxfto-Ba, ^oKour-da, etc. Language no doubt offers 
 analogies to such a re-creation of a grammatical form by the 
 fresh addition of an element, which is there already, but 
 obscured (e.g. reov-s /ieov-s, p. 143): and the comparatively 
 small number of forms (and that in an early dialect) in which 
 -o-da obtains may be due to the fact that an awakened gram- 
 matical knowledge saw the needlessness of such a repetition 
 
viic] Veri Inflection. 173 
 
 of the same element, and drove it out of the literary language'. Person- 
 In any case the literary dialect of a nation is formed by the 
 survival of the fittest among a number of spoken forms ; and it 
 is quite conceivable that this form in -aQa may have been one of 
 such by-forms, holding its ground still iu Homer before the 
 introduction of writing, but then disappearing. Nor is it 
 necessary that the explanation of Greek -ada should be uniform 
 with that of Latin -sti or Teutonic -st, if we suppose that the 
 development in question took place after the separation of :these 
 different branches of the Indo-European family. 
 
 The original 2 sing, element, however, is perfectly recognis- 
 able, though weakened, in ex^is ; and it is not easy to conceive 
 auy motive for the addition of to. Moreover, the stage of word- 
 change at which ta-=-iva had sunk to s (s") implies that this 
 pronominal element was no longer recognisable in its earlier 
 form ; and it is probable that, granting the possibility Of such 
 repetition as is assumed, we should not find the repeated ending 
 in so early a form. 
 
 (3) Another explanation looks to the root is for explanation 
 of this <T, as of the o- in other verbal forms (i aor., future in -o-<a, 
 etc.) -ada would then =e(r0a, as -crav in £bo<Tav^e<Tav(T)^erant. 
 The Latin perfect forms vidi-sti, vidi-stis (estis), and vide-runt 
 (sunt) seem to bear out this view : and it is certainly difficult to 
 lose sight of the similarity between the Latin -sfi and Greek 
 -crda, though (as will be seen below) the analogy of the two 
 forms is not certain — one interpretation making the is of e.g. 
 ded-is-ti a tense suffix as in infin. is-se, and perhaps originally 
 a part of the verb sum. 
 
 (4) Curtius, comparing the form -a6a with the -other verb 
 terminations in which we meet with the combination <t6, regards 
 this a-6 as the result of phonetic change from an original tt. 
 The terminations which exhibit a6, are — 
 
 1. 2 Sing. Act. -(r6a (ecj)ri-crda). 
 
 2. I Plur. Mid. -ii€<rda (ord. luBa). 
 
 3. 2 Plur. „ -(T6e. 
 
 '■ It is possible that for this, and many other cases of the final settlement 
 of dialectical forms, we are indebted to the Alexandrian grammarians. 
 
174 ^<2^^ Inflection. [chap. 
 
 Person- 
 endings. 
 
 
 
 Verb Inflection. 
 
 4- 
 
 2 Du, 
 
 Mid. -(t6ov. 
 
 5- 
 
 3 Du. 
 
 „ -trdov. 
 
 6. 
 
 3 I^- 
 
 „ -(tBtiv. 
 
 7- 
 
 3 Sing. 
 
 „ Imper. -crflca. 
 
 8. 
 
 3 Plur. 
 
 „ Imper. -a-Bav. 
 
 9- 
 
 3 Du- 
 
 „ Imper. -a6a>v. 
 
 10. 
 
 Infin. 
 
 „ -a-6ai. 
 
 Certain dialectic forms, e. g. xp^"''"'*) «XfoT<B = xph<^6(o,.. iKiaOoa 
 (Locrian); eKe(TTai^=.('kci^6ai, Upca-Tai, xpv'^Tai, etc., seem to show 
 that 6 is not an original and necessary part- of these termina- 
 tions, but may arise from an original t: while such forms as 
 aiToFeiTTodda (Cretan) seem to be a double dental sound, the first of 
 which might become o- by dissimilation. Following up this 
 cltie, Curtius explains the Imperative middle forms (7,. 8, 9), 
 as arising from rr which becomes by dissimilation or,., and 
 finally, by the assimilating influence of r on a preceding tenuis, 
 it6.. Thus. (^a(r5M=^a-T-Tia-(T), the t-tco being the 3 personal 
 pronoun added twice for expression of middle voice. The Dual 
 forms (5, 6), would arise in the sam* way, there being no 
 original difference between dual and plural : and so would 
 the 2 pers. forms -o-5f, -vBov, the first r in their case being the 
 < of 2 pers. pronominal stem tva. Thus for seven of the above 
 forms (Nos. 3-9) the successive change tt, or, crd is established. 
 
 Tor. 2 sing. -<T6a the analogy of the parallel forms -o-^m, -oto>, 
 and -a-6<u, -orat suggests a parallel form -ora, corresponding to 
 the Latin -sti ; a correspondence which is further borne out by 
 Gfithio sais6-sl, i. e. sedisti (the exact parallel to which would be 
 a form sesi-sti, supposing it to exist). Assuming or of ora to 
 arise, as in the other forms, from tt, this tt may be due to 
 'progressive assimilation' from, tv of the pronominal stem tva 
 (as e. g. Terrapes, cp. with Indo-European katvaras) ; and thus 
 we have tv, tt, or, 116 as the scale of change from tva to vBa. 
 e(j)r](T6a would thus represent an earlier ^iprjTTo, just as ^atrOo 
 represents ^otto). 
 
 [Where the data are so few and uncertain, as in the case 
 of this 2 sing. -a-Sa, it is difficult to pronounce positively in 
 favour of one among several conflicting theories. The high 
 
■VIII.] Verh Inflection. 175 
 
 authority of Curtius claims attention to his view, and it isPerson- 
 perhaps based upon wider induction from observed facts than 
 some other views : but Gurtius, like others, has to depend upon 
 assumptions at one stage or other of his argument, and our 
 decision will after all be only an estimate of the comparative 
 probability of unverifiable hypotheses. Iti such cases it seems 
 better simply to call attention to the known facts, and to the 
 most plausible theories that are based upon them, without 
 attempting to lay down that one is right' and all the others 
 wrong.] 
 
 2 Phi/ral. If I' plur. -maB=ma-si, i.e. ma-tva, ' I + thou,' 2 Pto. 
 we should expect in 2 plur. a form =tva-tva, expressing 'thou 
 + thou.' No suoh direct evidence as the Vedic -masi of 
 I plur. is forthcoming; Sanskrit has only -tha (primary) and 
 -ta (secondary), as in bMra-tha pres., abhara-ta imperf , while 
 Greek in all tenses has the weakened form -re. But Latin has 
 -fis, which may represent -tas, i. e. ta-si (' thou + thou') : and 
 Sanskrit in the dual retains a stronger form thas. There is 
 therefore evidence for an original -tas or -tTias; which is susceptible 
 of either of the two explanations offered for i plur. -mas (p. r^o). 
 The Latin imper. form -tote, however, and Vedic Sanskrit -tat 
 seem to point to a doubling of the 2 pers. pronoun-stem. 
 
 2 Dual : Sanskrit -thas primary, -tain secondary. Greek 
 -Tov throughout, perhaps formed like -/lev of i plur. (p. 170) by 
 addition of -v, or corresponding to Sanskrit -tarn, which may 
 be (as explained by Pott)=<v-aOT, -am being an appendage as 
 in aham, vayam (see p. 144, and below on 3 dual). 
 
 3 Sing. The demonstrative pronominal element ta- (in t6-v, 3 Sing. 
 o5-ro-9, is-to, etc., see p. 150) is weakened to ti- / Sanskrit as-ti, 
 Greek fV-ri, ndrj-n (Doric). This n becomes -o-i in Ttdrj-a-i, 
 cjjria-l, etc., by the usual assibilation of t before t, (as in nXoiros, 
 nXovarios ; iviavros, iviav(nos) ^. -<n is also retained in a number 
 
 of conjunctive forms in Homer — ideXjia-i., dirricn, ndS^iri, etc., and 
 in a few used by later writers— e.g. miJ,jTKjjcTi Hesiod, Op. 
 301, Ifjo-i Theognis, iiroTniiirpfja-t Arist. Lys. 348, opo-Jo-t ApoU, 
 Rhod. iii. 1039 : so wapa(j)dairj(ri, optat., in II. X. 346. 
 ' See above, ch. iv. p. 78. 
 
176 Verb Inflection. [chap. 
 
 Person- The ordinary 3 sing, termination in -ei is sometimes explained 
 
 as arising e.g. from (^e'pe-(ri= original (j>epe-Ti by loss of <r between 
 two vowels. But it seems better to exj)lain it on the analogy 
 of 2 sing, in -ns, by one or other of the processes mentioned on 
 p. 170, (fifpe-n becoming qjepeir, and t then falling off, as an 
 inadmissible final sound. This would throw light on the 
 original quantity of such forms as legit, regit in Latin, if we 
 suppose an original lege-ti, legeit, legit. In any case the final -t , 
 of Latin 3 sing, is the secondary form of -ti/ Latin thus 
 retaining the inflection consistently in all tenses, while Greek 
 has lost it from the ordinary Conjugation of verbs, except in -to 
 of imper., Latin -to. Oscan has -tud, and Vedic Sanskrit -tat 
 (see above on 2 plural), which point to a repetition of the 
 pronominal element. [The t of 3 sing, inflection survives as s 
 or th in English, he carries, carrieth; as t in German ist.] 
 
 3 Plur. 3 Flur. : Sanskrit -nti, -n ; Greek -vri (Doric), -v ; Latin 
 
 ■Mt. 
 
 Thus, primary, bharanti, (jiepovri, ferunt. 
 secondary, abh.aran, e^epov, ferebant. 
 
 In Greek the termination appears only in Doric eVi, (pepovn, 
 etc. In Attic Greek t becomes a-, v disappears,, and the vowel is 
 (raised; so ipip-o-vri becomes 4>^pov(n. An old inscription of 
 Tegea (Arcadian) gives the forms Kplvava-i, KeXfiavai, which 
 appear to be a transition stage between -ovn and -oucri. The 
 Lesbian dialect has a-i with i before it in place of v ; (jjalai, xpv- 
 wrouTi, etc : so f^opiouTt. in Theocr. xxviii. n (an Aeolic poem). 
 Latin retains throughout the stronger form (see above, p. 1 7). 
 
 The imperative 3 plur. in both Greek and Latin exhibits 
 peculiar forms : t^ep-d-vra-v (Doric ipep6vTa>, hatin fermito) seems 
 to = Vedic Sanskrit -ntat {t lost and v added), and to correspond 
 to Sanskrit -ntu of 3 plur. imper. The other Greek form -t<o- 
 aav is a later formation, unknown to Homer : it is e. g. 3 sing. 
 cfxpera + a-av:=(TavT, the remains of 3 plur. of ia-p-i (asmi), asanti 
 (p. 17). This -aav is also used to form a later 3 plur. opt. 
 V LOTmrj-crav, and appears in its primary form in 3 plur. perf. act., 
 
 e.g. lijd(Ti=:fiS-(TUim. 
 
 [There is no evidence here for a formation analogous to that 
 
vrii.] Verb Inflection. 177 
 
 assumed for i and 2 plur., so that 'they'='he + he:' and all Person- 
 that we can say is that -ti probably represents the demonstra- 
 tive pronoun -ta (as in 3 sing.), and that the element an or n 
 in some way or other may give the notion of plurality.] 
 
 3 Dual: Sanskrit has primary -tas, secondary -tarn; Greek 3 Dual. 
 -Toi/ is primary =-te(s) + J/; -tj/j- secondary =:-ta.m. Thus bh^r- 
 atas (pres.) = (peperov, Abharatam (imp.) = i^epi-Tr)v ; -rav of 
 3 dual imper. also = -tarn, and so=-tt;i'. 
 
 \-tas might be explained as=-te, -sa (ep. -thas of 2 plur., 
 P- 175); -**«* »»«2/ Ise (as Pott) = to + a»i, a pronominal ap- 
 pendage (see above on 2 dual).] 
 
 Middle and Passive Inflections. 
 
 The name 'Middle' voice, as applied to the Coniugation of Middle or 
 Irreek Verbs, conveys no notion of the real distmction of forms in 
 Middle forms, viz. their reflexive character, expressing the effect Passive, 
 of the action of the verb upon the subject and not (as in the 
 ' Active ' forms) upon an external object. The term ' Middle,' 
 implying something between the Active and Passive Voices, 
 would naturally suggest that these latter . are the original, the 
 Middle a later development of language: whereas it has been 
 established with tolerable certainty that language has generally 
 developed the Passive from the Middle Voice in Verbs. In 
 Sanskrit, for example, we find belonging to each tense two 
 distinct sets of verbal terminations, corresponding (as 'we shall 
 see) to the Active and Middle forms of Greek ; but both active, 
 and sometimes applied indiscriminately to transitive verbs. 
 These are called respectively (i) ' Parasmai-pada,' 'word' 
 directed to another,' because the action passes {pa/rasmai) to 
 another object (cp. the term ' transitive ' from transire) ; and 
 (2) ' Atmane-pada,' 'word directed to oneself,' because the 
 action is restricted dtmane, 'to oneself (dat. sing, of dtman, 
 ' self'). These two schemes of terminations partly answer to 
 the ' active ' and ' middle ' voice of Greek Grammar. Thus, 
 
 ' Pada, =!i,n inflected word as opposed to the uninflected root. The 
 term refers only to a scheme of terminations, and does not necessarily 
 carry with it the associations of ' voice ' in the ordinary grammatical use 
 of that term. 
 
17^ Verb Inflection. [chap. 
 
 (PmSto) ^^^° * ^^^^ '^^ conjugated in both ;padas, 'Atmane-pada ' does 
 
 Infleotioiis. not alter the idea expressed by the root, but directs the action 
 
 in some way towards the agent or subject : e. g. pa/^ati, ' he 
 
 cooks,' palate, 'he cooks for himself;' namati, 'he bends,' na- 
 
 mate, ' he bends himself 
 
 Passive verbs in Sanskrit are conjugated in Atmane-pada. 
 But while in Greek and Latin a verb in the Passive voice 
 corresponds in form to the same verb in the Active voice, 
 the terminations only being changed ; in Sanskrit a passive 
 Verb is a separate derivative from the root (as e.g. causal 
 desiderative or frequentative verbs are) formed on one invari- 
 able principle without any necessary connection with the 
 conjugational structure of the active verb, but using the 
 Atmane-pada terminations, e. g. bhdr-a-ti (^tpet), bMr-a-te 
 ((j)epcTai, middle); but bhri-ya-te {(peperat, pass.) by insertion of 
 the stem suffix ya ^. 
 
 The evidence of the Sanskrit Verb, in addition to the obvious 
 fact that in Latin but one form (and that, as we shall see, in 
 its origin refleadve) serves for Middle and Passive, while in 
 Greek (though there are some special Passive forms) the same 
 form serves for both in certain tenses, supports the conclusion 
 that the original distinction is between 'Active' and 'Eeflexive' 
 terminations; but we may speak of these later under the 
 currently accepted terms of ' Middle ' (or Medio - Passive) 
 inflections. 
 
 Middle The Middle or so-called ' Passive ' Inflections of the Latin 
 
 (Passive) . , „ 
 
 foims in verb may be considered first, not as bemg older, but as exhibit- 
 ing most distinctly this reflexive character. They are formed 
 (with the exception of 2 pers. plur.) by suffixing the reflexive 
 pronoun se to the Active Voice ; the s of se generally pass- 
 ing by the euphonic laws of Latin into r, which is the familiar 
 characteristic of the Passive terminations. Thus to take the 
 Present Tense: — 
 
 1 Sing. : amo-se, amove, amor. 
 
 2 Sing. : wmad-se, amarise, amaris (or possibly, by intro- 
 vention of a connecting vowel, from the ordinary amas, amas-u- 
 
 ' See Monier Williams' 'Sanskrit Grammar,' §§ 2436, 461. 
 
 Latin. 
 
VIII.] Verb Inflection. 179 
 
 se, amasus, amaris. See on ? sing, and cp. a form utarus=^ Middle 
 
 . .... ^ o f (Passive) 
 
 utaris ou an inscription). forms in 
 
 c- r / ,■..-, . Latin- 
 
 3 bmg. : amat-u-se, amatur (u bemg perhaps a connecting 
 
 vowel). 
 
 1 Plur. : amamu(s)-se, amam/wf (or ? amamus-io-ee, cmia- 
 mur-v/re, amamur), 
 
 2 Plur. : amamini (sc. estis) is really a nom. plur. of a 
 participle formation analogous to Greek -iievo-, the singular 
 of which is found in Old Latin for 2 and 3 pers. imper. ^-ae- 
 famino, progredimino (i. e. jpraefaminos sis). The formation of 
 cMnamini {estis) is therefore precisely analogous to that of the 
 perf. pass, amatus sum, es, est, etc. The form amaminor (2 plur. 
 imper.) given in grammars is probably due to false analogy, 
 r being added as the characteristic passive sign : and is supposed 
 by some to have had no existence except with the grammarians. 
 
 3 Plur. : amant-u-se, amantur. 
 
 The same formation is traceable throughout (except where 
 a participle with auxiliary verb is employed, as in perf. and 2 
 fut.) : thus imperf. amabam-se, amabar : imperat. araa-se, 
 rnnare, amato-se, amator : amanto-se, amantor. 
 
 Prom this it appears that ' Deponent ' Verbs are wrongly 
 so termed, as if they had laid aside {deponere) a passive 
 meaning; They are rather to be looked upon as the survival 
 of an earlier stage of language prior to the superseding of the 
 original Middle or Eeflexive by the later Passive force of these 
 inflections. 
 
 The Middle Inflections in Greek have more affinity with Middle and 
 those of Sanskrit, and the explanations offered of both rest flections in 
 upon much less sure ground than that given of the Latin Theories of 
 medio-passive. Like the active person-endings they are capable nectiou vvith 
 of a primary and a secondary form (-licu, -a-m, -Tm, etc., pres. forms. 
 -Ii.r)v, -(TO, -TO, etc. imperf.) ; and these are obviously formed by 
 some increase of or addition to the corresponding active ter- 
 minations. But as to what the precise connection is, ' Gram- 
 matioi certant et adhuc sub judice lis est.'- I give some of the 
 solutions that have been proposed : but the question cannot be 
 determined, with any certainty. 
 
 N 2' 
 
i8o 
 
 Verb Inflection. 
 
 [chap. 
 
 Middle 
 
 OPassive) 
 
 Inftections 
 
 in Greek. 
 
 Theories 
 
 of tlieir 
 
 formation. 
 
 I. That in -fiai, -a-at, -rai, etc. we have a similar formation 
 to that of the Latin Passive — viz, the addition of the refliexive 
 pronoun (sva) to the pronominal elements from which the 
 active person-endings arise. This -sva, Greek -crFe, -a-e or -at, 
 would if thus suffixed give such forms as ma-si, sa-si, ta-si, 
 nta-si, and the falling out of s between two vowels in Greek 
 (p. 66) would leave the Greek -(uu, -a-ai, -rai, -vrat. The 
 recommendation of this view, if it could be maintained, would 
 obviously be that it brings Latin and Greek into harmony 
 upon a point where otherwise they must be regarded (and 
 have generally been regarded) as at variance. Most compara- 
 tive grammarians, however, appear content to accept such 
 variance in the formation of these inflections as fundamental, 
 regarding the Latin (shared by Lithuanian and Keltic) as 
 later; and uphold one of the two remaioiug theories, viz. — 
 
 2. That -/iot, -<rai, -rat, etc. are formed from the active voice 
 by Vowel Intensification (see p. 53), ai, Sanskrit e, being the 
 natural raising or intensification of i to express a change of 
 meaning. The objection to this view is that we find Vowel 
 Intensification employed in the formation of stems, as an 
 agent in Word Formation, but not in Inflection, which in all 
 cases consists in the addition of suffixes. 
 
 3. That -fiai, -a-ai, -rai, etc. are abbreviations from ma^mi, 
 sa-si, ta-ti, etc., i. e. that language expressed the ' reflection ' 
 of the action upon the agent by adding the pronouns bwice 
 over to the verbal stem, once as object case and once as nomina- 
 tive. Against this view it is urged (i) that, if in the doubling 
 of pronominal elements in the plural of active inflections (see 
 above, p. 170) both elements remained, so to speak, in the 
 nominative case (e.g. mas^zego + tv), it is inconsistent that 
 precisely the same phenomenon in the middle voice should 
 give the different result of me {mihi) + ego. [But this dif- 
 ficulty is less, if we bear in mind that the formation of the active 
 inflections may have taken place at a different and much earlier 
 stage in the development of language. The same elements may 
 have combined in different ways at different |)eriods — in dif- 
 ferent strata or layers, so to speak, of word formatioii,] (2) that 
 
Yin.] Verb Inflection. i8i 
 
 if -fiat, -a-oL, -Tat ai'e explained in this way, we ought to find Middle 
 similar forms in the plural {^=: mas-mas, thas-thas, etc) ; (3) that Inflections 
 while the dropping out of s and t supposed by this theory in 
 2 and 3 pers. may be justified by phonetic analogy, that of to in 
 I pers. ma{rn)i cannot be so justified. [It is, however, parallel 
 in Sanskrit bhare=bhar-a-nie (mai) compared with 0cp-o-/iat.] 
 
 Upon the whole, this latter theory meets with most favour, 
 being adopted by both Bopp and Schleicher, and now by Curtius, 
 who in Das Verbum (p. 80) retracts the opposition which he 
 ofiered to it in Tempora und Modi. "We may therefore adopt' 
 it as presumably the correct account of the Greek middle in- 
 flections, or at any rate the best ' working hypothesis ' for 
 deducing an explanation of them. 
 
 The middle Inflections of the Greek Verbs will then be as 
 follows : — 
 
 1 Sing. : Primary form, ma-mi, mm,, Greek -fiai, Sanskrit 
 S. [In Sanskrit the initial m disappears, and we find bhard 
 ^bliar-a-nid=:0e/Dojtiai.] Secondary m,am,j Greek -fi-qv of imperf. 
 mid. Sanskrit has e as in primary form ; abliare = e^epo'/xi?i'. 
 
 2 Sing. : Primary form, -sa^si, -sat, Sanskrit -se: Fn 
 Greek -o-m is retained in verbs in -fu and perf. tense ; tora-o-ai, 
 
 reTvyjfai (jt-a-ai). Epic forms like \i\aUai, Sl^rjai (Od. xi. lOo), 
 oprjai (Od. xiv. 343) have only lost the initial o-; later -eai was 
 contracted into -et, as in (fiepeL=(f>e'pe(^(r)ai ; and later again into -17, 
 as in <t>epri. 
 
 Secondary form -sas, -sa, Greek -tro, retained in imperf. of 
 -/u verbs (eVWetro) and plup. tense irirv^o (Tr-<To). Epic forms 
 with loss of o- only — ipApvao, irapia-Tao, Weo, etc. ; cp. the forms 
 op(To, Se^o, Xf|o from opo-f-o-o, etc. In ordinary Greek c(o-)o 
 becomes -ov, e(^e/>ou=e^£peo-o : so in 2 aor. of -ju verbs, t6ov, 
 cSov and imper. 6ov, 8ov. 
 
 3 Sing. : Primary form, -ta-ti, -tai, Sanskrit -td, Greek -rai 
 retained throughout. Secondary, -ta-t, -ta, Greek -to {i^epero, 
 Sanskrit a-bhar-a-ta). In the imperative we find -(rOa, a form 
 which recalls the -a-Ba of 2 sing, act., and other forms (see 
 p. 173) and the explanation of which is equally uncertain. The 
 most plausible suggestion is that it may arise from -ttco (repre- 
 senting -TO) of 3 sing, imper. act. doubled), rr becoming or by 
 
i83 Verb Inflection. [chap. 
 
 Middle dissimilation (p. 79), and then ad under the influence of the 
 Inflections, spirant. See however, above, on -aBa of 2 sing. act. 
 
 In the Plural terminations it is still less easy to arrive at 
 even a plausible suggestion for their origin: and for i and 2 
 plur. especially such suggestions are but guesses. 
 
 1 Plur. Greek has -/ic5a both as primary and secondary 
 form, with a variety in -fitaOa, found in Homer and later poets, 
 but not in Attic prose, and possibly a mere phonetic varia- 
 tion metri gratia; for in a majority of cases the form -fieOa 
 would not scan. Others however see in -/jte-aBa the form -vSa of 
 2 sing, act.; and on Curtius' view that <t6 here arises from tv, 
 -fieuBa would represent ma-tva-i. Sanskrit has for primary form 
 mahe, Zend maide, which points to a primary form madhai, 
 secondary -madJia, whence Greek -^6a. It is sugge.sted that 
 mad?iai=ma-tva{s)i, mato(«)i='I + thou, to thee' (the reflec- 
 tion of the action being in this case expressed by the repetition 
 of one of the two elements 'I,' 'thou,' which make up the i 
 plur. act.) This would make .ne6a=-iie(r6a. 
 
 2 Plur. Greek has -a-de both as primary and secondary 
 form. Sanskrit has -dhve (primary), Vedic -dhvai and -dhvam 
 (secondary, m being perhaps a later addition), a Vedic imper. 
 
 -in -dhva being found. 3 often disappears in Sanskrit before 
 dh : hence we may infer an original -sdhvai, -sdhva, the Greek 
 equivalents to which would be -a6Fm, -crSFe. This reconciles the 
 Sanskrit and Greek forms, and Curtius' explanation of ad as 
 arising from tt (above, p. 174) presumably covers sdh also: 
 but the origin of this termination also is uncertain. A similar 
 suggestion to that given above for i plur. is made, viz. that 
 sdhvai, a6Fe=tva-tva-{tv)i, 'thou -i- thou, to thee.' 
 
 3 Plur. Sanskrit primary -nte, secondary -nta, correspond- 
 ing to Greek -vrai, -vto. In the Epic forms -arm, -aro (i<p6ap- 
 arai, Kearo, etc.) the a is not a substitute for v, but is the 
 thematic vowel, which in this case has not sunk from the 
 original a sound retained in Sanskrit bliara-iite abharajita 
 {(j)ipovTai, e<j)ipovTo). In Ionic this a is found even after vowels 
 (jSe/SXijaTat, etc.), whence the idea that it stands in place of v. 
 The imper. termination -aOav is on the analogy of other similar 
 
VIII. j Verb Inflection. 183 
 
 forms, and crBa-a-av is a later formation analogous to Ttn-trav of Middle 
 
 . / ,\ (Passive) 
 
 imper. act. (see p. 176). Inflections. 
 
 Different explanations are given of this termination. Schleicher's, 
 
 that it arises from doubling the active termination {^-ant, -anti, 
 
 whence by omission of the second nt, oMtai), assumes that the a 
 
 (Grreek o) preceding nt is not the thematic vowel, but part of 
 
 the inflection — a view which may be true, but has not been 
 
 adopted here. Another suggestion is that ntai-=ntati by the 
 
 addition to the plural idea ' they ' (expressed by 3 plur. act.) of 
 
 the element 'to him;' so that it='he + he to Mm:' another, 
 
 that these 3 plur. middle forms -vrai, -vto are the singular forms 
 
 -Tai, -TO, increased by the addition of the nasal sound for the 
 
 purpose of expressing plurality. But though we have noticed 
 
 before (p. 55) the employment of 'nasalisation' in the formation 
 
 of tense-stems-', there is no evidence for its employment in the 
 
 formation of terminations. All, in fact, is guess-work. 
 
 1 Dual : -iii6ovz=-ixi6a of i plur. with v iipeXKva-TiKov. An 
 Aeolic form -fi^B^v is also mentioned. [Sanskrit has vahe=: 
 yahai (primary), and vahi=valia (secondary).] 
 
 2 and 3 Dual : -a-dov, -a-dqv, and imperat. -a-dav obviously 
 correspond to the act. forms -tov, -ttjv, -rav, as -o-^m of 3 sing, 
 imperat. mid. to -rm (see above, p. 161). In all these forms, 
 the change from r to cr6 seems to carry with it the reflexive 
 meaning: for a possible explanation see above, p. 174, on -o-^a 
 of 2 sing. 
 
 II. Verb Inflection : Mood-signs. 
 
 The function of signs of mood and their position in Verb Mood-signs. 
 Inflection have already been noticed (p. 161). We have now 
 to distinguish the signs for (a) ^ Conjunctive,' (6) ' Optative ' 
 tenses : these two groups of tenses being included under the 
 general term ' Subjunctive (Dependent or Hypothetical) Mood,' 
 by which the supposition of a fact or action is distinguished 
 from its direct statement in the ' Indicative Mood.' 
 
 [a). The special characteristic of the conjunctive forms in Comunctive, 
 classical Greek and Latin appears to be a long vowel (m, ij, a, e) 
 preceding the termination. We find however in Homer a 
 ' Compare Peile, Leot. V. pp. 93, 4, ist edition. 
 
184 Verh Inflection. [chap. 
 
 Mood-signs, limited number of conjunctive forms, such as l-o-fiev, /SX^-e-tqi, 
 Conjunctive. OT^-o-^fv, <^5i'-e-Tat, ak-e-Tai (conj. of SK-to), in which a short 
 vowel 0, e (i. e. original a) denotes the modal element : and the 
 comparison of these with a conjunctive form occurring in Vedic 
 Sanskrit (e. g. han-a-ti, indie, han-ti, as-a-ti, indie, as-ti, from 
 asmi, sum), points to the conclusion that originally the con- 
 junctive was distinguished from the indicative by the insertion, 
 or addition to the stem of a — in precisely the same way, ex- 
 ternally, as the indicative with a thematic vowel from the 
 primitive indicative, so that' conj. as-a-ti : indie, as-ti : : indie, 
 bhar-a-ti : indie, bharti, Latin /er-(<i). And as there is always 
 a possibility that formations outwardly similar may have been 
 originally one and the same, we may perhaps consider with 
 Curtius that the thematic vowel d and the conjunctive suffix 
 originally served the same purpose; and that as bhara-ti, 
 i. e. 'bearer he' (see above, p. 167), developed on the one hand 
 the meaning of ' he is a bearer,' ' he bears,' so on the other hand 
 it might develop the meaning ' he may be, or is tp be, a bearer,' 
 'he is intended for bearing' — i.e. the force of a conjunctive. 
 This however is speculation. The facts to be considered are 
 (i) the highly probable fact that the conjunctive suffix was 
 originally a (0, e) ; (2) the certain fact that, with a few ex- 
 ceptions already mentioned, it appears as a (a, q, a, e). Now in 
 the ordinary conjugation of Greek verbs this long vowel seems 
 naturally accounted for by the combination of the mood-sign 
 with the ' thematic ' or ' connecting ' vowel at the end of the 
 stem ; these two uses of the vowel a, whether originally 
 identical or not, having of course become distinct with the 
 development of verb-forms. But in the conjugation of verbs in 
 -fii no thematic vowel is used (p. 169), and here the long vowel 
 is most plausibly explained as the result of analogy, a, r) being 
 regarded as the characteristic mood signs. Thus from elfu=. 
 icr-fu we have in conjunctive. — 
 
 I Sing, ia-a-ni, iaa, Ito, S>, 
 
 3 Plur. iaSi-vTi, iSivri (Dor.), eaai, Siai, 
 following the analogy of the ordinary conjugation. 
 
Till.] Verb Inflection. 185 
 
 In Latin, the conjunctive suffix a, answering to Greek <b, ;;, Mood-signs. 
 appears in the pres. subj. of and, srd, and 4th conjugations ^ " 
 (consonant and e and ^ stems) ; e.g. mone-d-m, leg-a-mus (=Xey- 
 Mjaes), leg-a-tis (^Xiy-ti-re), audi-a-m. In a- stems (ist conjug.) 
 the mood sign is e (am-e-m) and a few verbs have i, e. g. sim, 
 nolim, possim, edim, duim : but this e and % are both probably 
 optative forms (see below, p. 186^7). The imperf. and plup. 
 subj. in all verbs have e as mood-sign ; amar-e-mus, regiss- 
 e-mus. The perfect has originally I, which however (from 
 confusion with the completed future indie.) often becomes i in 
 dactylic poetry: just as in the completed or 2nd fut. indie. Xis 
 often treated as ^. Thus we find — 
 
 Perf. Subj. -en- : dederitis (Ennius), fueris (Horace), respueris (Tib.), 
 
 dederis, -credideris (Ofid). 
 ■eH- : egerimus, respexeris- (Virg.), dixeris (Hor. in 
 hexameters suspexeris). 
 2nd Eut. Indie, -eri- : viderimus (Lucretius), dixerUis (Ovid), (Virg. 
 Georg. ir. 59). 
 -eri-: dedenHSjtransieritis, etc. (Ovid), /ecerimMS (Catullus), 
 dederis, miscueris, etc. (Hor. in hex.), dederig (freq. 
 in Prop, and Ovid). ' 
 
 [As in Latin ■ the conjunctive and optative coalesce into one 
 subjunctive Mood, we might expect a mixture of conj. and 
 opt., forms such as we actually find. Koby (Lat. Gr. i. § 5'93) 
 suggests that the proper Latin mood-suflSx was t (seen in 
 the Greek optative), which contractfed with a preceding a to e 
 e.g. ama-s, ama^i-s, amies; but as i suffixed to the present 
 indie, of any other than a verbs would give the same form when 
 contracted, an a (seen in Greek conj.) was substituted. This, 
 however, would not be true of consonant stems proper, where 
 there was no opportunity for contraction, e.g. leg-o, le-gi-m : 
 and it would have to be supposed that the d form was extended 
 by analogy to these. On the whole it seems best to admit the 
 confusion of forms, and explain each separately, without re- 
 ducing them to uniformity.] 
 
 (6). Optative. The suffix is ya{ja) (retained in 3 plur. act. Optative 
 of Greek \eyo-ie-v) usually raised to jd, or i : in Greek, le, irj, i. 
 
i86 Verb Inflection. [chap. 
 
 Mood-signs. The verbs in -fu retain the longer form of the suffix (t?j) in the 
 Active Voice, the shorter (i) in the Middle, e. g. : — 
 
 I Sing. iiZo-iijV compsired with Sito-l-piqv 
 
 1 „ SiSo-Jtjs „ SiSo-L-ia)o 
 
 2 „ SiSo-iij „ Sl5o-i-to 
 
 3 Plur. StSo-ii-v „ SiSo-i-vTo 
 
 = SiSo-T-evT. 
 
 Verbs of the ordinary conjugation have the shorter form of the 
 sufBx, which coalesces with the ^preceding o (whether this be 
 regarded as connecting vowel or an addition to the stem, 
 assimilating all stems to a stems, makes no difference here) into 
 the diphthong ot, e.g. (^epoi-ju, rimro-i-fu. In 3 plur., however 
 (as with -^i verbs), the longer form is retained in its most 
 primitive shape (if=jd), e. g. (/jepo-ie-y, Tmrr-o-K-v. With the 
 longer form of suffix the secondary, with the shorter form the 
 primary person-endings are found. ' Contracted' verbs in -<b 
 employ both forms of the mood-sign with corresponding variety 
 of person- endings : thus from T-tfidm (=Ti/ido-/ii) we find pres. 
 opt. Tifiao-i-fu, Tifimfii, and Tt/iao-iij-i', Ti/uoriu. 
 
 The strong (2nd) aor. opt. is formed like the present opt. : 
 
 e.g. So-irj-v, 8o-[-iirjv, Tvit-o-i-ixi, etc. (present, diSol-qv, SiSoifirjv, 
 
 TVTTTOlfllj. 
 
 The weak (ist) aor. employs i as mood-sign, retaining its 
 characteristic stem -letter a, Xia-a-i-fu, \va-a-i-iirjv. 
 
 The passive aor. employs «j as mood-sign with secondary 
 person-endings, \vSe-ir)-v, rvne-iq-v. 
 Optative Optative forms are traceable here and there in the Latin 
 
 Latin. verb. Compare for example the parallel forms of pres. opt. 
 
 from root as {es, es) in Sanskrit, Greek, and Latin : — 
 
 I Sing, s-ya-m ( = as-yi 
 3 „ s-ya-s 
 3 „ s-ya-t 
 
 1 Dual s-ya-va 
 
 2 „ s-ya-tam 
 
 3 „ s-yd-tam 
 
 1 Plur. s-ya-ma 
 
 2 „ s-ya-ta 
 
 3 „ s-yua 
 
 [(trj-aav a later form, see p- 176.] 
 
 tttjv ( = i<r-trj-v) 
 
 s-ie-m ( = es-ie-m) 
 
 sim 
 
 itijs l^~ ^(T-irj-s) 
 
 s-ie-s ( = e3-ie-s) 
 
 sis 
 
 £11; ( = ia-lr)-T) 
 
 s-ie-t ( = es-ie-t) 
 
 sit 
 
 etrj-TOVj elrov 
 
 
 
 el'ffTrjv, eiTrfV 
 
 ' 
 
 
 fttj/ifv, etiifv 
 
 
 Bimus 
 
 eirire, etre 
 
 
 sitis 
 
 (liv{ = ia-l-ivr) 
 
 a-ie-nt ( = eB-ie-nt) 
 
 siut 
 
vin.] Verl Inflection. 187 
 
 The evident correspondence of this old Latin form siem, later Optative 
 sim, with the parallel optatives of the Sanskrit and Greek Latin, 
 verbs, is irresistible evidence that in other so-called conjunc- 
 tive forms in -im, -is, -it, we have optative formations. Thus 
 velim=vel-ie-m, duim (common in Plaut. and Ter. and in old 
 legal language, cp. Di te perduint used by Cicero) =(^aim=<Zas- 
 ie-m, which corresponds to Greek bo-ir)-v, and to a (Vedic) 
 Sanskrit form de-ya-m^ ^oedim from edo (Hor. Epod. iii. 3 ; 
 Sat. II. viii. 90^ : temjperint, coquint (XII Tab.). 
 
 There is also reason for supposing that the subjunctive 
 present of a stems, in which e is the characteristic letter, is 
 an optative form. Beside stet is found Oscan sia-ii= Greek 
 (TTa-ir^-ir), which points to sta-ie-t, sfa-l-t, as the origin of the 
 Latin form. Thus amem=^ama-i-m=^ama-ie-m : cp. Umbrian 
 X>orta-ia{t)-=portet, and Greek nua-o-irjv, nfuir)v. 
 
 The 'future indicative' of consonant-stems (3rd conjugation) 
 in -em, -es, -et appears also to be an optative form, e.g. dicem 
 =deicai-m : a being here the vowel which in Greek appears 
 as o (see pp. 36, 54), as in (jjepo-i-fM, TimTo-i-fu. Thus we have 
 
 Skt. bhara-i-ina(s) Gk. <pep6-i-fi.es hat. fera-i-mus 
 
 bhare-ma (pepoi-fiev /ere-mus^. 
 
 III. Tense-stems. 
 
 These have been briefly classified above (p. 162), and we 
 may proceed to discuss them in the order there observed. 
 
 I. Perfect-stem : — 
 
 The most characteristic feature of the Perfect-stem in Indo- Formation 
 European languages is Eeduplication, i.e. doubling the verbal stem, 
 root. The force of the Perfect Tense is to express completed %wa. 
 action; and for this purpose language seems to have availed 
 itself of the same means or instrument, by which (as we have 
 noticed above, p. 32) frequentative and desiderative verbs are 
 
 ' Curtius quotes also the Vedic forms dhe-y-am = Oeir/v, gfie- ya-s = 
 yvo-'tTj-s. 
 
 ' In Virg. Aen. xii. 8oi, 'Ne te tantus edit tacitam dolor,' Eibbeok's 
 correction, edit, is accepted by Conington : Forbiger, Gossrau, Heyne, 
 Wagher, and others, retain edat. See Couington's note, ad loc. 
 
 ^ The following forms are cited from Zend: — apa-barois = dwo(pepois = 
 au-feres (ab-feroris); bara-yen = (pipo-itvir) =feren,t (fera-int). 
 
i88 Verb Inflection. [chap. 
 
 Perfect- often formed, and any strengthening of the idea of a word is 
 expressed. The earliest conceivable form would be a simple 
 repetition of the root, with a further root indicating the subject, 
 — vid vid ma. The 'agglutinative' stage of language would give 
 vidvidmay and the inflectional stage is marked by frequently 
 raising the .vowel of the second root and shortening the first 
 by loss of its final letter^ e. g. vivaidma ; the process which is 
 seen in so many Greek perfects, e.g. XeXooto (stem Xitt-), neiroiBa 
 (md-), etc. 
 
 Its relation It is indeed not improbable that the Perfect may have been 
 
 sent-stem. a development from the reduplicated present with an intensive 
 meaning. It has the primary person- endings and no augment : 
 its distinctive feature, reduplication, is employed in the forma- 
 tion of certain present-stems ; and while many so-called perfects 
 have a distinctly present force (e. g. eyvaKo, KeKTrjftai, Kexpaya, olSa, 
 odi, novi, memini) in Greek and Latin, it is found that in Vedic 
 Sanskrit, the oldest accessible type of Indo-European language, 
 the distinction between ' intensive ' present and perfect is slight 
 and fluctuating. These facts certainly point to a closer connection 
 between the present and perfect formations than appears in 
 later developments of Indo-European speech, and perhaps in- 
 dicate that the perfect, so far from necessarily implying past 
 or completed action, was at first a mere variety of the intensive 
 present. 
 
 Greek The Greek language both in the form of its perfect-stem, 
 
 by the almost universal . emplo3rment of reduplication, and in 
 its usage, by restricting. the perfect-stem to the expression of 
 the idea of completed action, displays an. antiquity superior 
 to that of Latin ; which, as we shall see below, forms its 
 perfect-stem in four or five different ways (by reduplication 
 -least of all), and uses its perfect-tense as an aorist. Reduplica- 
 tion, therefore, gives to the Greek perfect-stem, in spite of 
 differences in the mode of formation, a unity which it is vain 
 to look for in Latin. 
 
 The redupii- The reduplicated syllable usually contains the initial letter of 
 
 Gated sylla- r j j ^ 
 
 ble. the root with the vowel e, representing original a, the commonest 
 
 root-vowel. Apparent exceptions are due to the phonetic ten- 
 
viii.] Verb Inflection. 189 
 
 dency towards easier articulation : thus an aspirate is repre- Greek 
 sented by a corresponding tenuis ; Trecjyrjva (^av-), cp. ndrjfu ; while 
 of two initial consonants only one is repeated, and that only 
 
 when it is a mute followed by X, fi, v, p, e. g. yeypa<pa, iriiikryya 
 
 {irKay-), n^iri/evKa (ttj/u-). [Exceptions are yv, yX, and sometimes 
 0X — t-yvcD-Ka, f-^Xda-rrj-Ka : and on the other hand the stems Kra 
 and /iva have Ki-Ktrj-jiai, fii-itmifiai. In jTCTTTajKa an e has fallen 
 out between jtt.] 
 
 In all cases but those specified, a stem beginning with two 
 consonants has only r for its reduplication, — surova, e^rjTrjKa, etc. 
 The similarity of sound with the augment (p. 163) which is 
 thus occasioned is, of course, purely accidental : and it is in- 
 correct to say that any verb forms its perfect ' by prefixing the 
 augment.' 
 
 Initial vowels are raised, as opdo-a, Sipdaica : but some stems 
 with initial a, f, o, take ' Attic Eeduplication,' i. e. either 
 (i) doubling the whole root (oB-md-a, root 08-), or (2) repeating 
 the first syllable only of the root or, stem (a\et<f>a}, stem aXi^-, 
 aK-ri\i<j)-a ; e\avva, eXa-, cX-^Xa-iea). This ' Attic ' Reduplication 
 is most frequent in Homer, e. g. apr\poTai, oKdXripm, etc. Herodo- 
 tus has dp-mpr)-K.a from aipea> '. 
 
 The forms e-aXw-xa, c-dy-a, f-oiK-a, e-avrj-pai, are due to the 
 loss of an initial consonant of the respective stems (f ). Homer 
 has e-oXw-a (f eXtt), e-opy-a {hpy, our work) : see above eh. iv. 
 p. 68. 
 
 Certain verbs with initial consonant have n for reduplication, 
 e. g. fiXijp^a, iiKrj<l)a, cljiaprai. 
 
 The root-vowel is generally, but not invariably, raised. It 
 appears that originally this raising was confined to the singular 
 number ; a fact which' explains the apparent anomaly of 'ii-p.€v, 
 tore, etc., in the dual and plural of olba. The conjugation of 
 this particular perfect-stem is remarkably illustrated by Com- 
 parative Philology. The verbal stem is hh, vid, which redupli- 
 cated and raised as above would give vivaida: but the re- 
 duplicated syllable has disappeared in all the kindred languages. 
 Thus we have : — 
 
 ' See Curtius' 'Greek Grammar,' § 275. 
 
190 
 
 Verb Inflection. 
 
 [CHAl". 
 
 Perfect in 
 Sanskrit, 
 Greek, and 
 
 Latin. 
 
 
 Sanskrit. 
 
 Greek. 
 
 Gotliic. 
 
 Latin. 
 
 Stem. 
 
 vid- 
 
 F^5- 
 
 vit- 
 
 vid 
 
 I Sing. 
 
 ved-a 
 
 folS-a 
 
 vait 
 
 vidi 
 
 2 „ 
 
 vit-tha 
 
 foXa-ea. 
 
 vais-t 
 
 vidistl 
 
 3 .. 
 
 ved-a 
 
 potS-e 
 
 vait 
 
 vidit 
 
 I Dual 
 
 vid-va 
 
 
 vit-u 
 
 
 2 ,. 
 
 vid-athus 
 
 fta-rav 
 
 vit-u-ts 
 
 
 3 .. 
 
 Tid-^tus 
 
 fla-Tov 
 
 
 
 I Plur. 
 
 vid-ind(3) 
 
 fta-jjLiv 
 
 vit-u-m 
 
 vid-i-mus 
 
 2 » 
 
 vid-a 
 
 fta-Tf 
 
 rit-u-th 
 
 vidistis 
 
 3 " 
 
 vid-us 
 
 fiaaai 
 
 = fiS-aavTi 
 
 vit-u-n 
 
 viderunt. 
 
 ofacS^ In Sanskrit can be seen the regular working of laws of 
 
 accent which required this change from sing, to dual and 
 plural in the perfect stem. In Greek, analogy has carried 
 the raising of the stem right through the conjugation of this 
 tense, e. g. XeKoma, XeXoi7ra/ie» (for Xe-XOT-/iei) : but the impress 
 of the laws in question remains, as we see in otSa, "cruev; in 
 eiKTOv from e-o«-a(stem eix-), fVe'7rj5-/i6i'plup. plur. from ■ni-TioiB-a 
 (stem m6-) ; and in the shortened plural forms of perf. ^e^afuv, 
 Te8vaii€v, TCTXafieu, cora/iev, fieSijitcv. Latin has lost all trace of it ; 
 but it survives as above in Gothic, and in modern German Ich 
 weiss, ' I know,' Wir wissen, ' we know.' [A similar change of 
 quantity, inexplicable at first sight, between SeiKm-fu and deU- 
 vvfiev, Safivrifu and ddfivafiev, and similar words, is illustrated by 
 the working of the accent in Sanskrit, where exactly the same 
 change is produced, but with greater regularity. In all these 
 cases, Greek has lost the consciousness of the original motive 
 power for the change, retaining with uniformity of accent the 
 results of a difference (cp. e. g. v6d-a, vid-mA with oiSa, la-fiev). 
 These survivals, however, were but few, and the natural 
 process of analogy brought about the complete uniformity of 
 XeXoma, XeXowra/XEi'.] 
 
 Two forms of Perfect Active must be distinguished in 
 Greek : — 
 
vm.] Verb Inflection. 191 
 
 1. Strong Perfect, formed directly from the stem — Trpdo-cro, ■ strong* 
 
 iti-npay-a; tUtco, tc-tok-o; XetVo), \e-\oLn-a: a being the COn- Perfect. 
 
 necting vowel between the stem and inflections, as in XcXo«r-a- 
 /ici/. (Schleicher and others regard the a, like o of (pepoftev, 
 as part of the stem.) The Strong Perfect occurs almost en- 
 tirely in the case of ' radical ' verbs (i. e. whose verbal stem 
 is a root, e. g. \i-a) ; and is generally the older and rarer 
 form. 
 
 2. Weak Perfect, formed from the stem by insertion of k, — 
 fo-ToK-Ka, Kc-Kpi{y)-Ka ; the only form in use in vowel stems, and 
 the most common with stems ending in ,; 8, 6, p, ,1, X, p. The 
 origin of this element k is unknown : it occurs rarely in Homer 
 (about twenty^ out of nearly three hundred known instances) 
 and then only with vowel-stems : and is evidently an element 
 of stem formation, which is perhaps traceable in the isolated 
 aorist forms eSmica, %6r\Ka, r\Ka ; in present forms such as Skk-Ka ; 
 and in aorists such as ^'pnXaKov, eirra-Kov. Curtius suggests a 
 comparison with the Latin ya-c-io, which, if root fa-=^6f, would 
 be a present form corresponding to e-dr)-Ka. 
 
 The aspiration of the final stem letter in forms like yi-ypa(f>-a 
 (ypoTT-), ivrp/ox-a, eiXri^-a, etc. is probably a mere phonetic altera- 
 tion without any definite reason. It is unknown to Homer, 
 who has e. g. xeKonas, not Kexo^ms the usual Attic form : and is 
 found in comparatively few verbs. Curtius ■' enumerates thirty- 
 eight aspirated perfects, most of them not found before Polybius. 
 Bopp regarded these aspirated perfects as a distinct formation, 
 a view which is sufficiently refuted by Curtius (Elucidations to 
 Greek Grammar, § 272, pp. 123-128, English Translation). 
 
 The term ' Perfect Middle ' applied to e.g. yi-yov.a is Perfect 
 erroneous. A perfect Middle or Passive can only be formed Passive, 
 in one way, viz. by affixing Middle person- endings without 
 a connecting vowel to the reduplicated stem, as \e-\v-pai, 
 TeTvppat^Te-Tvw-p^i. The final consonants of consonantal stems 
 change by the laws of assimilation before the initial p, o-, t of 
 the inflections, as in the following table : — 
 
 ' These are enumerated by Curtius, ' Das Verbum,' II. p. 210^ 
 ^ Ibid. pp. 200, 201. 
 
193 
 
 Verb Inflection. 
 
 [chap. 
 
 Perfect 
 Middle and 
 Passive. 
 
 Final Letter of 1 
 Stem aasimilated.J 
 
 Before 
 
 Before 
 
 Before 
 
 Examples. 
 
 
 
 (Seepp. >4-78.) 
 
 
 
 Stem. 
 
 Perf. Pass. Inflec. 
 
 AH Gutturals to 
 
 7 
 
 «® 
 
 K 
 
 Tr\tK- 
 
 Tii-TrXfy-itai 
 \4\eK-Tat 
 
 „ Dentals „ 
 
 a 
 
 rejected 
 
 It 
 
 me- 
 
 )> 
 
 ireireta-imt 
 
 irenet-aai 
 
 ir«ir«i(r-T(U 
 
 „ Labials „ 
 
 M 
 
 T(W 
 
 ■n 
 
 
 yiypa/i-imi 
 
 yiypat/nu 
 
 yiyparrat 
 
 Formation 
 of Latin Per- 
 fect-stem. 
 
 1. Itedupli- 
 catiou. 
 
 The Pluperfect Middle amd Passive differs from the Perfect 
 only in prefixing the augment and in having the secondary 
 Person-endings. The Pluperfect Active is a composite tense 
 and will be treated below. 
 
 Latin Perfect-stem : — 
 
 There are several different ways of forming the perfect-stem : 
 viz. — 
 
 (i.) Reduplication; only in about twenty-seven verbs, and 
 with some of these only in archaic Latin (e. g. te-tuli, te-tini, 
 sci-cidi). Of two initial consonants, the second is treated as 
 initial, and both are retained (unlike Greek) in the redupli- 
 cated syllable; e. g. ste-ii (sta-), spo-pond-i (spend-). 
 
 The vowel in the reduplicated syllable often (as in Greek) 
 sinks to e, especially where the root vowel is a or a weakening 
 of a; dedi (da-), cecini {can-), peperi (par-), cecidi {caed-o^ 
 eaid-), te-tuli (root tol=taT). In compound verbs the reduplica- 
 tion often disappears, e.g. pepuli, expuli; cucurri, decv/rri: but 
 remains in compounds of do, sto, — ahdidi, ahstiti. Reppuli, 
 rettuli, repperi, etc. (sometimes explained as the result of 
 assimilation from red, the earliest form of re) perhaps = 
 repepuli, re-tetuli, etc., e disappearing. 
 
 If the root syllable a of present sinks to e (or i) before two 
 consonants or r (p. 61) ; e becomes u before I {pepuli, pello, 
 sepuUus, sepelio). 
 
VIII.] ' Verb luflection. 193 
 
 (ii.) Raising the Stem-vowel (without reduplication. — Greek Latin Per- 
 employs both); fdve-o,favi; Sgo, egi; jdcio,jeci; lego, legi; 2. naiaiog 
 video, vldi; rumpo (stem rup-), rwpi, etc. Some explain the 
 length of the root syllable in these perfects by the absorption 
 of a reduplicated syllable : i. e. jeci=jejici, legi^legigi ; or, 
 where v is final stem consonant, by absorption of a sufiSxed v 
 (see below iv) : i. e. fa/vi=/dv-vi. But the analogy of Grreek 
 perfect-stems (above, p. 189) where vowel-rsiising and reduplica- 
 tion go together, perhaps points to a similar account of this 
 formation in Latin : one part of the process is lost, but the 
 other remains. 
 
 (ui.)' Suffixing -s (perfect in -si) to consonant-stems; a later s. Suffiiing 
 form, sometimes found side by side with the older reduplicated 
 perfect : e. g. punxi {-ed) with pupvgi ; panxi {pang-si) with 
 pepigi ; intellexi (Jeg-si) with legi. For illustrations see Koby's 
 Latin Grammar, i. §§ 670-675. The termination -s-i is sup- 
 posed to ^es-i, a perfect formation from the stem -es, and 
 therefore analogous to the -tra of Greek weak (first) aorist stem. 
 Strictly speaking, this perfect is a ' weak ' or composite tense, 
 and (with the perfect in -vi or -ui, mentioned below) is some- 
 times classed separately under the head of 'Weak Perfect- 
 stem ;' numbers i and ii being the ' Strong Perfect-stem :' but 
 it seems more convenient to arrange all varieties of the Perfect- 
 stem together. 
 
 (iv.) Suffixing -u (-ui) to consonant-stems, or -v (-vi) tol-Sutaxing 
 vowel-stems, as with most regular verbs in a, I, amavi, au- 
 di-vi ; e stems with a few exceptions (aholevi, delevi, and 
 quievi, etc. from inchoative pres. quiesco) drop the final 
 e and form the perfect as though from a consonant 
 stem, mon(e)ui, ferb-ui, and from some a- stems are found 
 similar forms, e. g. crep-ui, cuh-ui (rarely crepa-vi, cuba-vi) ; 
 also from pres. -io, infin. -ire (J being dropped), aper-iti, 
 salui. 
 
 The perfect form in -vi, -ui, is found in a considerable class 
 of verbs with a Present-stem (see below, p. 204) increased by 
 n or sc, e. g. lino, livi, or levi ; sino, si-vi ; cre-sco, ere-vi. 
 In stemui from sterno, trivi from tera, the stem originally 
 
 
 
194 feri Injkction. [chap. 
 
 Latin Per- consonantal becomes a vowel-stem by metathesis of tbe rowel 
 lect-stem. •' 
 
 and r : pds-ui is from p8s-no, contracted, pono. 
 ''"' "^ ■"'■ In certain verbs wbose stems end in -u {acuo, arguo, tribuo, 
 statu-o, etc.) the -ui of the perfect arises from loss of v, u being 
 the stem-lrtter, e. g. statui:=statu-vi. In some other verbs the 
 apparent Jdentity of perfect- and present-stem may arise from 
 loss of reduplication {pamdi, verti, etc.). 
 
 N.B. The perfect-stem formed by suffixing v is frequently 
 modified by the omission of v in all forms except i and 3 sing, 
 and 1 plur. of perf indie, and the contraction of the vowels thus 
 brought together : e. g. amdsti, amdstis, amdrunf, amdram, 
 amdssem, amdsse. Sometimes however the vowels are not con- 
 tracted after loss of v, e. g. ie and sometimes ii, as avdieram, 
 audiero, audiisti, as well as avdiati ; so frequently from peto, 
 eo and their compounds. 
 
 Sometimes both forms of Compound Perfect, in -si and -ui 
 are combined in one verb, e. g. met-o, messui:=met-s-ui : nexui 
 ^nec-s-ui (stem Tiee-). 
 
 The ending -vi, -ui is generally recognised as ^ fu-i the 
 preterite of stem/u- (Sanskrit bhu- in bhav-armi=«a;i«io, orior; 
 3 sing. 2 aor. a-tohtl-t : Greek (j}v-a>, (fiv-Tevm, etc. ; fu-am, fu- 
 iurus, fo-rem, fore). The original bh represented by /"in fu-i, 
 etc. (p. 69), may have passed into h ; then hui would easily 
 lose its aspirate, and become -ui or -vi. Whatever the process, 
 it is evident that vi=ui and that v must not be considered 
 as representing the f of fui. The formation, then, of the 
 compound perfect in -vi is exactly analogous to that in -si; 
 a preterite form of stem/w- being used in one case, a preterite 
 of stem es- in the other. 
 
 Inflection of The terminations are the same for all four classes of perfect- 
 Perfect- .... 
 stem. stems, being distinguished throughout from the Greek perfect 
 
 by the characteristic vowel l^ (found in old Latin in all 
 
 ^ Corasen (TJeber Aussprache, etc., i. p. 609, 2nd edition), quotes from 
 the poets, fuit, rediit, vidU, dedii, stetlt, and many others. Laohmann, on 
 Lucr. iii. 1042, instances petiit, abut, redut, penlt, from various passages 
 in Ovid, and ' Italiam fatis petiit auctoribus,' from Virg. Aen. x. 67, where 
 however most MSS. and eitors read ' petiit fatis ;' and goes so far as to 
 maintain that, the final it being necessarily long, Virgil would not have 
 
Tin.] Verb Infleetion. 195 
 
 persons except i plur. and often written ei). In 2 sing, and Latin Per- 
 plur. we find a suffix is- {is-ti, is-tis), with which the er- of 3 
 plur. {er-unt=:es-unt) is identical ; cp. also the infinitive ter- 
 mination -is-se. These forms then point to a suffix Is as 
 characteristic of the perfect indicative, whose complete forms 
 would he fec-ls-in{i) (l&tei fec-t), 
 
 fec-ls-ti, 
 
 fec-ls-t, ( „ fe&,i), 
 
 fee-ls-mus, ( „ fecimus), 
 
 fec-^s-tis, 
 
 fec-is-o-nt-=:feoerwn,t. 
 s in Latin not unfrequently falls out before m and t ; this 
 would account for the later forms of i and 3 sing. ; and of i 
 plur. also, except that here the l is always short in poetry, and 
 no forms in ei have been preserved. We must therefore sup- 
 pose that in i plur. the tendency to shorten the penultima, 
 which is seen at work in 3 plur. tiderunt^, etc., and in the 
 forms of perf. subj. dederimus, etc. (where t is the characteristic 
 mood- sign) prevailed to such an extent at so early a period, 
 as altogether to obscure the original quantity. [In the case of 
 3 plur. the syncopated forma dedrot, dedro, dederunt, on old 
 Pisauran inscriptions ^, show the early prevalence of such a 
 tendency.] 
 
 Others (e.g. Schleicher, Comp. § 291) suppose two forms of 
 perfect-stem, one in Is the other in I, to account for the different 
 
 shortened it, but must have written, e.g. in G. ii. 8i, Aen. ii. 497, exit not 
 exnt, in Aen. v. 274 transit not transiit. Lachmann's extreme view, how- 
 ever, is repudiated by Munro on Lucr. 1. u., and Conington on Aen. ii. 497 ; 
 the former pointing out that Ovid is singular among the poets of his day 
 in lengthening the final it of perfects, which, though undoubtedly long 
 temp. Enaius, had come to be universally shortened like so many other 
 final sounds in Latin. 
 
 ' Virg. Eel. iv. 61 {tidirunt), Aen. ii. 774 {steth-unt). Miscwerunt in 
 Georg. ii. 129, iii. 283, may possibly be trisyllable {-cue by synizesis). 
 Lucretius frequently shortens the er ; Ennius not so often : and it is 
 probable. that this quantity was a later poetical licence with perhaps some 
 foundation in the tendencies of ordinary pronunciation. 
 
 ^ These inscriptions (chiefly votive, to female divinities) are given in 
 Wordsworth's ' Fragments and Specimens,' p. 167. On the marks of their 
 antiquity (not later than the Hannibalic war) see Mr, Wordsworth's 
 notes, p. 408. 
 
 03 
 
tg6 Verb In/lection. [chap. 
 
 fe°-t"' ^''^'''' P^i'sons of the perf. indie, but this seems hardly necessary. 
 The formation above noticed in is finds a parallel in certain 
 aorist formations in Sanskrit, e. g. from root vid, ' to know,' 
 sing, a-ved-im (Vedic), aved-is, a-ved-it : plur. a-ved-ish-ma, 
 a-ved-ish-ta, a-ved-ishus. Here Sanskrit has lost the inflectioi* 
 -ti from 2 sing, (as cp. with Latin is-ti), but in i plur. retains 
 the sufiBx (ish-ma cp. with i-mus) ; both are defective in 
 1 sing. 
 
 N.B. If this account be correct, the -ti (older ~t&i) of 2 sing. 
 is the only instance in which Latin retains the i of 2nd' person 
 pronoun (see p. 170) Another explanation of the perfect 
 forma (just alluded to), regarding i as the stem-ending (or 
 connecting vowel) throughout, makes the 2 sing, and plur/ 
 -sti, -stis analogous to the Greek 2 sing. -aOa; and accounts 
 for the 3 plur. -erunt as a composite form with es-onti 3 plur. 
 of sum (root es) analogous to L(Tairt=:ptd-aaim (p. 176) so that 
 dederunt=^dedi-sont (instead of ded-is-o-nt on the other view). 
 This view is plausible from its simplicity, and the harmony 
 between Latin and Greek forms which it conceives ; and, con- 
 sidering the obscurity in which the early history of grammatical 
 forms is really involved, it is perhaps unsafe to say that any 
 fairly plausible view is untrue. The other view, however, is 
 most approved by philologists, and is therefore given as pre- 
 sumably the truer. 
 
 2. 'Simple' or 'Strong' Aorist-stem \2nd Aor.\ 
 The Strong Aorist-stem exhibits, with few exceptions, the 
 stronR Pure Verbal-stem, sometimes reduplicated ; e. s. c-Xm-o-v from 
 
 Aoristsener- ^ t ■ i p i 
 
 raii.v=pure XtiVo), ayay-Hv from ay-a. It IS Only formed as a rule from 
 Verbal-Stem. , . , . , , , , . ,. . 
 
 verbs m which the pure verbal-stem is distinct from the 
 
 present-stem (enlarged), e. g. Mira, stem Xwr- ; (jteiya, stem 
 (ftvy- ; jSaXXo), stem /3aX- ; and but seldom from any but ' root- 
 verbs,' whose stems cannot be traced back further. Verbs 
 whose present-stem = pure verbal-stem, e. g. apx-a>, Xu-m, Xiy-a, 
 etc., form no strong aorist, because in these cases it would 
 coincide with the imperfect. With aya, however, the redupli- 
 cated form ^yayov avoids this confusion : and with some other 
 verbs the change of the vowel in the pure verbal-stem forms 
 
viii.] Verb Inflection. 197 
 
 a strong aorist distinct from the imperfect, e. g. rpeir-a, stronsr 
 
 jf Aorist- stem. 
 
 (Tpair-ov. 
 
 Grreek has two main classes of Strong Aorist forms, corres- 
 ponding to the two principal conjugations, (p. 169) : — 
 
 (i.) Without thematic-vowel, usually from vowel-stems *, e. g. 
 
 Act. i-Br)-v, i-Bfj-s, i-Sr) : i-Se-pLfV, e-Oe-re, i-Bt-aav (compound) 
 
 (but c/Sai/), 
 Mid. i-ek-mv, l-fleo-o, «-9c-to ; i-ei-ntea, i-Si-aBe, l-BevTo. 
 iBov, 
 
 So ffirjv, e(j)dr]v, crXi/v, eyvav, iSKav, ?(j>vv, etc. : and certain Epic 
 middle forms from consonant-stems without a connecting-vowel, 
 e. g. SKto, Sckto, X€k-to, jrnX-To, juk-to, Sipro ; XexSai, op-dai, hix-Oeu 
 (infin.) ; aKp,evos, appevos, Siypevos, and a<r-pevos Used adjectivally 
 {=a8-pevos). The imperatives Xe^o, 8e'|o, Spa-o are more pro- 
 bably weak aorist formations (see below) : Kf-K\v-6i, (ce-rXu-tc, 
 are examples of reduplicated forms. 
 
 (ii.) With thematic-vowel, as in ordinary conjugation : — 
 
 Act. i-\in-ov, i-Km-e-s, l-Mir-e, ) , ^ . 
 
 ,,^.j , , , , , , , ( and so on, as Imperfect. 
 
 -M-ia. t-AtTT-o-fiTjVf e-Aiir-c-ffo, e-Ktir-e-ro, ) 
 
 k\lTTOVf 
 
 To this belong most of the reduplicated forms, frequent in 
 Homer, e. g. ■neiriQ-ov, eearov:=iFeFeiT-ov (usually without augment 
 
 eiiTOv), im(j}paSov, r)vvna!rov, TjpvKaKov. Reduplication here probably Eedupli- 
 
 does not (as in the perfect-stem) express past or completed 
 action ; for this is expressed by the augment, and the combina- 
 tion of the two elements would not be analogous to that found 
 
 ' The original quantity of the root-vowel in some of the forms here 
 cited is a matter of some uncertainty. In i-Bi-rriv, Bsivai, 6k-ais, and 
 kindred forms, bi-do-Tai^ bovvai^ So-ais, etc., <pa~6t^ fpa-T6s, etc., the short 
 vowel of Be, So, (pa, appearing as it does in a majority of forms, is presu- ' 
 
 mably the primitive root-vowel. On the other hand, the Indian gram- 
 marians allow no roots in a, but only in 8. ; so that Sanskrit da answers to 
 Grreek So, Latin dare (but do-nuni), Sanskrit Sh& to Greek Be ; and in 
 forms like yvu-vax, yvaj-T6-s, yvw-ffis, etc., l3iat-T/ai, &\Qj-aLS, d\$i-vai, etc., 
 re-Tpoi-fiai, e-Tpii-Bi]V, etc., the long vowel extends to the greater n umber if 
 not to all the forms, and seems to be original. Schleicher maintains that & 
 is the primitive form in all Sanskrit roots ; and readers of his Compendium 
 will find this assumption there carried out. This view may or may not be 
 correct, but we have no data reaching far enough back into the history of 
 European speech to enable us to determine the question, 
 
198 
 
 Verb Inflection. 
 
 [chap. 
 
 strong 
 Aorist. 
 
 Traces of 
 Aorist in 
 Latin. 
 
 Present- 
 stem' in 
 relwtion to 
 Pure Verl al- 
 stem. 
 
 in the pluperfect. We must therefore look to other uses of 
 Reduplication, e. g. the expression of intensive meaning (above, 
 p. 52). Curtius (' Temp, und Modi,' pp. 150-164) enumerates 
 32 reduplicated aorists, in 7 of which {^kuxov, apope, fitSae, 
 \e\a6ov, \e\axov, ireirapeiv, KexaSmv) he traces a Causative mean- 
 ing ; in 1 1 ((cf(cXfTo, rivliraire, tpruKaKf, ^vaKJJOv, kikXvBi, XfXa/3f(r5at, 
 aiiiTCTraKaiv, itriirKrfyov, TeTapmirBai, iire<f)pa8c, Terayav) an intensive 
 meaning ; in 2 (irfTiKovro, ttijtIBoito) a special transitive sense ; 
 while in the remaining 10 {akoKKov, ^yayov, rjveyKov, KeKvB&a-i, 
 KexapovTO, XeKoKOVTO, fiejicaroifv, irecpiSoipriv, ctct/iov, i've(j>vov) no 
 
 special influence of the reduplication can be detected. In his 
 later treatise ('Das Verbum,' ii. pp. 21-32) he enumerates 41 
 reduplicated aorists ; but is more cautious in expressing an 
 opinion as to the exact force of the reduplication. Thus he 
 only mentions ek/kXct-o (cf. keXcto), k(kKv6i, (cf. k\v6i), ^yiWan-f, and 
 perhaps rerdyav (as cp. with tangere) as examples of intensive 
 force, while citing the same verbs as before for causative force. 
 From these Greek forms and a comparison of Sanskrit, in which 
 reduplicated aorists are formed almost entirely from verbs of 
 the, loth class (principally causatives), Curtius arrives at the 
 conclusion that in the reduplicated aorist the reduplication 
 (Verdoppelung) belongs not to the tense-formation but to word- 
 formation : and that its original import was to give an intensive 
 or causative meaning, irrespective of time. 
 
 Traces of an aorist formation in Latin are supposed to lie in 
 certain old forms, e. g. in tago, tagis, an old pres. form of tango 
 mentioned by Festus (Forcell. quotes Plaut. Asin. ii. 2. 106, 
 but the reading is doubtful) exhibiting a pure verbal-stem tag 
 {6i.y) beside pres. stem tang ; in jpagunt (XII Tab.) by present 
 pango, cp. i-iray-riv, irriym-fu ; and in parentes (=01 TtK-ovres), 
 beside farientes (01 TUT-o-vres). 
 
 3. Present-stem. 
 
 The Present-stem is (as has been already pointed out, p. 160) 
 in many cases different from the pure verbal-stem, by combina- 
 tion of which with the various suffixes of person, mood, and 
 tense, all the forms of the verb may be explained. Under the 
 heading 'Present-stem' is in fact included a series of morpho- 
 
VIII.] Verb Inflection. 199 
 
 logically distinct formations, each of which had oriffinally its Present- ' 
 
 ., . ..,. stem in 
 
 own special meaning (e. g. inchoative, intransitive, durative, Sanskrit. 
 passive, intensive, causative, desiderative, iterative) : but in 
 Greek and Latin, while a variety of forms remains, distinct 
 functions have disappeared, or survive only in a few special 
 cases (such, e. g. as the forms for inchoative and desiderative 
 verbs). By Sanskrit grammarians the special modifications of 
 roots to form the present-stem of verbs are taken as the basis 
 of a classification of verbs : and the ten ' conjugations ' of San- 
 skrit grammar are ten classes of verbs arranged according to 
 the formation out of roots of verbal-bases or stems, which then 
 receive a common scheme of terminations, in the four ' con- 
 jugational tenses ' (present, imperfect, potential, and imperative) 
 which alone are affected by the rules of stem-formation. In all 
 other tenses there is one general rule for forming the base or 
 stem of all verbs, i. e. in all except the four ' conjugational 
 tenses ' all Sanskrit verbs belong to one common conjugation. 
 For Greek and Latin grammar, in which no such elaborate 
 system of stem-formation and euphonic combination of stems 
 with inflections can be traced, the most practicable classification 
 of verbs (as of nouns) is found to be a purely phonetic classifica- 
 tion, according to the final letter of the stem (see pp. 167-9): 
 but in the various formations of the Present-stem we have the 
 outlines of a system akin to that of Sanskrit, which may to a 
 certain extent be made the basis of a classification of verbs 
 according to stem-formation, but without the corresponding 
 distinctions of meaning which give its point to such classifica- 
 tion. 
 
 The Present-stem is generally speaking an enlargement of formation 
 the Verbal-stem, either by strengthening this latter or making stem, 
 additions to it. For strengthening a root, language employs 
 two principal means, — Eeduplication, and Vowel-strengthening 
 (see above, pp. 51, 53); and to these may possibly be added a 
 third, viz. Nasalisation. The operation of these is seen in the 
 formation of ' Intensive ' Verbs in Greek, in which Eeduplica- 
 tion is often combined with Vowel-strengthening (e.g. vijv/ib, 
 7ra(7raXX<i>, Tronrvia) and Nasalisation (e. g. irafK^aivco, Panfiaivo), 
 
300 Terh Inflection.- [chap. 
 
 ciassifica- yoyyv^to, etc. The employment, separately, of these three means 
 Bent-atems. of stem-strengthening gives us three distinct classes of Present- 
 stem ^ : and if we take first (as probably earliest in order of 
 time) those verbs in which the Present-stem is identical with 
 the Verbal-stem, we shall thus have four classes of Present- 
 stem, viz. : — 
 
 1. Verbal stem unaltered — Xiy-ai, ypi.(p-u, ayai, cado, tego, etc. 
 
 2. „ „ reduplicated — Si-Soi/Ji, mn(f)Tai, iibo, sero ( = seeo). 
 
 3. Stem-vowel strengthened — ^tiya (<pvy-), \e'nra (Xht-) diico {due), fldo 
 
 (fides). 
 
 4. Nasalisation :— 
 
 (i) By insertion — eKiyx'^' <"piyy<»t tango, findo, fundo, etc. ; 
 
 (2) By addition — le&nvoi (xa/i-), iaieva : forms in -vv-iu and -av-a : 
 
 ster-n-o, sper-n-o, stern-u-o; 
 
 (3) -^y both these — \aii0-av-co (\aP-), imvO-Av-oi (imB-), etc. 
 
 To these may be added three more classes, viz. : — 
 
 5. Addition of « sound — tvjtt-o) (twit-), etc. pecio, flecto. 
 
 6. „ „ ya (ja) (pronominal ?), which appears, 
 (i) as simple i sound — /iTjvla, sal-io, 5oK-e-a ; 
 
 (2) in a diphthong — Sal-ai, iMto/iai, ipatva, Kreiva, etc. ; 
 
 (3) by assimilation into a double consonant. 
 
 7. Addition of sJi, 6k. (Tk, Lat. so (Inchoative and Iterative verbs). 
 
 Of these classes, it seems best to regard 2, 3, 4, and 5, as 
 showing a merely phonetic increase of the root ; 6 and 7 only 
 as formed by the addition of distinct (pronominal) stems. Some, 
 however (e. g. Schleicher and — ^though less positively — Curtius 
 in his latest work'') regard 3 and 5 as also exhibiting an 
 additional pronominal element {na, nu or n, ta or t). In the 
 absence, however, of data respecting the original development 
 of these forms we must regard this as an open question : all 
 that we can say is that, e. g. tvttto (rvTrri), and Sclkw are like 
 ayo {aye), fully developed, possibly nominal, stems as far back as 
 we can trace the growth of language. 
 
 ' It sometimes happens that two or more of these methods are employed 
 in forming from the same stem verbs. of a kindred signiiication, e.g epvy- 
 yavoj, ipevyu — TU7X(ii'a;, revxoi — TTwOdvofJiai, TreiuOopai — XavOdvoi, A.i^0ai, etc., 
 etc. See Curtius, ' Tempera und Modi,' p. 81. 
 
 ' 'Tempora und Modi," pp. 67-123 : 'Das Verbum,' I. pp. 199-392. 
 
VIII.] Yerh Inflection. 201 
 
 The Present-stem receives in all cases the primary form of 
 the Person-endings : and under each of classes 1-4 fall Greek 
 verbs of both principal conjugations (p. 169 ), affixing the 
 terminations to the stem with or vithout the addition of a 
 thematic vowel, e. g : — 
 
 1. (Unaltered) \iyiii (Xiy-o-pu) and ^a-iti. 
 
 2. (Reduplicated) 7riir(e)T-iu (stem ttct-) and ri-Brj-iu, 
 
 3. (Vowel raised) 7rei6-a> (stem m9-) and «-/ii (stem i-). 
 
 4. (Nasalised) mrvd-a (stem irer-) and ■mr-i.v-vv-ni. 
 
 Verbs of the remaining three classes (5, 6, 7) belong almost 
 entirely to the ordinary or -m conjugation, characterised by the 
 thematic vowel. In Latin the other or -^t conjugation is almost 
 lost, except in isolated forms like es-t, vol-t, fer-t, i-mus : and 
 the ' thematic ' or ' connecting vowel ' characterises all Latin 
 conjugation. 
 
 I proceed to examine the different classes of Present-stem Formation 
 rather more in detail, following mainly the remarks of Curtius stem. 
 (' Temp, und Modi,' and ' Das Verbum,' as above). 
 
 1. Curtius ('Temp, und Modi,' p. 74) suggests that among the 1. Verbal- 
 ' unaltered ' present-stems should be included verbs whose stems altered. 
 have undergone ' strengthening,' but in which the strengthened 
 
 form has become stereotyped so to speak as the only existing or 
 traceable form, and the unstrengthened form is quite obscured, 
 e. g. yeia, 8eva, \ei(jiai, afiei^o/im ; and (with nasal) jungo, pre- 
 hendo, scando, incendo ; also disco, dico, ftdo (on the ground 
 that their stem-vowel is only lengthened, not increased). He 
 allows however that philologically these forms may be assigned 
 to the 3rd and 4th classes respectively : and it seems to be 
 a needless hair-splitting not so to class them. 
 
 2. Keduplicated Present-stems are rare in Latin, which (as 2. EedupU- 
 we have already seen in the case of the Perfect-stem) has sent-stems. 
 retained this primitive method of strengthening but little : 
 
 it seems that gigno (gi-gen-o), si-sto {=l-<Trri-)u), sero=se-so 
 (stem sa- in sa-tv/m), and bi-bo are the only certain examples : 
 Schleicher (§ 295) adds sido=sis-do:=si-sedo, from root sed- in 
 
 In Greek the vowel of the reduplicated syllable is generally 
 
3oa Verb Inflection. [chap. 
 
 T'ormation i (not € as ia perfect-stem, p. i88), e.g. 8t-8p- (80-), X-aTa-=. 
 
 stem. ' o-i-o-TO, ri'-^c (root 6e-, on change to r see p. 50) : 'l-ri-jur:=yi-ya-mi : 
 SlCriiJiat=8iSyrifiat (by assimilation, p. 76). Compare also KL-xpni^i 
 (xpu); and S1-81) (II. xi. 105), Si-Sevrau (Od. xii. 54), imperat. 
 from a stem Si-Si; (root 8e- of Se-a) ; pi-^as part. pres. stem 0a : 
 m/i-n^rj-fii and TTifi.'TTpfriiu {jtKa- and npa-) introduce a nasal into 
 the reduplication. In these forms the final vowel of present- 
 stem is often raised in the singular only, e. g. 8iSa-fu but 81S0- 
 fjLcv, Irj-iii but h-jiev : compare Sanskrit da-da-ml, dad-mas, where 
 d is lost. 
 
 In the ordinary conjugation we have yiy{e)va, win(e)Ta>, filiwa 
 =iu-fi,ev-a>, to which Schleicher adds "tC<'>=^'^^i/<^ (p- 76) = ie8y(»= 
 si-sedyo, from root e8=sed, see above on Latin sido. [But it 
 would be simpler to rank if <o =18^0) in class 6 with suffix ya (ja): 
 for even if sido be rightly explained as above, it is not necessary 
 to assume a precisely similar development in Greek from the 
 
 same root cS=:«e(Z.] TiyvaxTKo), Si-SpdaKio, TiTV(rKOfw.i, irKJyava-Ka) 
 
 come also under class 7, being formed by addition of ctk. In 
 the intensive forms TrmiraWa, SmBaX\a>, Ttomvoio, SeiBicrKoiuu, etc., 
 the reduplicated syllable is intensified, no doubt as being the 
 significant part of the word : but as the consciousness of the 
 meaning conveyed by it was lost, emphasis was no longer laid 
 on that syllable. On the contrary, it became weakened ; and 
 what was originally a formative element became merely me- 
 chanical, the intensive or frequentative or desiderative force 
 disappearing altogether. Thus lu-fieo-pjii (root ma-, p.€-, in 
 me-t-ior, etc.) originally='I frequently measure myself,' i. e by 
 some one, and so ' copy,' ' imitate,' has entirely lost its fre- 
 quentative force. Latin imifor, imago are possibly weakened 
 forms of mi-mi-tor, mi-ma-go, formed on the same principle 
 from the same root. • 
 3. Vowel of 3. The vowel of the verbal-stem or root is raised irregularly 
 in the pres. indie, of some primitive verbs, e. g. ft-fu, els, el, ela-i ; 
 but t/ifK, i-re (stem t) : <^ij-/i«, stem (pa-. The Latin stem i- is 
 raised to i in is, it, imus, itis ; but So, e-u-nt- 
 
 In the ordinary conjugation of Greek verbs the raising is 
 more regular throughout the present-stem, the unstrengthened 
 
 Verbal-stem 
 
VIII.] Verb Inflection. 303 
 
 form being often visible in 2 aor. (see above, p. 196), e.g. 06uy-i» Formation 
 
 { fk \ \ ' A\%'/i/Nyi\ ' / \ / / ofPresent- 
 
 y<pvy-), KuTT-a {j\.m-), Kriaa \Ka6-j, TrjKia (tok-), rpmy-a (rpay-) , stem. 
 
 ir\ma, the two last having the second stage of intensification (see 
 PP' 53) 54). Certain verbs in -tm from stems in v have had the 
 stem raised to eu, but the v of the stem has passed into the con- 
 sonantal sound F, and has thus been lost in present-stem (as in 
 gen. y\vKe-osz=y\vKeF-os, see p. 119), remaining as v before a 
 consonant in other parts of the verb. Thus pe-a)=p(F-a>, pev- 
 (iop.ai (stem pv- in ippv-rjv) ; compare also wXe-o), x^-'") ''■«-(«. 
 
 Curtius arranges the verbs under this head in two divisions ; 
 (a) those which exhibit completed strengthening by an addition 
 of vowel sound, i. e. fi, ev from i, v ; as in dXfiKJia (cp. nKr/Kicj^a, 
 Xitt' e\alco, Xlirapos, etc.) ; fISofuu (root FtS) ; neidto (f-mO-ov) ; 
 Kfidai {kv6i aor., KeKvBatri); irevBopai (Homeric, cp. TivB-eadai etc.); 
 pea, x™ = <rf)ef<», X^'fo) (cp. i-ppvrj-v, e-xv-TO, etc.): (6) those in 
 which the strengthening only appears in the increase of quantity 
 of the stem-vowel, e. g. a to u or ij (a being by its nature in- 
 capable of receiving additional vowel sound, see p. 53), as in 
 \rjda (\ad-), TjjKa (tok-) ; or i, li to i, v (instead of to ei, ev), 
 as in rjSopai (root erf aS of avSava, cp. eaS-oi/, etc.), X^^o) (Xe-'ka6-ov), 
 
 Tpi^-a (e-TpL^-rjv), (j)pvya> {(-(jjpvy-riv) ^. This simple increase of 
 quantity is all that is exhibited by the Latin present-stems 
 which fall under this head — the weakness of the Latin vowel- 
 system having all but extinguished diphthongs and made a full 
 increase, such as from i, v to ft, ev, impossible. Dlco (root of 
 in-dia-are, Greek SU-rf) and fidq {fides) are analogous forms 
 to rplfico {rpiffri) : but the change was probably much more 
 formal and meaningless to the Romans than to the Greeks, 
 who seem to have retained some consciousness of its purpose. 
 
 Sanskrit exhibits the same processes as Greek. Thus — 
 
 i of root is raised to e=ai : sidh, sedh&mi (cp. XiTr, XeiVo)). 
 
 u „ „ „ to 6=au : ush, osh&mi ('bum,' cp. tpvy, 
 cjxvymy 
 
 i is not raised to 1 as in Greek Ixa : but u is sotoetimes raised 
 to ti, e. g. giihami ('veil,' cp. KeiSa and Zend gaozaiti). 
 
 ' Curtius ('Das Verbum,' I. pp. 218-226) enumerates 58 Greek verbs 
 under this cl»ss, giving to its two subdivisions the titles ' Diphtbongische ' 
 and ' Mouophthongische Zulaut.' 
 
204 
 
 Verb Inflection. 
 
 [chap. 
 
 4. Nasal 4. The different results of the principle of Nasalisation in 
 
 sound in- .•■/.. 
 
 sertod. the formation of Present-stems may be thus arranged : — 
 
 (i) Nasal introduced into the body of the root, chiefly in 
 Latin, e.g. tcmgo (old form tago, p. 198), pango (older pago), 
 fr'ango {fractus,fragor),fingo {fig-i), linguo, tundo,jungo {jug- 
 um), etc., etc. This, the simplest kind of Nasalisation, is common 
 to Latin and Sanskrit, but almost unknown in Greek ; tr^lyya 
 {(TiftLy-iws), cXcyxco being perhaps the only cases where it alone 
 is employed, though it is combined with a nasal syllable (no. 3) 
 in a good many stems, such as Xafi^-dv-a, dtyy-dv-a (Xa(3-, ^ly-), 
 see below. 
 Appended. (2) Nasal appended to the root : — 
 
 (a) After vowels — iriv-a>, nv-a, <l)6lv-a>, (jiddva, SivcD as Compared 
 with emov, Tia, e<j)di-To, (jiSd-jjievos, Sv-ca. The roots yev, rev, fiev, 
 4>ev (in yi-yov-a, rov-os, lii-fiov-a, ^oV-os) are perhaps nasalised 
 forms of still older roots which appear in the forms yi-ya-a, 
 ra-Tos, fii-jioa, Tri-^a-fuu. In Kplvco and KKlvto the nasal passes 
 into other tenses also. 
 
 (6) After consonants — Kdjiv-a (e-Ka^-ov), hdK-va, and Tiiivu, ; 
 spemo, temno, pono-=posno, posin-o (positus). 
 
 (3) Addition of nasal syllables — ve, va, vrj, w, and av, e. g. 
 iKve-Ofiai, Kvve-a, olxvi-a j Kipvd-a, jriT-vd-a, SeiKa-vd-o-jjuu (icepdv- 
 Wfii, irerdv-wfu, SiUw-iu) ; ^eiyvv-fu, pr^y-vvfu, oWvfu^oXwiii, 
 a-KiSi/rj/ii, KLpvriiii • 'iKav-a, aii^dva, afutprdva ; and (with inserted 
 nasal also) Xapfi-dvai, 6i,yydvai, -jfavhavta, etc. 
 
 Schleicher (Comp. § 293) regards these nasal syllables as 
 pronominal additions. Curtius, on the other hand, considers 
 them as purely phonetic additions growing out of the simple 
 nasal sounds inserted or suffixed to produce a greater fulness 
 of tone, analogous to the intensification of vowels. According 
 to him, therefore, the Latin forms pa-n-go, etc., in division i, 
 into which the nasal enters only as an extension of consonantal 
 sound, are more ancient than the forms in w-ju, etc., common 
 in Greek, where the nasal combined with a vowel forms a dis- 
 tinct syllable. [See 'Tempera und Modi,' pp. 63-66, where 
 the phonetic character of these nasal additions is elaborately 
 illustrated by analogies from Sanskrit; 'Das Verbum,' I. pp. 240- 
 263 ; and compare above, chap. iv. p. 55.] 
 
 Addition of 
 r.asal sylla- 
 bles. 
 
VIII.] Verb Infieetion. 305 
 
 5. The strengthening of the verbal-stem by addition of the s- Addition 
 dental tenuis t is chiefly found in Greek : e. g. in two verbs 
 
 only after a vowel, viz. avvra and apina (Attic for avva, apiai) ; , 
 in two after a guttural, viz. ireKra, beside weUm and ttckm, and 
 TiKTa (stem TfK-) ; and often after labials, and <p being changed 
 by assimilation to n (p. 74), e. g. ^Xdirr-a (/3Xd0-7), Epic 0Xa0- 
 
 frai), Ka\vwT-a> (^KoKiPrj), tvitt-w (e-run-ovj, iplirr-io (later form for 
 
 epe0-(o), daiTT-a (rdcfi-os), etc. The only analogous forms in Latin 
 are pect-o, flect-o, nect-o, {nexui:=nec-s-ui), plect-o (wXeK-ia). 
 Schleicher regards the < as a pronominal stem ta ; but it is 
 more probably a purely phonetic increase of sound, as e. g. in 
 tttoXls, KTciva, TTTokeiios, beside 7r<iAir, Kaivm, irokejios ; compare 
 
 iiTtT-io-s from stem in-, Latin sup-. 
 
 6. The insertion oi ya {ja) between stem and person-ending, 8- Addition 
 which is the characteristic of the fourth class of verbs (chiefly 
 intransitive), and also of the passive conjugation (see p. 178) in 
 Sanskrit, appears in many Greek and Latin verbs. The y (J) 
 
 sound seems to have been uncongenial to Greek organs of speech; 
 
 accordingly it is, generally speaking, either vocalised into i or 
 
 passes by assimilation into some other sound. The forms which 
 
 it assumes may be thus arranged : — 
 
 (i) y (j) sound appears as a vowel : in Greek: 
 
 (a) As I, in JSi'co (Sanskrit svidya.mi, compare i&pa)s = (rFcS-pa>s), 
 
 firjv-i-o) (root IMV-), ead-i-m (Horn, ead-a, strengthened from cS-m). 
 
 The I is sometimes long (jufvUv, II. ii. 769 : compare Aesch. 
 
 Eum. loi) : so that perhaps these forms should be reckoned 
 
 parallel to Latin audire, etc. («'=*/, -see below). 
 
 (6) As e, in doK-i-m, ■ya/i-e'-o), Kvp-e-a (xup-o), Kvp-crm), TcaT-iofiat 
 (i-iraiT-apjfv, ttckttos), (jiiKeai (Epic <JHKai and i(j}l\aToj, irrvyeai 
 (e-a-Tvy-ov), the -eco of these verbs, which in other forms exhibit 
 a shorter stem, being different from the -ea of ordinary deriva- 
 tive verbs (see Appendix B to oh. v. p. 103), though probably 
 the distinction was forgotten. 
 
 (11) The y {j) sound (vocalised into i) appears in a diph-y.OJastina 
 thong : 
 
 (a) Combined with the final vowel of a stem- — Saia, root 
 da, ' divide,' or du {dah), ' burn,' (so i-ba-rj), fiai-ofuu (jua-o-ojuot, 
 
2o6 Verb Injledion. [chap. 
 
 ^j?™"*!™ efiturdiujv), vala {tvaatra), Kala (Attic Kaa, fut. Kava-<a, stem Kafr), 
 
 stem by yo K\aia, mnia {omi-trm). 
 
 (6) Thrown back within tte stem and combined with its 
 vowel (as e. g. in dixeiva>v=:dnepiav, /ifXmva^ /ieXavia, aarfipa^ 
 aarepia, and many others) : thus ^alva-=.(^av-jti> {i-^dv-rjv), KTeiva 
 :=KTfvj(o (e-KTov-a), fiaivo/uu (another formation from root i^av-, 
 see fujv-l-a above : and so with many verbs ending in -fiaiva, 
 derivatives from nouns in -/xa(T)=an older -/uaf, e. g. ovoiiaivio, 
 hufiaaim, Bavfialva ; TeKfiaip-oiiai (^TCKnap), Kadaipa (itaBapos), ifieipa 
 (i/ifpos) ; aipw, fipco (Latin sero), cjidetfia (^e(j>0dp-Tjv), xalpa (J^dp-yv), 
 Kpiva {KpXv-a, fut.). 
 
 y(j)as«)n- ^jjj) The y (j) sound passes into a double consonant by 
 
 assimilation, asginiilation (see above, p. 75) : 
 
 (a) By pure assimilation from X; to XX, e. g. SKKoimi (Latin 
 
 sal-i-o), oreXXo) (f'-CTToX-iji'), <T^dK\ii)z=(r(j)aKj<o (e-tr^aX-iyv), ;3aXXm 
 
 (e-^aX-ov), o(l>£\Xa)^6(f>iXja, which also passes into oc^eiXm (11. 6, 
 above). See Curtius, ' Das Verbum,' I. pp. 300-303. 
 
 (6) From kJ, xji yjj Vt ^j to tra, e. g. (pvXda-a-a {(pyXaK-ja), 
 Tapdaa-a (rapd^-jo, compare Tapax-rj), dXXd(ro-a> (aWdy-ja, compare 
 dKKay-rfj, XtVorojxai (Xirjo-pai, compare Xtr-^), KopvtrvcD {Kopi6j-a, 
 Kopvd-os). The process of change in these cases has already 
 been described, ch. iv. p. 75. Full lists of forms in illustra- 
 ,tion are given by Curtius ('Das Verbum,' I. pp. 311-317). 
 In noun forms we may compare ^<ra-av=rjKJav (^k-iotoj), iXda-trwv 
 =^eKdxj(ov (eKdx-iTTos), KiKiacra^KiKiya, ;(apif(r(Ta=;(apte>T'^a : the 
 two latter showing the feminine Bv&xja (ya), which in /jteKaiva, 
 a-wTeipa, etc., noticed above, passes back into the stem as the 
 i sound of a diphthong (11. 6). 
 
 (c) From 8/ (and sometimes yj) to f: e.g. €^op.ai {e8jop,at, root 
 e'S- of 16-05, sedes), Sfa (oS-mSa), (fipd^ia {^(ppaS-ov), <rxi'f<a (root 
 o-^tS in o-X'S"?), tXufo) (/cXuSmk) : also xpd^a {^^^Kpdyja, cp. KE- 
 Kpay-a), aTa^to {aray-iiv), pl^a (e/)e|a=epfyo-a, pexBf, etc.), c'XeXi- 
 ffti/ {l\c\ix-6t]). 
 j (f) in Latin In Latin the i sound remains, e. s. in verbs in -io of qrd con- 
 
 Pressnt- . ,. i ^ ^ / ■ \ 
 
 stem. jugation before and u {capi-o, ccvpi-unt) and the conjunctive a 
 
 (capi-ant), and so called fut. indie, e (ca^ii-ent). Where the 
 thematic vowel becomes i, the two coalesce {cajpis, capit, 
 
Till. J Verb Inflection. 307 
 
 capimus, capitis). We should have expected I throughout : Fonnation 
 
 1 - T - , n J, • • , .-i.- of Present- 
 
 and capis ana captt were probably the original quantities ; stem by ya 
 capimus and capitis arising partly from analogy with capio, 
 capiuni, partly from the general tendency to shorten an un- 
 accented syllable, aided by the desire to distinguish the forms 
 from those of the 4th conjugation of derivative stems in l 
 (avdvmus, auditis, etc.). Other Latin verbs of this class are, 
 e.g. jaci-o {jac-tus), fodi-o {/ossus=fod-tus), fugi-o [fug-i), 
 aio:=agj-o (root ag in ad-ag-ium), etc. etc. In ero^esio, 
 a present formation from the root es (compare ea-ofuu for 
 i(r-cryofiai), the i sound has disappeared before and u. 
 Assimilation analogous to that observed in Greek is possibly 
 seen in pello (ttoXXcb, pe-pul-i)=pel-jo, percello, tollo (te-tul-i), 
 vello, fallo, curro, etc. But this kind of assimilation is not 
 familiar to Latin, which e. g. keeps sali-o beside Greek aWoficu, 
 alius beside aXXor, medius beside iiea-a-ns, melior beside imWov : 
 and it is possible that in these and similar forms with U, rr, we 
 have a doubling and so increase of the consonantal sound with 
 the same object as that of the vowel increase in verbs of class 3 
 (p. 202). 
 
 [N.B. In this class have been included only those verbs 'Derivative' 
 
 ■- . . . Verbs with 
 
 in which the suffix ya ( ja) appears to have been used in the for- suffix^a (t/a) 
 
 V^ , , , ■ , , . ™t includ- 
 
 mation 01 the present-stem from a verbal-stem, which latter is ed under this 
 traceable in other forms of the verb. From these must be sent stem. 
 distinguished a common formation of derivative verb-stems 
 by the addition of the same suffix ja (j) to nominal stems, 
 in the conjugation of which the i sound (or its effects) is 
 retained throughout all tenses ^. These are in Greek the ' con- 
 tracted' verbs in -aa, -e<o, -om (from an original -aja, -eja, -y<i>= 
 Sanskrit -ayami, the regular termination of one class of verbs 
 (loth) in Sanskrit) from which the y {j) sound has dropped, 
 
 e.g. Tifiaa^=Tiiiiaja>, from noun-stem rifia-; <j)opea>=:^opeja>, from 
 
 stem (jiope- (0opp-) ; 6p66a=:6p66j-a, from stem 6pdo-. Correspond- 
 ing formations in Latin are the ordinary ist and 2nd conjuga- 
 tions, and verbs in u-o of the 3rd, e.g. amo=.amao, from arniajo; 
 
 ' See Appendix B to Chap. v. p. 103. 
 
2o8 Verh Inflection. [chap. 
 
 derivative moneo, from monejo ; statuo, from statujo : the a, e of the ist 
 
 T CJTDS Willi • • ^ » 
 
 iufflxja. , and 2nd being the result of combination with the connectmg- 
 vowel, as in the contracted forms Tiii.a>nev=Ttf).a.-o-nev, (jiopovfievz= 
 <i>op4-o-ficv; while in the -no forms ( = -00)), the vowels remain 
 uncombined {statuis, statui-mus), except in the supine stem 
 {stat'utum=statu-i-ttcm). Greek verbs in -ia> where t remains 
 through all tense-stems, e.g. ISia (root 28), fU]vi(o, KovLa=:Kovtj-<o, 
 seem parallel to Latin 4th conjugation forms in -io as compared 
 with 3rd conjugation cuplo, etc. : and the long i found in some 
 of them {kovTco, iirjvla) shows a contraction of i sound with 
 another vowel. The terminations -a^io, -ofia, -i^a probably con- 
 tain the suffix ya (j) assimilated (see p. 76) : and in -atva, -vva 
 (Xejati/tt), Bapa-iva) the i sound is thrown back into the stem as 
 above in 11 6. Latin desiderative forms in -tur-io are formed 
 from nominal stems in -tor by addition oija («'), e. g. partur-i-o 
 (jpartor), esurio = ed-turio.J 
 
 7. Addition 7.' The verb forms in -o-kib, -sco ('Inchoative' verbs) ^ are 
 
 ot<rK-(sc.). . „ . . , ' ... 
 
 especially interesting because we can in this case prove a par- 
 ticular meaning for the additional element in the present-stem, 
 such special meaning having (as already pointed out) been lost 
 sight of in the other forms that have been discussed. The 
 Inchoative (or Inceptive) meaning is obvious in many verbs 
 both Greek and Latin (especially the latter), e. g. •yTjpd-o-K-o) 
 (cp. sene-sc-o), t)fia-a-K-ai (pube-sc-o), dvafiia-a-K-0-p.m (revivi-sc-o) ; 
 and can be traced in many others, e.g. p,i-pvfi-a-K-(o {re-min-i-scor), 
 aKh-fj-ijK.-ci> (cp. adole-SC-o), yi-yva-tTK-a (=gnoseo), and fii-fia-o-K-o), 
 ' I make to learn,' which is the correlative (with causal sense) 
 of di-sc-o, ' I learn.' In other forms (e. g. jSXmcrKa), 6paia-Ka, 
 paciscor, ulciscor) there is no historical trace of the meaning. 
 The ' Iterative ' forms of imperf. and aor. in -o-kov, common in 
 Homer, are an isolated preterite of this formation of the present, 
 e. g. ex^-a-K-ov, Kf-o-K-oK, fiev-e-a-K-ov, etc. Curtius (Elucidations, 
 pp. 142, 3) explains the connection between the two thus: — 
 The Inchoative meaning consists essentially in the fact that the 
 action comes to pass gradually; and the gradual realization 
 
 ' See Curtius, ' Elucidations,' pp. 141-144. 
 
■^iii.] Verb Inflection. 309 
 
 (which language originally intended to denote by these present- formation 
 forms) and the repetition of an action were regarded by Ian- stem by o-k- 
 guage as nearly akin. Hence these iterative forms in -o-kov are 
 the opposite to the sudden ' momentary ' action of the aorist. 
 
 The forms in -ana, -sco are also interesting as showing Connection 
 the especially close connection between the Greek and Latin Greek and 
 branches of the Indo-European family. Sanskrit has something ative forms. 
 like it in the addition to a very few verbs of hh, the regular 
 representative of sh in Indian languages ' : but there is no trace 
 of that specific meaning of the additional element which in the 
 two classical languages is retained to so great an extent as to 
 give the name ' Inchoative ' to the class of verbs. The mode 
 of adding the ok-, so-, is also very similar in the two languages : 
 ' We need only compare (g)no-sc-o, {g)na-sc-or, cre-sc-o with 
 
 yi-yva-a-K-a, iTi-Trpd-a-K-a), Ki-K\r]-crK-<o, the derivative fifid-tTK-a, 
 yrjpa-aK-a vrith Latin ira-sc-or ; oK-i-a-K-o-iim, trrep-i-aK-a with 
 Latin ap-i-sc-or, pac-i-sc-or ; and SiSa-o-K-ai, Xa-o-K-w, in which 
 a guttural (i. e. of stem SiSa^-, Xqk-) is lost with disco (cp. 
 doc-eo), to perceive that the laws of formation are the same.' 
 Both languages unite the Inchoative element to a consonantal 
 stem by the intervention of a thematic (connecting) vowel (1, i, 
 or e) ; but whether Ctirtius' statement ' that the genius of lan- 
 guage, which is ever intent on delicate distinctions, has separated 
 the Iterative forms from the Inchoatives, at least in part by 
 the connecting-vowel,' is sufficiently borne out by the evidence, 
 may be doubted. The thematic or connecting vowel seems to 
 be merely euphonic (p. 166); and though language sometimes 
 avails itself of purely euphonic differences to express differences- 
 of meaning (see above, p. 36), there is no proof that it has done 
 so here. 
 
 The origin of the element o-k-, sc- is unknown. 
 
 Imperfect (Greek). 
 
 Formed from the present-stem by prefixing the augment, 
 with secondary person-endings : e. g. (a) with connecting-vowel, 
 
 ' Thus from root gam (go) is formed gsJchhS,-mi ; from ish (wish) 
 ikTcJiSriai ; from yam (restrain) ya&iAS-mi. These forms stand for 
 gaskami, etc. : Mh representing sk, na in i:hhS,y& (shade), Greek axta. 
 
3IO, Verb Inflection. [chap, 
 
 ^°™**J°" E-i^fp-o-i-, ?-(j)ep-es, etc., 3 plur. e-<f,ep-o-v=e(l>epovT : (b) without 
 ia Greek, connecting-vowel, indri-v, -ri-sii), -i/(t), (stem-vowel raised in 
 singular), i plur. e-Tide-fiev, 3 plur. e-riBe-o-av (a compound 
 formation, see p. 176). From elpi are found two forms of 
 imperfect, (a) eov=ecr-o-u, with connecting- vowel and augment 
 omitted ; (6) ^v=:^a--v with the augment and with n- dropped ; 
 or, with V also dropped, ^. Sanskrit forms from the correspond- 
 ing stem as- a ist preterite as-a-m^a-as-a-m, the vowel a 
 being appended to the stem to make the inflection easier. This 
 appears in another form of i sing, imperf. from fi/x« (ftr-fii), viz. 
 ^a=:^o-a (Ionia ea^ without augment), and in 3 plur. ^aav^kaant, 
 or ?<rav^erant ; in Latin eram=^esam (=:asani without aug- 
 ment); and finally in the weak (i aor.) termination -a-a (see 
 next page). Putting these forms together, we have — 
 
 1 Sing, ftsa-m, ^a=^ffa(/i), 
 
 2 „ asi-s, 
 
 3 „ asi-t, §t = ^(r-€, 
 
 1 Plur. fts-ma, 
 
 2 „ as-ta, 
 
 3 „ fl,san(t), ^aav(T) or l(rav(T) = erant. 
 
 A similar formation in Latin from the stem fu-, i. e. fu-am, 
 is generally supposed to survive in the termination -bam of the 
 Latin composite imperfect (see below, p. 220). The length of 
 d throughout iu erdm, fuam is a fact of which there is no 
 explanation. In Sanskrit the vowel a is always raised to a 
 before m or v of the person-endings (e. g. bhar-ft-mi, bhar-ft-vaa, 
 bhar-a,-mas of ist sing, dual and plur. : but bhara-si, bhara-ti, 
 etc., throughout the rest of the pres. indicative) ; and traces 
 of this (the reason for which is unknown) may remain in eram, 
 fudmus, extended by analogy to the whole conjugation of the 
 tense. 
 Eelatiotiof 4. 27ie Weak or Compound Aorist (i aor.). 
 
 Weak to . -^ . ^ ' 
 
 Strouj? I. The function of this tense is the same as that of the 
 
 Strong Aorist, viz. the expression of momentary action in past 
 time. But whereas the Strong Aorist is formed in general 
 only from verbs which form a present-stem distinct from the 
 pure verbal-stem (see p. 199), the Weak Aorist is formed from 
 
 [?"(=? 
 
 -ia-v). 
 
 era-in = eaam, 
 
 ^a-6a. 
 
 
 era-a. 
 
 ?iy, 
 
 
 era-t, 
 
 ?jius. 
 
 
 era-mus. 
 
 ^"], 
 
 
 era-tis. 
 
 A-orist. 
 
Vin.] Yeri Inflection. 31 1 
 
 all verbs whose present-stem is the same as the pure verbal- l'o™«>t'on 
 
 , „ , . '^ of Weak 
 
 stem (e.g. apx"") ^eyto, ypd<j)a), or a nominal-stem increased byAorist. 
 
 j (e. g. cXTTtfo), (jivKda-a-a, rekea, see p. 205). Comparatively few 
 verbs, e.g. those with vowel-raising or dental suffix in the 
 present-stem (abovej pp. 202, 205), as jreiBa, Xein-m, tottoj, and 
 several verbs in -fu, have both forms of aorist : and in some 
 cases where both forms are found, they are used to denote an 
 intransitive or neuter, and a transitive or active meaning re- 
 spectively, e. g. eo-Ti/v, ea-Trja-a ; f^rjv, e^-qa-a. Many verbs, how- 
 ever, along with the weak aorist form, exhibit a so-called 
 2 aorist pass, formed from the simple or strong aorist-stem 
 
 with the addition of e, rj (see p. 221): e.g. (eiymfU, e^ev^a, 
 
 e^iyriv: ^XaiTTO), efi\a^a, f/SAa/Sijv. In the later periods of the 
 language the newer form, i. e. the weak or compound aorist, the 
 use of which had become widely extended with verbs from 
 which it was impossible to form a simple aorist (e. g. the large 
 class of derivative verbs in -aa, -ca, -oa, -eva, -ifm, -afm, -aiva, 
 -vva, etc.), appears to have superseded the older form, even 
 where the conditions for a strong aorist formation were found, 
 
 e. g. 0\dnT(i>, E;3Xa\^a, but not ej3Xn/3oi'. 
 
 2. Formation of the Weak Aorist ^ The ist preterite of the Formation 
 
 of Weak 
 
 verbal stem as- (dsam, dsts, dstt, see above, under head of Aoristfrom 
 Imperfect, p. 210) is added to the pure verbal-stem like an stem o»-. 
 auxiliary verb. The initial a of as disappears as in Sanskrit 
 (a)sm.as, Latin (e)sum; and in i sing, the nasal /x or k falls 
 away, as it does in ace. sing. jrdSa^padam, pedem. The aug- 
 ment is prefixed, as in strong aorist and imperfect. Thus e. g. 
 e-SfiK-o-a (usually written c8fi|a) corresponds exactly to Sanskrit 
 a-dii-sham (sh here euphoniae gratia for s) ; the retention of 
 
 ' The characteristic of this formation being the letter a, it is sometimes 
 called the ' sigmatio' aorist. This element s (<t), representing the root as 
 (€<r-) of the substantive verb, enters into verb-formation in various ways : 
 (I) in single person-endings, auoli as Greek eSo-ffac (p. 176) ; (2) in a pre- 
 terite or aorist form — the 'weak' aorist in -aa; (3) in a perfect stem in 
 Latin -si (p. 193) ; (4) in the pluperfect of Greek and Latin, ■^Si-{ix)a, vide- 
 ram. (p. 219); (5) in 'futurum exaotum,' \iXv-aoimi, sohe-ro (p. 223); 
 (6) in the Greek future in -(rw = s + ya (p. 213), and the Latin future 
 formations in -so, -sim, -sere (p. 217); (7) in the Latin subjunctive forms 
 lege-rem, lege-rim, legi-ssem (p. 223) ; (8) in desiderative formations, e.g. 
 Sanskrit pipl-slia-ti, ' he wishes to drink,' cp. Greek Spaaeia, Latin vi-so. 
 
213 Verb Inflection. > [chap. 
 
 Formation the full vowel sound a involving the loss of the final nasal, 
 
 ofWeak ,.,.,.,, .... , , 
 
 Aorist. which IS retained where original a is weakened to o in strong 
 
 aor. and imperf. {f-rvir-ov, e-Ttmr-ov), and in accus. of o- stems, 
 HTiro-j/. This retention of a becomes characteristic of the weak 
 aorist, the only regular exceptions being 3 sing, indie, act. 
 eS«^e (=a-dLk;-Bha-(t)) and 2 sing, imper. act. Seliov. Several 
 Homeric forms however exhibit the weaker vowel sound, e. g. 
 l^ov, -cs, -f. Epic aor. of iKa> j a^ere, Xucreo, e^Tja-ero, ibvaeTO, opcreo, 
 olare ; and the shortened forms Spa-o, "Ki^o, bi^o with e omitted, 
 i. e. op(r-e-(<7)o, X«cre-(o-)o. [Possibly however these forms repre- 
 sent an older formation of weak aorist with the element a and 
 connecting-vowel « and o (as in imperf. and strong aor.) instead 
 of the permanent stem-vowel a in Xva-a-, deiKo-a-, etc.] 2 sing, 
 indie, mid. ikia-a>=:eX.vara-{a-)o, 2 sing. imper. mid, \va-ai is 
 anomalous ; we should expect Xuo-a-o-o, \va-a analogous to pres. 
 imper. Xvov, from Xueo^Xue-o-o. 
 
 The double a- common in Homeric forms may sometimes be 
 explained by the first a- being part of the verb-stem, e. g. eaa-a 
 from fvvvfu^icrvviu, root p^s j cSiKcurira, Kopiira-a, and similar 
 forms from verbs in -fw, where the first o- is due to assimilation 
 of final 8; Saa-a-ao-Sm^SaT-a--, stem 8ar-; and perhaps ircKeaa-a 
 from stem i-eXes, the full form being lost in pres. TeXcm. It is 
 more probable however that in this last case, and possibly in 
 some of the others, a-a- is due to the epic licence which we see 
 in 'OBva-a-eis beside 'oSvaeis, etc. ; and this is certainly true of 
 the forms with double "■ from vowel- stems, tkaa-a-a, KOTia-a-aa-- 
 6ai, etc. 
 
 With stems in X, p, p., v the laws of Greek euphony did not 
 preserve the o- of the weak aorist in contact with these con- 
 sonants (except pa- in a few Epic forms, cKepa-e, Kvpa-as, (j)vp(ra, 
 &pcra ; and Xo- in one form cKeXtra, which survived to later 
 times). In Aeolic the a- was assimilated (p. 74) to the stem- 
 consonant, e. g. ivepparo, etrreWav, iyevvaro, erfvva (cp. Homeric 
 SfpfXKa), and possibly this may have b^en the older process. 
 Other dialects dropped the n- and lengthened the stem-vowel 
 in compensation J e.g. eveiij.a=iev€p,-<ra, EO-rfiXa=l'o-TeX-(7a, iyei- 
 
 vaTo=iyev-(raTo. In Doric this was a pure lengthening of the. 
 
VIII.] .Verb Inflection. 313 
 
 vowels ii, e ; e. g. e(j)dva, stem (j)av- ) ayyrjKa, stem ayyeX- : Ionic JPormation 
 and Attic raised a to rj, and e to ei, e. g. etprjua, ayyeCKa ; I, and V Aorist. 
 
 were simply lengthened in all dialects, e. g. hlKa, rjtivva. 
 
 a- was dropped after p in forms like £KTia=€KTifa (with com- 
 pensatory lengthening) ; ej^ea, Homeric ex^va=exfFa (root x" 
 raised to x^"=^X^F)- Similarly the o- has dropped from eveyKa, 
 ftira to avoid the collision of too many or of incompatible con- 
 sonants. 
 
 The vowel of all vowel-stems is lengthened before a- in the 
 weak aorist and future, eVoi'ijo-a, rroirjcra (Troie-a)), eXvo'a, Xvtrco (\v-a>). 
 In derivative verbs in -am, -em, -om, which a,\l=: -ay Sni formed 
 by suffix ya {-ja), the length of the vowel is natural as expressing 
 a contraction ; and from this large class of verbs it may have 
 passed by analogy to others. A few exceptions are seen in 
 
 such forms as iKokeo'a, KaKiaa ; rjpocra, ap6(ra (from ap6a)\ ^veaa 
 
 from alvfto. 
 
 In conjunctive forms a is lengthened to a, rj by the addition 
 of the mood-sign (see above, pp. 183-4), and the endings are then 
 similar to those of pres. conj., o- alone marking the tense, e. g. 
 Xva-cD, \va-rj-s, etc., Xi5-o-<»-/iat. In optative forms the suffix i 
 makes with a a diphthong — Xva-a-i-p.i, etc. The 'Aeolic opta- 
 tive ' in -a-eia seems to be formed with the suffix ya [ii], le, p. 186), 
 but with the indicative weak aorist terminations, e. g. Xuo-Eia, 
 -as, -e, etc., instead of Xvaelrjv, which would be expected on 
 analogy of Tideirjv, etc. The a of weak aorist-stem in these 
 forms has sunk to e ; but a feeling that a was characteristic 
 of this tense led to its retention in the suffix -ta, which usually 
 becomes ic or irj, though the letter there had really nothing 
 to do with the tense formation. 
 
 5. The Future Tense (Greek). 
 
 The characteristic Greek future termination in -o-m is not (as Origin of 
 has by some scholars been supposed) connected with thfe weak-o-(u(=eaim). 
 aorist -era, except in being originally a tense formation from the 
 same root as (er). From this root as (es) language developed 
 a present form by the addition of ya (see above, p. 205), viz. 
 as-i/d-mi=zm Greek fV-o-i'o) (a hypothetical form), the middle of 
 which, ea-topai, becomes Itro-o^m. The suffix ya, {ja, i) is perhaps 
 
314 Verb Inflection. [chap. 
 
 Fonnation identical with the root t, ' to eo,' seen in l-uev, i-re : and if this 
 of Greek . > s > r i 
 
 Future -<ru. be SO, as-yd-mi or «o--tca='I go to be,' a natural mode of ex- 
 pressing future time by the addition of an auxiliary verb 
 analogous to Je vais /aire in French, ' / am going to do ' in 
 colloquial English, and the Latin form datum iri for fut. 
 iufiu. pass. 
 Traces of The form -o-iffl, thus derived, has in most Greek dialects snnk 
 form -o-iu. (by omission of i) to -nai. Doric however preserves traces of 
 the fuller form, sometimes with i, sometimes with e, following o-. 
 Thus on inscriptions are found ^oaSrja-Uo, oTrevala, jrpa^io/ies 
 (7rpay-(rio-/ifr), ^oaBrjirlovTi, }(api^wiu6a, etc.; Theocritus has future 
 
 forms like ol<Teiifiis = ol(Teones (15. 133), avXrj(TevvTi (7. ^l), atrcviiai 
 
 (3- 38) ; Aristophanes, employing the Doric dialect, gives forms 
 
 such as So^fire, rjae'ire (Ach, 741) 747)i oyopatrovvres (Ach. 75") J 
 
 and in Thuc. v. 79 (terms of a truce between Lacedaemo- 
 nians and Argives — both Dorian) we find eo-a-ciTatzi^iaa-eeTai^ 
 fa-a-Urai. We also, in non-Doric Greek, meet with middle 
 forms known as 'Doric future;' e.g. e'o-a-eiTai (Horn. II. ii. 393, 
 xiii. 317), KKavo-oiiicda (Ar. Pax 1081), liKivaoviieBa and 
 irXevofidQai (ThuC. i. 1 4 3, viii. l), (jjev^oi/ieda (Eur. Hel. 
 500, etc.) The existence of these forms indicates that in 
 non-Doric dialects -ata (-crem) originally prevailed, but gave 
 way to -a-a. 
 
 With this etri<a='I go to be,' other verbs are compounded in 
 order to acquire a future, just as e. g. the perfect-stem in Latin 
 is compounded with era in cecid-ero, amav-ero ; the initial e of 
 the root « being lost in the process as in the weak aorist forma- 
 tion (above, p. 211) and in Latin sum; so that -ato becomes 
 the normal future termination. The future of stems in X, p, 
 H, ii, e. g. TtvS>, <j}avS>, luva, vepa, |3aX«, etc. (which evidently arise 
 from -i(Tco, by loss of a- and contraction of -ea>, so that iiev&= 
 fiieveaz=iicvc-(Tio) show an e between the stem and a- which is 
 sometimes supposed to belong to the root fs; on which sup- 
 position there would be a distinct and older class of future 
 forms, with the addition of the fuller form -ea-uo to the verb- 
 stem. The analogy however of certain Sanskrit forms, e.g. 
 tan-i Bhyft-ini=r£i'-€-(r»a) (whence reve-tra, Tev(-a>, rcf-u) seems 
 
^v^m.] Verb Inflection. 2,15 
 
 to justify the view taken in Curtius' Greek ' Grammai-, S 262 \ Greek 
 ,-. .. ' Future, 
 
 that the f is a phonetic insertion between the stem and the 
 
 future suffix, in satisfaction of the laws of Greek euphony which 
 (as we saw in the case of the weak aorist) did not tolerate 
 an o- in close juxta-position with X, n, i>, p. In the case of the 
 weak aorist of such stems as e. g- (pav-, & disappeared from 
 the contact — £-<t>r]v-a=^^av-a-a 1 in the future it was retained 
 in the first instance by the intervention of e (^av-c'-o-a)), but 
 then disappeared in accordance with another euphonic tendency 
 to drop <r between two vowels (p. 66). And as in the case 
 of the weak aorist a few exceptional (chiefly Epic) forms remain 
 in which a- survives in contact with p and X, so in the future we 
 find exceptional (chiefly Epic) forms like Kepaa, Kvp<ta, depaofuu, 
 and KiXaa, which show the shortened form in -o-(o=:-o-iai, and 
 that at quite an early stage of the language. Other futures 
 which show traces of this e between the stem and the future- 
 ending are ibovp.ai,z=i&i-trop,ai, ir«rovixai,=:ir«T-i-(ropai, fiaj^ovpai, 
 Epic iiax(<^oiiai, iia)(iiTvofi.ai, 
 
 T has similarly been lost and the vowels contracted in ;8i^5 
 (^■^z^iPda-a, /Strati)), (XS>, fia/xS, the so-called 'Attic futures'^:' 
 o- has been lost without contraction in the Homeric forms avva 
 (11. iv. 56), cpva (xi. 454), ravia (Od. xxi. 1 74). These forms 
 have become like present-forms by loss of a-, but there are 
 others which really are present formations to which a future 
 
 ' Curtius now molines ('Das Verbum,' II. p. 306) to the supposition of a 
 double aeries of stem-forms, e.g. m«», mana, whence /ttv- of aorist iiuiva = 
 i~lifv-aa, nivt- of future nevai =,iieve-{a)a. The i of Sanskrit tanl-shya-mi 
 would thus be a weakening of a in the stem-form tana-, corresponding to 
 Ttrf- of Greek r(ve{<r)ai =Tevfaiai. He allows that in the Greek examples 
 the vowel has "become a merely phonetic adjunct ; and it is apparently for 
 the sake of consistency with his present views about the 'thematic vowel' 
 in ordinary conjugation (p. 1671, that he ia now unwilling to regard it as 
 originally phonetic. Whether, however, it is necessary to strain after such 
 consistency, in face of other undoubted examples of purely phonetic 
 insertion of a vowel (pp. 8,^, 166), may be doubted. 
 
 ^ Other examples of 'Attic future' are SikSc (Hdt. i. g']) = SMaaeiv, 
 «o\q. (Ar. Eq. 456) = KoA-afffi, ireXiu (Aesch. P. V. 2S2, cp. viKaaai, Eur. El. 
 1332), diroffweSS (Soph. O. T. 138), 7a/i« (Aesch P. V. 764, etc.),Ka0(5ovfmi 
 (Ar. Kan. 200), and many futures in -lai, -lovfiai from 1 stems. The term 
 ' Attic future,' applied by old grammarians, is xeaUy incorrect, many of the 
 forms in question being found in Homer; while in some verbs (e.g. Si*afeu) 
 the Attic dialect invariably retains the a. 
 
3l6 
 
 Yerl Inflection. 
 
 [chap. 
 
 Two forms 
 of Latin 
 Future. 
 
 Future in 
 -bo. 
 
 "meaning has attached, notably et-ju, ibo : compare the forms 
 (chiefly Epic) tdofiat, Trio/xai, /Se'o/^at or jSfto/tai, Srja, etc. 
 
 Future (Latin). 
 
 Here we find two distinct forms : i. a modified form of pres. 
 subj. (with consonant and i- or m- stems) which like sim, etc. 
 (see p. 186) is probably an optative form. With a- and e- stems 
 however this form, if used for the future, would lead to confu- 
 sion with pres. subj. in the one case {amemus), with pres. indie, 
 in, the other {monemus) ; and with these verbs, accordingly, we 
 find another form, ama-bo, mone-bo. A similar future in -ho 
 (besides the more usual form in -am, -es, -et) is found from 
 1, stems in earlier writers (Plautus, Terence, etc.), e. g. aperiho, 
 adgredibor, scibo, etc. ; but none of these forms survived in use 
 in the ist century B.C. except ibo, quibo, nequibo. Propertius 
 has leiiibo ; and we find veniet for the more usual venihit (future 
 of veneo=:venum, eo) in the Lex Thoria, 112 e.g., and exiet in 
 Seneca. [The pulcrior exiet of Hor. Od. iv. 4. 65, though 
 accepted by Orelli, has very little MS. authority; evenit is 
 probably coiTcct.]. Dicebo,jideho are also quoted from old Latin. 
 
 This termination -60, like -bam of the imperfect and -ui, -vi 
 of perfect (see p. 194), is generally supposed to be a tense-form 
 of the stem fu-, ' to be,' whence fui, fore, etc. Opinions how- 
 ever differ as to what precise tense-form it represents, two 
 explanations being given ; — 
 
 1. -bo=fu-o, a present formation : thus ama-bo would be 
 analogous to ' I am to love.' 
 
 2. -ho-=bu-i-o, fu-i-o ; a form analogous to la-la, esio, ero, 
 and='I go to be' (see above, p. 214). This latter is more 
 generally adopted, perhaps on account of the close analogy 
 which it presumes between Greek and Latin in the formation 
 of the future, Greek taking one form of auxiliary (eo--), Latin 
 the other {fu-). 
 
 Was this future in -bo the original form of all futures in 
 Latin 1 Curtius (' Temp, und Modi,' p. 324) thinks not, because 
 it is contrary to all analogy that language should proceed from 
 a compound to a simpler form. The most primitive usage, he 
 thinks, was to employ the optative form {dicem, faciem) as 
 
Till.] Tei-b Inflection. 217 
 
 future: the form in -ho beinff a later form, and as such applied Latin Fu- 
 
 , , . . turem iio. 
 
 mainly to the derivative verbs of ist, 2nd, and 4th conjugations, 
 
 and but little to consonantal stems. Forms therefore like sugebo, 
 
 dicebo, viveho (which are very few in number), are not relics 
 
 of an earlier formation for consonantal stems, but anomalous 
 
 later forms on the analogy of a- and e- verbs. 
 
 The ordinary ' futurum exactum ' or ' Completed Future ' 
 {amav-ero,cecid-ero, eic.) is, as has already been noticed (p. 214), 
 a compound form; the future (or pres. with fut. signification) 
 of stem es-, i.e. ero:=esio, being added to the perfect-stem, the 
 characteristic i of which vanishes ; amav-ero, scrips-ero, tetig-ero, 
 ded-ero, etc. 
 
 In the older language of Plautus, old laws, and formularies, Futures in 
 etc., is found a series of future forms with characteristic s ; '.sere, ' 
 tiz. indie, -so or -sso {facso, amasso, etc.) ; subj. -sim or -ssim 
 {faxim, cmsim, locassim) ; infin. -sere or -ssere (reconciliassere, 
 Plaut.) ; jjass. indie, -sttur or -ssitur {mereassitur, Lex Thoria ; 
 Jussitur, Cato; faxitur). A number of other examples are 
 given in Eoby's Latin Grammar, i. §§ 619, 620 : but almost the 
 only forms which survived after Terence are faxo, f axis ; ausim, 
 ausis. Terence has besides excessis (And. iv. 4. 21) and apel- 
 lassis (Phorm. v. r. 15); Lucretius has cohihessit (iii. 444); 
 Cicero, difaxint; Catullus, recejpso (xliv. 19) and tepefaxit 
 (Ixviii. 29); Yirgi\,jusso (Aeu. xi. 467). Two explanations of 
 these forms are given : — 
 
 I. On the analogy of amassem, amasse, consuessem, etc., and 
 other forms acknowledged as syncopated (dixti, extruxem, con- 
 sumpse, etc., see Wordsworth's Introd. xviii. 12, p. 149), these 
 forms are regarded as formed by the addition to the perfect- 
 stem, of -so=-ro of the ordinary 'futurum exactum,' ihe e pre- 
 ceding this -ro (amavero) being on this view regarded as the 
 i of perfect-stem shortened to e, as abl. -t to -e (p. 125). Thus 
 amasso = amavi-so =ainave-so::= amavero. [Schleicher in fact 
 classes the forms in -so under the head of ' futurum exactum,' 
 of which he distinguishes (a) the shorter and older form -so 
 added to pure verbal-stem ; (b) the longer and later form in 
 which -so is added to the perfect-stem, including the ording,ry 
 
31 8 Verh Inflection. [chap, 
 
 LatoiPu- forms amavero, etc., and the syncopated forms a/masso, etc.] 
 -aim, .sere. To this view two objectious are made ; (i) that it does not 
 account for forms like cap-so, rap-so, faxo (fac-so), ^oMbesso, , 
 etc., where the present- and not the perfect-stem seems to be 
 employed ; (2) that it does not properly account for the double 
 s ; the view that this ss is not a compensation for the loss of 
 V or ui being not only conjectural, but contrary to the analogy 
 of other contracted forms. Amaveso, by loss of v, would more 
 naturally become amaeso, amaso, amdro ; or if it did become 
 amavso, it would contract to amauso or ammo (cp. amdram^^ 
 amaveram, nauta=^navita, aetas=-aevitas, etc.). Accordingly 
 others explain these forms as — 
 
 2. Formed from the ^e«e»<-Btem, like the Greek future in 
 -tra : -so, -sim, 'sere being respectively a future indicative, sub- 
 junctive, and infinitive, formed by the addition of s to the stem ; 
 a fi.nal i or sometimes e of the stem being dropped, as in fac-so 
 (faxo) from stem/«cj/ sponso {spondso) from stem aponde; 
 ausim (aud-sim) from stem avde. Other e- stems preserve the 
 vowel, prohibe-ssit, cohibe-ssit, etc. The double s in these forms 
 and those from a- stems may possibly be due to a mistaken 
 analogy from the forms amasse, consuessem, etc., for amavisse, 
 consnevissem, etc. : or it may have been a mode of marking the 
 accent, or of preserving by additional stress the characteristic s ; 
 a single s between two vowels, as we have seen (above, p. 66), 
 being very rare and almost always changed to r. It may be, 
 therefore, that in pronunciation of these forms the desire to 
 retain what was characteristic of meaning, viz, s, struggled 
 against the phonetic tendency to resolve s into r ; and that the 
 success of this effort affected orthography in the Bs of the forms 
 in question. 
 
 The objection urged to this explanation is the difficulty of 
 regarding ss as merely the result of accent in pronunciation; 
 a view which has already been set aside in the explanation of 
 the superlative termination -issimus (see above, p/ 134): but 
 upon the whole the difficulty appears less than those which 
 attend the other explanation. 
 
 [N.B, The verbs araesso, capesso, facesso, laeesso are prob- 
 
VIII.] Verb Inflection. 319 
 
 ably similar formations, originally future, from arcio (=a(^-ao, Latin Fu- 
 
 ' ° ■> ' •■ tures in -so, 
 
 see p. 65), capio, facio, laeio ; but they have been treated as etc. 
 
 present-stems, and so received fresh inflections of tense and 
 mood. Similarly mcesso {=incecl-so, p. 75) is formed from 
 incedo : and petesso or petisso (Cicero, Tuscul. II. xxvi. 6z : 
 Lucretius, iii. 648, v. 810) is probably a like form from^eio.] 
 6. Tenses forrried from the Perfect-stem (Pluperfect, etc.), 
 (a) Greek Pluperfect. A preterite of aorist form from the Pluperfect; 
 root eo-- (eV-a/i, eVa-f, etc.) is added to the perfect-stem, the Latin, 
 augment being prefixed. Thus froni Tret6a>, perfect-stem Trejroi^-, 
 we have pluperfect i-imroiB-Kraiy), whence Epic eneirotdea, con- 
 tracted naturally into inftroidrj, which is found in old Attic ; 
 -V being added as secondary form of i pers. inflection. In 
 3 sing, however a became t, iireiroiB-ea-eir), imnoiBie ; and the 
 natural contraction was to ei in eirfiToWei, This et having become 
 usual in 3 sing, was transferred by a false analogy to i plur. 
 
 and AndX {ineitoideaafies, eKOiBeajiis, iTtenoWrmes), giving e.g. -ei/ifv 
 
 as its termination, and then to i sing, giving -eiv instead of 
 -))v ; the extreme point of confusion being reached when in 
 3 plur., where alone the full form was retained [eirciToiBea-avljyi 
 and there was never any contraction, the ei representing a con- 
 traction was introduced, giving -eia-av as the termination. But 
 this -eurav of 3 plur,, though always given by grammars, is 
 rarely found in the best MSS. of Greek authors : and many 
 good MSS. of Plato and Thucydides give in i sing, -ijc, not the 
 later and incorrect form -«>'. 
 
 (6) Latin Pluperfect. Here -eram, -as, -at, etc. added to the 
 perfect stem is obviously a coiresponding formation to Greek 
 €(ra{ii) ; latin retaining fuller forms in i sing, and 3 plur. 
 Compare e.g. ^Sea=:§Se!Ta, the older form of ^Sew, with Latin 
 iiideram, tlie pluperfect form vidi=Foida : — 
 
 gSfa 
 
 vld-eram 
 
 i-FeiS-e{(r)a 
 
 vld-eras 
 
 i-F€t&'e(a)as 
 
 md-erat 
 
 f-fe£5-<<r)e 
 
 md-eramus 
 
 c- rctS- f (o")a/i€s 
 
 md-eraMs 
 
 €~Feid-^a)aT€ 
 
 md-erant 
 
 i-Ff&-tvav 
 
220 Verl Inflection. [chap. 
 
 Except that the Greek pluperfect has the augment, the two are 
 identical throughout. 
 
 (c) ' Futurum exactmn' (Greek), by the addition of n- (see 
 above, p. 213) to the lengthened perfect-stem in active voice 
 Tedvri^a (-/ctrm), icrTrj^a ; to the perfect-stem in the middle forms 
 XfKv-aofim, Treirpa^-ofxat^^ireirpay-aoiiai, yeydyjf-ofmi, etc. 
 
 (d) ' Futurum exactum ' (Latin), by addition of er-o, etc., to 
 perfect-stem (see above, p. 214). 
 
 7. Imperfect Tense (Latin). 
 iJaMn Im- The imperfect in -bam, like the future in -bo, appears to be 
 -6am. an exclusively Italian formation, found in all Latin verbs except 
 
 sum, and supposed to be formed from the parallel root fu- as 
 eram from es. Fu-am, then, is the original of -bam, ; the process 
 of change being according to some the loss of u and change of 
 / to h, -fuam, -jam, -bam ; according to others, loss of / and 
 hardening of the u (consonantal =m) to b, -fuam, -uam, -bam,^ 
 This termination -bam, -bas, etc., is added directly to most pure- 
 vowel-stems, e. g. dd-bam, sta-bam, qui-bam, i-bam, and to 
 derived vowel-stems in -a, -e {ama-bam, m.one-bam) : but with 
 derived verbs in I (4th conjug.) and consonantal stems, a long 
 vowel is inserted between the stem and the termination, e. g. 
 audi-e-bam, reg-e-bam. In old Latin poetry this e is often not 
 found with l stems, e. g. ai-bam, sci-bam (Plant., Ter., Lucr., 
 Gatull.); servi-bas (Plant.); insani-bas (Ter.); saevi-bat (hucr. 
 V. 1003), etc. ; and so in later poetry — audi-bant (Catullus, 
 Ixxxiv. 8) ; largi-bar (Prop. i. 3. 25) ; leni-bant (Virg. Aen. iv. 
 528), nutri-bant (vii. 485), redimi-bat (x. 538), etc. These 
 instances are chiefly poetical, and the probability is that -^bam 
 is not an original form, but a contraction for -iebam : and 
 further, that for all derived verbs the form was originally 
 -ebam, but that with a- and e- stems it coalesced with the 
 stem-vowel (mone-e-bam, mone-bam). The origin of this e is 
 doubtful : some explain it as the lengthening of the ordinary 
 connecting-vowel : others as being merely transferred from the 
 derived verbs in e-, by false analogy, to the 3rd and 4th conju- 
 gations ; others (e. g. Bopp) as part- of the suffix -aja, which 
 enters into the formation of derived verbs in a-, e-, %- (p. 207), 
 
viir.] Yerb Inftect'wn. 3,2,1 
 
 and therefore confined in the first instance to these derived J^*'™ ?"• 
 
 perrect. 
 
 verbs. In support of this latter view it is urged that the 
 parallel future form in -io is general with some derived verbs 
 (in a- and e-, ist and 2nd conjugation), not unfrequent in 
 others («- stems), and very rare in consonantal stems. If this 
 view be correct, the e with consonant verbs reg-e-ham, etc., will 
 be the result of false analogy from the vowel or derived stems 
 in a-, e-, i-. Another suggestion, that in the long e we have 
 the effect of a stem-vowel coalescing with the augment prefixed 
 to the auxiliary (e-bam), contradicts all analogy, not only of the 
 Latin language, which exhibits no trace of having used the 
 augment, but also of the Greek, where in compound tenses 
 the augment always leaves the auxiliary and takes its place 
 at the beginning of the whole compound, e. g. e\v<ra not Xu-eo-a : 
 so too in Sanskrit, a-dik-sh.am, not dik-a-sam. 
 
 The original quantity of d in the termination -ham ig pre- 
 served throughout, except in 3 sing., which was shortened, (as 
 amat, monet, regit, etc., see p. 176) in dactylic verse, from Ennius 
 downwards; the old quantity being seen in Enn. Ann. 141, — 
 
 ' Noenum rumores ponebat ante salutem ; ' 
 and (perhaps as an intentional archaism) in Virg. G. iv. 137 ; 
 Aen. V. 853. 
 
 8. Aorist Passive (Greek). 
 
 The two passive aorist-stems in the Greek verb are dis- Greek Aorist 
 tinguished from other passive forms by active person-endings : 
 whence it seems probable that their passive meaning lies in the 
 stems themselves, i. e. in the elements e (i?) and 6e [dfj) appended 
 to the verb-stem. The precise connection, however, of these 
 elements with the meaning in question is matter of conjecture : — 
 
 (i) For the 'strong' 2nd aorist-stem e (17) is added to the 'Strong 'or 
 root, and treated as a root-vowel, the augment being prefixed. 
 Thus from root <pav- is formed the aorist stem e-(j>av-e ; indie. 
 i-<^a.vr\-v; imper. (j)avri-di; conj. cpavc-a, ^ava>] opt. (^ai/e-ti)-i»; ij 
 has usually been regarded as a raising (Steigerung) of f : but 
 some regard 17 as the original form, and e a shortening from it. 
 The origin of £ (i/) is uncertain : Curtius (' Temp, und Modi,' pp. 
 329-30) suggested that it arose from the root jd (2/0)=' to go,' 
 
iZ'i Verb Ir^ection, [chap, 
 
 GrfiekAorist which in Sanslcrit is employed in the formation of passive verbs 
 
 (cp. venum eo or ^eweo= passive of venum do or vendo), and 
 
 which e.g. in irjiu has a causative force='I make to go.' ' But 
 
 this is only a conjecture : and it is equally probable that e is a 
 
 mere increase of the stem, such as is found e. g. in the derived 
 
 verbs, whose stems are sometimes treated as if their final letter 
 
 were the final letter of the root itself ; compare Aeolic (pCKji-fu. 
 
 This, in fact, appears to be Curtius' present view (' Das Verbum,' 
 
 II. p. 322). 
 
 'Weak'or (2) The 'weak' or ist aorist-stem is distinguished from the 
 1st Aorist. , , , 1 , \ 
 
 other by 6 between the verb-stem and e (17). We may say either 
 
 that € {rf) is appended to the verb-stem increased by 6 (instead of ' 
 to the pure verbal-stem as in 2 aor.) ; or, more probably, that 6e 
 (6ri) is appended to the pure verbal-stem ; analysing e.g. iirpax6r)v 
 (stem Ttpay-") into ivpix'^T"- The form probably stands in more 
 or less close connection with numerous other formations in which 
 the same consonant 6 appears ^ : e.g. the present-stems reXfda, 
 (j>Sivv6a>, fuvvBo), irprjda (stem npa- of Tri'fwrpijjut), TrX^flm (77X0-), etrBia 
 (Epic)=e8-5<o (by dissimilation, see p. 79), from which with a 
 further suflSx u-^ja (i/a) is formed the present-stem ea-6te in 
 ea-Bia, or the preterites ea-xf-d-ov, riiiiva-d-ov, rj^epi-B-ovro, etc. 
 This 6ri is possibly identical with thejroot Be- (dJia)=.' to place,' 
 used in the sense of 'to do' or 'make,' with the same force 
 originally, in composition with other verbal stems, as our 
 English auxiliary verb did in such expressions as he did come. 
 If this be so, the formation would be originally active : how it 
 comes to have a passive meaning is an unsolved difficulty. 
 Sanskrit has a compound verbal stem crad-dhd., ' to believe,' 
 from which is formed a present Crad-dadhami, i. e. crad or 
 crath=' trust,' 'belief,' and da-dha,mi=Ti-5i)-;«. To this cor- 
 responds the Latin credo=cred-do (cp. condo, per-do, etc.) ; dain, 
 this and similar forms representing dha of da-dha-mi, rl-Btj-pi, 
 as dii- in dare represents do- da- of 81810^1, da-dft-mi : and English 
 do, did is from the same root. Thus in e-Brj-v, i-Trp&x-Bn-v, cre-do, 
 and English do, did we have one and the same element. 
 
 (3) From these two aorist passive-stems are formed two 
 ' A full list of these is given by Curtius, 'Das Verbum,' ii. pp. 340-345. 
 
Tin.] ^ Terl Inflection. 223 
 
 futures by the addition of the ordinary future termination Greek 
 (see p. 213) with middle peyson-endings, ^ai/ij-o-o/im, XuS^-o-ojuai. Passive. 
 They are apparently late formations ; in Homeric Greek the 
 weak aorist in -Brjarofuu does riot occur at all, the strong aorist 
 only in fityrja-fo-dai (II. X. 365). 
 
 9. Tenses of the Subjunctive Mood in Latin. Latin Sub- 
 
 (i) The Present Subjunctive lias been explained above, pp.present.' 
 185, 186. 
 
 (2) Imperfect Subjunctive. The -rem which in ist, 2nd, and 4th Imperfect, 
 conjugations is added directly to the present-stem, and in 3rd con- 
 jugation apparently with a connecting vowel e {ama-rew,, mone- 
 rem, audi-rem, dic-e-rem), probably :=-seTO, i. e. -erem or -esem ; 
 an optative form of eram, esam analogous to amem. Thus in 
 I plur. eremus, eram/us ; amemus, amamus. This -erem (-esem) 
 therefore originally =esa-i-m (see above, p. 187). For the 
 double s in essem, the ordinary imp. subj. of sum, different 
 explanations have been suggested. Pott, followed apparently 
 by Roby (Latin Grammar, § 609), makes essem^es-sem, the 
 first s belonging to the stem, the second to the suffix -sem 
 (-rem). But in the plup. sub], fecissem, evidently=/eci -|- -se»i, 
 and the two forms can hardly be treated differently : moreover 
 we have traces in Old Latin of both forms with a single s, e. g. 
 eset, esetis, esent on S. C. de Bacch. (Appendix I. ii) ; fuisem 
 is also said to occur. The form essem from edere is a case of 
 assimilation =eal-sem/ sumlaxlj ipossem-=pot-sem {pot-est, pot- 
 ts), ferrem=:fer-sem, veUem=vel-sem (see above, p. 74). 
 
 (3) The Pluperfect Subjunctive is apparently formed in the Pluperfect, 
 same way from the perfect-stem, by adding -sem, the s becoming 
 ss in the ordinary conjugation of all verbs ; am,avi-ssem, reid- 
 ssem. Schleicher (Comp. § 301) thinks that the forms viocem 
 (Virg. Aen. xi. 118), extinxem (Aen. iv. 606), confluxet (Lucr. 
 i. 987), etc., are not, as is usually supposed, syncopated forms 
 of vixissem, exstinxissem, confluxisset, etc., but relics of an older 
 formation - by the addition of -sem to an older perfect-stem 
 without i or is : the ordinary forms in -issem being an addition 
 of -sem to the -is, which, as we have seen above (p. 195), there 
 is reason to suppose was characteristic of the perfect-stem, 
 
224 
 
 Verl Inflection. 
 
 [chap. 
 
 Perfect. 
 
 surviving in the terminations is-ti, is-tis, is-se. THe comparison 
 however of these forms with those acknowledged to be syn- 
 copated, e. g. dixti, intellexti, misti, accestis, consumpse, traxe, 
 promisse, etc., rather points to the conclusion that the pluper- 
 fect forms in question are later contracted forms. 
 
 (4) The Perfect Subjunctive is formed by adding -sim=.-siem 
 to the perfect-stem in i- : thus feci-siem, which becomes fece- 
 rim ; the formation thus bearing some analogy to that of the 
 perfect subjunctive passive by the auxiliary sim {amatus dm), 
 just as the ' futurum exactum ' amav-ero bears to the fut, pass. 
 amatus ero. The original length of the i in sim^siem appears 
 infuenSj dederis (Hor. Od. iv. 7. 20), dederUis, etc. ; though, as 
 has been already noticed (p. 185), the confusion with I of the 
 ' futurum exactum ' has led to frequent intermixing of the quan- 
 tities in both tenses ' ; t is properly characteristic of perf. subj., 
 i of completed future indie. 
 
 Infinitives and Pabticiples. 
 inflnitire The Infinitive is' not a ' Mood,' its various forms being 
 
 not to be \. , , 
 
 classed as a nothing more nor less than cases of verbal nouns : and Phi- 
 *Mood.' ^ . _ . 
 
 lology and Grammar alike must begm by getting rid of the 
 
 misnomer ' Infinitive Mood,' for the proper understanding of 
 its real nature and historical development. The analysis by 
 Grammar of the syntactical uses of the infinitive points to its 
 being a verbal-noun, 'sharing the properties both of noun and 
 verb ^ : (i) of a noun, in that it expresses the action of the verb 
 in general, Uke nouns of action, and in Greek becomes a verbal- 
 noun by combination with the article; (2) of a verb, in ad- 
 mitting inflections of voice and tense, in governing the case 
 of the verb to which it belongs, in being qualified by adverbs 
 and not by adjectives, and (in Greek especially by combination 
 with hv) in sharing the functions of mood in oratio obliqua. 
 And the analysis by Philology of the forms of the infinitive 
 leads us still more surely to the same conclusion; so that there 
 is no class of grammatical forms of which the first origin and 
 
 ' For examples of this confusion, see Eoby, • Latin Grammar,' § 592. 
 ' See Koby's ' Latin Grammar,' vol. ii. §§ 1342, 3. 
 
Tni.] Ferb Inflection. 225 
 
 subsequent development can niore certainly be traced, and 
 a meaning more clearly seen to underly what meets us in 
 Greek or Latin Grammar as an apparently unmeaning form. 
 
 I. In Greek we find two forms of infinitive, (i) the older Greek infln. 
 
 Homeric infinitive m -jJtevm, -j>sv (eS-fievai, aiJ.vve-fiev) ; (2) the -(lerai, -n€j<. 
 form in -evat, -uv, {\ih.om-hai, XuTT-etv). -/Mevai corresponds to 
 Sanskrit mauS (manai), the dative of a sufiis man-, by which 
 a large number of nouns are formed in Sanskrit, Greek, and 
 Latin'; e.g. from Sanskrit gnS,, 'to know,' is formed (g)nElman, 
 Latin {g)nomen, that by which a thing is known, its name r 
 while from the corresponding Greek stem yvm- is formed yvm- 
 nav (yi'a)-f4oi/-os)=;'a knower,' the suffix -)iov, -f^ev, [^-man) being 
 used in Greek chiefly for forming masculine nouns, rXij/imj', 
 Tvoljirjv, etc. In Latin -men is a common termination of abstract 
 nouns in the neuter gender, e. g. teg-men, sola-men, ear-men, 
 tuta-men, etc. : and if we took the dative case of one of these 
 forms to express the object of doing anything, and said canes 
 fecit tutamini domum, we should have an exact equivalent to 
 the Homeric expression Kuvas erev^e ^vkaaixifisvai SofMov, 'he made 
 dogs for the protection of the house.' Thus the notion of 
 purpose or object is in reality the primary notion of the infini- 
 tive; and the expression in English of both dative case of 
 nouns and infinitive by the same preposition to ('I come to 
 say this to Mm ') reflects a philological truth. 
 
 [The above explanation of -invai is that preferred by Professor 
 Max Miiller, to whose ' Chips from a German Workshop,' vol. iv; 
 I am indebted for the statement here given. There is, however, 
 another explanation, viz. that -/lerot is the locative of a suffix 
 -mana {mana-i, cp. p. 126), which, as will be shown below 
 (p. 232), appears as a participial suffix in Greek (^epo-fieyo-s) 
 and in the isolated Latin form ama-mini (sc. estis, see p. 179) ; 
 -fuvai would, on this view, be a locat. sing. fem. of a verbal- 
 noun formed by this suffix, analogous to xo/m-i from stem x")"'-- 
 This explanation appears at first sight more suitable in the case 
 of I aor. infin., e.g. XCo-at from stem 'kvara- (p. 211) : but it 
 
 ' See Max MuUes's ' Chipa &om a German Workshop,' vol. iv. p. 33. 
 
in -evai. 
 
 335 Verb Infection. [chap. 
 
 Greek Infin- cannot show the same clear coincidence of form and meaning 
 ■i^evai, -iixv. as the other view ; and the analogy of -ntv-ai, when the con- 
 sciousness of its being a dative was lost, would tend to produce 
 similarity of termination in \va-ai, though the dative of stem 
 "Kvtra- would properly be something different {Kiira-ai, Xvira-, see 
 p. 128). The -m, however, was not felt to be dative any more 
 than locative, and would assert itself by analogy as the right 
 termination.] 
 
 The infinitive in -pevis probably an abbreviation of that in 
 -juvai ; though it has been suggested that -/ifw may be an 
 archaic accusative corresponding to Latin accusatives like teg- 
 vnen, etc., and expressing the general object of certain acts or 
 movements. 
 2. Infinitive But besides the form in -jievai, we find a form in -evai : thus 
 in Homer we find both "i-iuvai and l-h/ai, cfi-iievai (:=eo--/tei«it) and 
 eluai (^=zia--ivcu). Bopp and others have accounted for this form 
 by supposing the loss of ju : but it is more probably a collateral 
 formation from another sufiix -van or -an, added to verbal 
 bases in the Indo-European languages. ' By the side of d&mau, 
 the act of giving, we find in the Veda da-vaji, the act of 
 giving, and a dative d4-vto§, with the accent on the suffix, 
 meaning "for the giving,'' i.e. "to give." Now in Greek 
 this V would necessarily disappear, though its former presence 
 might be indicated by the digamma aeolicum. Thus, in- 
 stead of Sanskrit d&T^ue, we should have in Greek &oFc»tu, 
 Soevai, and contracted Sovvai. ... In the same manner ehai 
 stands for ia-Fivai, iir-ivcu, ievai, eivat. Hence Uvai stands for 
 iFcvat, and the accent remains on the suffix -van, just as it 
 did in Sanskrit^.' 
 
 The regular infinitives of the perfect active (XeXour-eVat) and 
 of the verbs in -fu {8186-vai, urrd-vai, nde-mt) should be referred, 
 according to Professor Max Miiller, to the parallel suffix -an, 
 dative -ane, for which again he quotes parallel forms in the 
 Sanskrit of the Veda. Schleicher, who regards these forms as 
 locatives, refers them to a kindred suffix -ana, which appears 
 
 ' Max Mfiller, ' Oliipa,' iv. p. 34' 
 
nil.] Ferl Inflection. 337 
 
 in the formation of substantives in Greek {hpeir-avo-v, riim-avo-v, 
 ayX-Avrj, ariip-avo-s, etc.) and Latin {jpag-ina, dom-ino-s, sarc- 
 ina, etc.). 
 
 The ordinary infinitive in -uv is generally regarded as formed Infinitive in 
 by transposition of -fw : e. g. <i>ipfvi becomes (jiipeui, just as 
 <l)cpea-i becomes (pipcis (above, p. 170). As to the form (pcpevi, 
 opinions differ : some regarding it as a phonetic corruption 
 of (jiepfvai, while others, who hold that (jjepimi is dative, suppose 
 the existence of a locative in -i side by side with the dative 
 in -ai^- The Doric form in -ev, e.g. dei8ev=aeiSeiv^ is thought to 
 arise from dropping the final i, instead of throwing it back into 
 the previous syllable ; compare the Doric form of 2 sing. o/ieX- 
 yes with the ordinary form ape\yeis=:dne\yein. An Aeolic form 
 (j>eprjv is found : and Curtius, comparing this with (pepev and 
 (pepeiv, postulates a common form (pepeev, in which ^epe- would 
 be the stem (with thematic vowel), -ev the termination. The 
 raising (in the arsis of a metrical foot f) of the last syllable 
 would give the Homeric infinitive in -eeiv ((jjvyhiv, ISictv, etc.). 
 The present infin. in -eiv, and strong aorist infin. in -eiv might 
 both arise from -eev ; the accent for the aorist being placed on 
 the thematic vowel {ij>vyiev, whence <j)vyelv), for the present or 
 the stem-syllable {<i>ipeeii, whence (jicpeiv). The termination -fv 
 may, Curtius thinks, be connected (by loss of s) with a (Vedic) 
 Sanskrit termination -sani (e. g. pra-bhti-sh&ni from root bhu, 
 Grk. <j>v). <p€pf{<T)fv would then correspond to a supposed form 
 bhava-sani : and Curtius is inclined to refer the Latin infini- 
 tive in re (^-««) to the same origin, making e.g. \iyea/=.\ey((iT)ev 
 =:lege-se (Jege-re). There does not, however, appear much 
 evidence for these hypotheses. 
 
 The middle and passive infinitives in -o-^m, -6ai {hlho-cBm, infinitive in 
 TimTe-a-dai, TeTv(f>-6ai) are explained by Schleicher and others 
 as dative feminine formations from a suffix dM, i.e. dhy-ai, 6yai. 
 Sanskrit exhibits forms in -dhya,i, which evidently correspond 
 to Greek forms in -aBai: e.g. bli4ra-d3iyM=</>epE-(r5at, sa^^ 
 dhyai=ore-(Tfc, yd,g'a-dhya.i=afFcj-^at, Li Zend also occurs a 
 
 ■' Max Miiller, ' Chips,' iv. p. 35. 
 Q 2 
 
aaS Verb Inflection. [chap. 
 
 form verezidydi=p€^e(r6ai, {verez=.Fpey or Fepy), and h'H-zhdydi^. 
 (I)ve-<r6ai ; the latter apparently showing the o- of -a-dai. The 
 origin of this o- is doubtful. It may possibly owe its origin to 
 analogy from other middle forms with <r6, viz. -a-Be, -cdov, -aSijv, 
 'ndav, in which a is perhaps original (see above, p. 173) ; or it 
 may be a phonetic strengthening of -6ai, as -\ua6a of -fifSa; 
 or it may be (as Bopp suggested) the reflexive pronoun (re=:€ 
 expressive of the middle and passive voice, as in Latin, pre- 
 fixed to the termination -6ai ; or -irBai may arise by regular 
 phonetic processes of assimilation and dissimilation from -Syai, 
 representing I. E. dhy&i. 6ym would become by progressive 
 assimilation -66ai or -r^at (see p. 174), and this by dissimilation 
 -(rOai. The existence of so many different explanations only 
 shows how little can really be known with certainty of the 
 origin of this form. In TeTv4>-0ai=iTerim--a-6ai, it has been 
 omitted for phonetic reasons, and tt assimilated to 6. 
 II. The forms of infinitive in Latin are : — 
 
 (i) Active ; -re in amare, monere, etc. 
 
 -se of perfect amavisse, and esse, posse. 
 
 -le in velle, nolle, malle. 
 (ii) Passive : -n or -4 in amari, moneri, regi. 
 
 -ier (archaic) in f wrier, etc. 
 
 Latin In- (i) The three terminations of infin. act. are really the same, 
 
 tive-jc, -le. -viz, -se^=-sei, originally the dative case of a verbal substantive 
 whose stem ended in -as, dat. -asai ; the -as being the same 
 termination as that of fem. substantives in -es (sedes, lobes), or 
 neuter in -us, -Mr=Greek -os (genus, robur, yivos). Sanskrit 
 has a corresponding dative formation, also used as infinitive, 
 e.g. gfiv-Ase (from jriv, 'to live') ; and Latin wAe-re= Sanskrit 
 vah-ase. 
 
 The older form -se is retained in the perfect amavis-se (the 
 perfect-stem ending in -is, see above, p. 195), in esse, 'to be,' 
 and es-se=ed-se, 'to eat,' and in pos-se=pot-se ; s is assimilated 
 to the final consonant of the verb-stem in ferre=:ferse, and velle, 
 nolle, malle =vel-se, nol-se, mal-se. When added to the present-, 
 stem s becomes r after the vowel of a-, e-, and i- stems, and 
 the connecting-vow^l by which it is attached to consonantal 
 
Tin.] Verb Infiedion. 339 
 
 and u- stems (rea-e-re, tribu-e-re) : the -i of I- stems becoming Latin Infiu- 
 e, as in cafe-re, present-stem capi-. ti^e. 
 
 [It should be noted that some philologists consider -ere=-ese 
 =-asai, and not -re, to be the infinitive suffix'. On this view 
 the penultimate e of reg-ere, cap-ere is part of the original 
 suffix, the ? of stem capi- disappearing before it : while in amdre, 
 monere, audlre, the initial e of the suffix is absorbed into the 
 long vowel of the stems ama-, mone-, audi-.] 
 
 The final vowel of the infinitive was originally e {=ei, '*'^^)>S^f™' 
 and traces of this are found in Plautus, e.g. True. ii. 4-74 
 (iambic trimeter) : — ■ 
 
 'Non aiides aliquod mlhi dare muni5soulum1' 
 and Pseudolus, i. 3. 136, trochaic tetrameter catalectic : — 
 
 'jfego soelestus ntino argentum prdmere possiim dome' 
 It is still oftener found before the 'caesura ' pause in tetrameters 
 and tetrameters catalectic, e.g. Asinaria, ii. 4. 14 (iamb, tetram. 
 catal.) : — 
 
 'Abso^de, ao sine me hunc pMere, qui Bumper me ira inc^ndit.' 
 So dare, Ter. Heaut. iv. 4. 2 (724) and other examples quoted 
 by Wordsworth (Fragments, p. 152) from Corssen. 
 
 The other Italian dialects have an infinitive form in -om, 
 -um, -0, apparently an accusative case of a verbal substantive 
 formed from the verbal-stem without any suffix, like venum, 
 2>essum, in venum eo (veneo), and pessum do ; and the ordinary 
 supine in -um to which attaches a dative or infinitive meaning, 
 e. g. spefitaium veniunt, ' they come to see.' 
 
 (2) Passive infinitive in ri-er, i-er (-n, -i). 
 
 (a) One explanation of these forms makes i-er=i-se, i. e. a Passive In- 
 passive or reflexive formation from the infinitive active ana- 
 logous to amo-r from amo- (see p. 178). Thus amari-er= 
 amare-se .' while for consonant-stems a shortened form of infin. 
 act. is assumed, e. g. dice- or did-, whence dici-er. It is, how- 
 ever, contrary to the phonetic analogy of Latin that -se should 
 
 ' Roby, 'Latin Grrammar,' i. § 611. 
 
 ' See Corssen, ' Ueber Aussprache,' etc., li, pp. 474, J, 2nd ed. 
 
23° ^erb Inflection. [chap. 
 
 Latin infta- thus become -er : amare-se, dici-se would naturally become 
 sive in -ier. amares, dices, or dicis, as in 3 sing. amaris=amasi-se (p. i']8). 
 And if tbe final r of -ier=re=se, as in amor, then no account 
 is given of the preceding e. 
 
 (6) To escape these diificulties Corssen devises a new theory, 
 viz. that -r=:-re=-se the reflexive pronoun, and that the rest 
 of the infinitive in -ier is a feminine substantive with a suffix 
 -sia (after vowels), or -ia (after consonants) : e.g. from stem 
 ama-, amasia-se, ama-sies, ama-rier ; from stem die-, dic- 
 ia-se, die-ies, dic-ier. 
 
 These substantives would be analogous formations to gloria, 
 curia, etc., and the vowel change from as to e analogous to 
 that between materia and materies. The theory is perhaps 
 more ingenious than convincing, the mode of composition which 
 it postulates being difiicult if not impossible to parallel ; and 
 though it avoids difficulties which beset other explanations, 
 it must be ranked with them as a hypothesis upon which little 
 evidence can be brought to bear in either direction. 
 
 Roby (Latin Grammar, §§614, 15) gives substantially the 
 same explanation as (a) above. "Without committing himself 
 to the phonetic change of -se to -er, he holds that the ordinary 
 passive suffix -r (I presume after it had taken that fwrn, and 
 its origin as=-«e had possibly been forgotten) was 'added to 
 the active infinitive in the form of er,' the final e of infin. act. 
 taking the form of i before -er on the principle of dissimilation 
 (see p. 62). The final r was then dropped, because of its ill 
 sound after another r, and ie contracted to ^. The stages of 
 change on this view are amare-er, amari-er, amwrie, amari. 
 For the shorter forms in consonant and i- verbs (diei-er, capi-er), 
 he accounts in the following way : if the process above described 
 had been followed in these verbs, then, because the penultimate 
 vowel of infin. act. was short (dieere), the syllable er would 
 have come twice over {diceriSr) ; but the instinctive desire for 
 economy of utterance dropped the first er, i. e. -ier was appended 
 directly to the final consonant of the stem. In the absence of 
 evidence for the date and exact process of the supposed changes, 
 this view is perhaps as likely to be right as any other. 
 
■'^ii.] Yerh Inflection. 231 
 
 The period of transition between the two forms (-ier, -i) can Transition 
 be approximately defined from 220-120 B.C. The ' Lex Acilia to°-i! '""^ 
 Eepetundarum,' circ. 123 B.C., offers the latest example from 
 inscriptions of the form in -ier, which form may therefore be 
 presumed to have passed out of common use after that date. 
 It also offers the first example from inscriptions of the other 
 form in -i {darei, § 9) : but the introduction of this form must 
 have been considerably earlier, as it is more common even in 
 Plautus and Terence than that in -ier. The dates mentioned 
 will probably represent with tolerable accuracy the period of 
 fluctuation, before which -ier, and after which -i, was the regular 
 use. In the poets of the later Eepublic and the Augustan 
 period, Lucretius, Virgil, Horace, etc., the form in -ier is an 
 intentional archaism. 
 
 Perfect Participle Active (Greek) : — 
 
 An Indo-European suffix -vat, -vant appears to have the Perfect Par- 
 !• , ■ »■ , . , , . ticipleAo- 
 
 meanmg or 'possession of or 'supply with' anything; e. g. tive Suffix 
 
 Sanskrit asva-van (stem asva-vat)=' supplied with' or 'pos- 
 sessed of horses.' In Greek this suffix appears as -Few, which 
 (with loss of digamma) is found in the adjectival termination 
 -ets, -e<r(ra=-«T— r, -cvr-ya : lx6v6-eK, vi<j>6-€is, ;^api-€ir, etc. (stem 
 Ixdvo-FiVT, vi(^6-FiVT, xapi-FiVT). In Latin it becomes -vans, and 
 (on analogy of 0- stems) -vonso, -voso* which (with loss of v) 
 appears in the adjectival termination -oso- ; fructu-bso, himin- 
 oso,/orma-oso (form-oso, an older form, formonso being actually 
 found in MSS. of Virgil, etc.). 
 
 The same suffix was also used to form a perfect participle 
 active, which (as the perfect itself) is expressed in many lan- 
 guages by an auxiliary denoting 'possession,' 'I have done.' 
 Sanskrit has a perf. part, in -v&n (nom. masc), -vat (neut.), 
 to which corresponds the Greek perf. partic. active -as= Fot-s, 
 neuter -os=foT (the stem appearing in oblique cases -dr-oy, -6t-i, 
 -oT-a); while the feminine -vla=^-v(Tya apparently results from 
 the combination of another suffix -vas {-us) with the feminine 
 suffix -ya {-ja), and corresponds to Sanskrit -ushi. Thus 
 ■Sanskrit toa-bhtl-vd.H.= Trepan, 'ba,-}3harV&t=Tre(j>v6s : vid-van, 
 vid-ushi, vid-vat=:fejSo)s, eiSuta, eldos (root ftS). The effect of 
 
33^ 
 
 Terh Infection. [chap^- 
 
 the digairima is seen in such forms as laTa-6Tes (Epic), which 
 later Greek contracted into iaT&Tes on analogy of other con- 
 tracted forms, hut which was preserved from contraction while 
 a consciousness of the original digamma in to-ra-f ores survived. 
 Present Participle Actiiie. 
 Present This (as well as the future and aorist participles) is found in 
 
 Participle ^ . , ^ - __ , , 
 
 Active -are*, Greek and Latin, and in other Indo-European languages, by 
 a shorter suffix -ant or -nt (with consonant and vowel-stems 
 respectively). Thus in Greek, from verh-stem ea--, <^ep-, we 
 have part, cct-oi/t-, (j)£p-ovT- ; from stem lara-, la-ra-vT- ; vt-s in 
 nom. sing, sinking to v or s with long vowel preceding. So 
 from ei/i' (^'"'-m'')" ^°'-^'^-^> ^^"T-s, e'-coy, later &v : from tcmiiu 
 (ta-Tu-), ia-Ta-vT-s, i(TTds : in I aor. act. Xvara-vT-s, Xvaas : and in 
 I aor. pass. \v6euT-s, Xu^fi'r. The feminine forms are due to 
 addition oi-ya {-ja), e.g. i(r-ovT-ya, iovr-ya, iovtra ; 'urraiia^='urTavT- 
 ya; and in i aor. \v(Ta(Ta=kvcravT-ya. In these forms the com- 
 hination it sinks to a- (as in 3 plur., p. 176) with compensatory 
 raising of the preceding vowel, and the y {j) sound disappears, 
 but remains in fem. substantives, — yepovirm:=y^povT-ya, etc. 
 
 In Latin the ordinary participle stem is ent- (older out-, unt-, 
 in e-unt-is, etc.). Praesens, absens preserve a participle of 
 sum, which exactly corresponds to that of elfiX given above, 
 
 , e.g. (e)s-ent-s^i<T-6vT-s; the termination -nts of nom. sing., 
 
 which in Greek becomes -av (or in i aor. -as), becoming -ns by 
 the phonetic laws of Latin. The feminine forms corresponding 
 to those of Greek are not used as participles, but are common 
 as feminine substantives, sapient-ia, ahundant-ia, etc. 
 Greek Middle and Passive Participle in -/xevos : — 
 
 Middle and The participial suffix -p.evo-, by which are regularly formed 
 
 Participle the present and future, middle and passive, the aorist middle 
 '*"' '' and perfect passive participles, seems to represent an original 
 Indo-European -mana, one of the many developments of the 
 element -ma in the formation of nominal stems (see above, 
 p. 102). In Sanskrit this appears as -mana, forming middle 
 and passive participles from the present, and strong aorist, and 
 future stems; in Zend as -mna, also participial; in Latin as 
 -mino,. -mno forming substantives like ter-minu-s, fe-mina, 
 
Tiii.J Verh Inflection. 2^^ 
 
 ahi-mnu-s, and with traces of a participial signification in the 
 isolated verb-form of 2 plur. passive, amamini, etc. (see above, 
 p. 179)- Thus we have, from the stem hhar, Sanskrit bhara- 
 m.ana, Zend hare-mna, Greek (jjipo-fievo-s, Latin feri-mini. 
 Latin Past Participle Passive in -tus, and Supines : — 
 The suffix -ta (above, p. 102V one of the commonest noun- ijatjn Past 
 
 ^ ' jr , /> Participle 
 
 suffixes among Indo-European languages, seems to have become Passive iu _ 
 at one period a regular mode of expressing the idea of a perfect Supines, 
 passive participle ('having been' . .). This function it retains 
 in the Italian languages, e. g. Latin seripto-, tJmbrian screih-to, 
 Oscan scrifto- ; but here also there are many traces of a less 
 closely defined use of the suffix in the formation of verbal nouns 
 both substantive and adjective ^, as in the Greek verbal adjec- 
 tive termination to-s, and nouns of action like Kpi-rtjs, ttohj-t^j, 
 etc. Such traces are found in the marly verbal substantives 
 in -tus, declined sometimes as 0- stems, more often as u- stems, 
 ' — gemi-tus, fremi-tus, par-tus, etc. ; and the ' sif^ines' in -vmi 
 and -u, which are obviously accusative and ablative cases respec- 
 tively of such a verbal substantive, often not to be distinguished 
 in form from the substantive itself actually in use. Compare, 
 for example, the substantive visus as used by Virg. Aen. ii. 212, 
 ' Difiugimus visu exsangues,' with the supines visum, visu of 
 the verb video. The perfect part, passive, supines, and such 
 verbal substantives, have therefore one common element of 
 formation, viz. the suffix ta- (to-, tu-) ; and the stem formed by 
 the addition of this suffix is sometimes spoken of as the ' Supine- 
 stem,' understanding by that term the base or stem common to 
 these various formations from verbal stems. And in treating 
 here of the formation of the past participle passive in -tus it 
 will be convenient in some cases to borrow examples from the 
 head of ' supines ' or ' verbal substantives.' 
 
 The addition of the element -to to the verbal-stem is marked Phonetic 
 in some cases by certain phonetic changes, which may be shortly Vowei-stems 
 noticed here. In the formation from e- stems, the stem-vowel of -to. ' '" 
 
 ' A list of Latin verba, with their perfects, supines, etc., is given in 
 Roby's ' Latin Grammar,' i. ch. xxx. pp. 239-264. 
 
334 T^erb Inflection. [chap. 
 
 Phonetic becomes shortened to ? in Latin, e. s. mont-tus (mow-), taei-tus 
 
 Changes on _ . , 
 
 addition of (tace-). With a- stems it generally remains as in amd-tus, but 
 
 Past Partio. . . ■,-,/■, \ 
 
 Suffix -(«* IS sometimes shortened to 1 domi-tus laomd-), attom-tus (tona- 
 
 to Vowel- \ /' \ 
 
 stems. re) ; this i being absorbed by a preceding v in adju-tws=:adju- 
 
 vi-tus (juva-), and lauttis^la/vi-tus (lava,-), cp. cautus=:cam- 
 tus {cave-), and fotus, motus, in which the v sound has also been 
 a.bsorbed into the preceding vowel. With I- stems the % re- 
 mains, as in avdl-tMS, moli-tus / but is occasionally dropped 
 out, as in sanc-tus {scmcv-tus being also in use), comjper-tus (but 
 jpen-^a). 
 
 From the cases above-mentioned, in which ?, preceding the 
 participial element -to, is a degradation of sound from a or e, 
 must be distinguished those in which ? is either part of the 
 stem, or a connecting-vowel. In gem-l-tus, vom-t-tus, gertA-tus, 
 frem-i-tus, and a few similar formations from consonant-stems, 
 it appears to be a connecting-vowel introduced for the sake of 
 euphony (see above, p. i66). Without it the forms from stems 
 ge^n-, vom-, frem- would, by the ordinary euphonic laws of 
 Latin, either have lost their characteristic m, becoming /rew-<MS, 
 von-tuB, gen-Vus, or have assumed the ugly forms fremptus, 
 vomptus, gempiw, by the insertion of p sound which seems to 
 follow naturally upon the transition from m to t, s, or I [cp. 
 sum-p-tus, sum-p-si, tem-p-lum (i-e/i-evos)]. Eoby (§ 698) sug- 
 gests that the forms ali-tus (post-Augustan) and molUus from 
 al-ere, mol-ere, may be due to a desire to distinguish these par- 
 ticiple forms from the adjectives alius, moltus ; and that strepi- 
 tus, geni-tus may have originally had stems in a- {strepa- and 
 gena-, cp. gna-scor, gna-tus). In pos-i-tus (pono^^p^s-no, pos- 
 vi) the connecting-vowel is employed, but not universally, in 
 the compound forms ; e. g. repos-ius, compos-tus and eomposi- 
 tus, sepos-tus and seposi-tits. 
 
 Verbs in -w, whose present-stem is formed by addition of 
 i=ya to the verbal-stem (above, p. 205), generally affix the 
 supine or past participle element direct to the verbal-stem^ 
 e.g. capl-o, cap-tus. Where i appears, it may be regarded 
 either as the i of present-stem, or more probably as a connect- 
 ing-vowel; e. g. fug-i-tum {fug%-o, fug-i, fUg-a), where the 
 
VIII.] Verb Inflection. 335 
 
 insertion of a connectinff- vowel preserves the characteristic g, Past Parti- 
 ,.,, ,. -J. cipi* Suffix 
 
 which otherwise would have become c before t, as in Jractus .<«». 
 
 (see p. 49). 80 we have eUci-ium, but iUec-tum ; and in 
 certain words the connecting-vowel, not used before the past- 
 participle and supine, is inserted before the fut. participle for 
 euphony (or-tws, or-i-tmrus ; mor-t-furus / par-i-turus, par-tus). 
 
 Verbs in -uo (m- stems) generally have u preceding t of par- 
 ticiple or supine, e. g. acu-tus, exu-tus, irribu-tus, etc. The stem- 
 vowel of such verbs being generally H, the u is perhaps due to 
 coalescence of a connecting- vowel with the stem-vowel, e.g. 
 inibH-i-tus, imhu-tus. A few verbs retain u, e. g. rU-tum (ac- 
 cording to Varro rutmm), the future participle being ru-i-twrus, 
 cp. ob-rutus : and clU-tus from clu-eo [liki-a), whence in-clu-tus. 
 
 With consonant-stems the suffix -t- appended to the verb- ^^^'jo^^^f 
 stem is sometimes softened to s, usually by the influence ofiiant-steins. 
 a preceding dental (all dental-stems having -sus, -sum in past 
 participle and supine '). The dental either drops out, the pre- 
 ceding vowel being lengthened in compensation (e. g. divl-swm 
 =divid-tiiini, man-sum=^'mand-tuin, or it is assimilated {mes- 
 sum=met-sum, quas-sum=-quat-s'um). Lapsus from lab-or 
 illustrates the same process after a labial : and fixus {=fic-siis 
 iorfig-tus) after a guttural. The guttural, however, often drops 
 out, e. g. par-swm,=-paro-tum, spar-sus^=sparg-tus. [Fig-tus 
 would naturally become fie-tus (see p. 49), and this would 
 lead to confusion vS&x fic-tus, part oi Jingo. And in this and 
 most other cases of change to s, otherwise than after a dental 
 (as above) or after a liq^uid and some other consonant {ard-, 
 arswm, etc.), it is noticeable that the perfect active, if any, 
 is formed with -si : so that the s of participle and supine may 
 be the result of analogy from that of the perf. act.] 
 
 Futwre Participle in -turus (Latin) : — 
 
 The termination of the stem of fut. partic. active in -twro Connection 
 appears to represent an Indo-European -tara, a variety of -tar, Participle 
 which is largely used in the formation of nouns of agency {-tar) Noun-suf- 
 or implement {-tra). In Sanskrit and Zend the suflfcs; appears etc. ' 
 
 * Por examples, see Eoby, § 707, 8. 
 
2^6 Verb tnflectiort, 
 
 as -tar, -tra. In Greek it appears as -rep in na-rfip, na-rep-os ; 
 <r<aTfip=(Ta)Tfp-s, etc.; or -Top in prjrap, prj-rop-os ; larap, (orop-os, 
 etc.; and a feminine -Tpm=.tar-ya is found in iroirjTpia, etc.; 
 while in other cases the i of the feminine suffix -ja {-ya) is 
 thrown back into the previous syllable, e.g. aasTetpa=au>Tep-ya, 
 S6Tcipa=86T€p-ya. Another specially Greek form is a femi- 
 nine stem in -Tpi8=tarid, formed by suffixing -tS, e. g. jrarpiy, 
 Tra-TplS-os. -tra appears as -rpo or -6po (neut.), and -rpa, -6pa 
 (fern. ), e.g. vm-rpo-v, apo-rpo-v, ptj-Spa, pax-rpa (root pax- of 
 patTiTa>=paK-y(o), Koiprj-dpa, etc. Latin has forms almost identical : 
 pa-ter, ma-ter, etc. ; nouns of agency, as vic-tor, censor=:eens-tor, 
 sponsor^spond-tor, etc. ; of implement, as ara-tru-m, claus- 
 Pru-Tn,=zclaud-tro-m, ros-tru-m=rod-tro-m. The suffix is also 
 further increased by -ya (id) to -trio-, -torio- {jpa-1/rio-s, victoria) ; 
 by -ic to -trie-, a feminine suffix, as in victric-s ; by -ino, as in 
 doc-t{o)rina. To the longer form -tara corresponds -turo- of 
 fut. partic. act., and feminine nouns of action, e. g. sepul-twra, 
 usuraz=ut-tti/ra. 
 
APPENDIX I. 
 
 Specimens of Latin Inscriptions from 250 B.C. to the 
 close of the Bepublic. 
 
 The following selection of Latin Inscriptions is given as 
 a fuller and more connected illustration of those gradual 
 changes in the form of Latin words, to which incidental allu- 
 sions have been necessary in the preceding pages. For the 
 text of the inscriptions cited I am immediately indebted to 
 selections made from the Corpus Inscriptionum Latinqrum 
 (ed. Mommsen) and Kitschl's Priscae Latinitatis Monumenta 
 EpigrapMca by Messrs. "Wordsworth (in ' Fragments and Spe- 
 cimens of Early Latin') and Eoby (' Latin Grammar,' vol. i. 
 Appendix B) ; of the general accuracy of whose citations I have 
 satisfied myself by comparison with the authorities whom they 
 have followed. The inscriptions are all in 'uncial' (i.e. capital) 
 letters. The vertical strokes denote the ending of lines on the 
 original inscription : but in the version (in italics) of the 
 Scipionic Epitaphs they mark the ' caesura ' of the Saturnian 
 metre. 
 
 I. Epitaphs of the Scipios : — 
 
 I. On L. Cornelius Scipio Barbatus, consul 298 B.C. (In- 
 scription not later than 240 B.C.) 
 
 Cornelius . lucius . scipio . barbatus . gnaiuod . patre | prog- 
 natus . fortis . uir . sapiensque — quoius . forma . uirtutei . pari- 
 suma I fuit — consol . censor . aidilis . quei . fait . apud . uos — 
 taurasia . cisauna | samnio . cepit — subigit . omne . loucanam 
 . opsidesque . abdoucit. | 
 
 Cornelius Lucius \ Scipio Barbatus, 
 
 Gnaeo patre prognatus | fortis vir sapiensque, 
 
22 8 Apjiendix I. 
 
 Cujus forma virtu \ U parissuma fait, 
 
 Consul, censor, aedilis \ quifuit ajpud vos 
 
 Taurasiam Cisaunam \ Samnium cepit 
 
 Subigit omnem Lucaniam, ohddesque abducit. 
 ' 2. On L. Cornelius Scipio, consul 259 B.C. (Inscription 
 perhaps earlier than No. i.) 
 
 honcoino . ploirume . cosentiont . r 
 
 duonoro . optumo . fuise . uiro 
 
 luciom . scipione . filios . harbati 
 
 cousol . censor . aidilis . hie . fuet . a 
 
 hec . cepit . Corsica . aleriaque . urbe 
 
 «^edet . tempestatebus . aide . mereto 
 
 Swnc unwm pluri/mi eon | sentiunt romai (i. e. romae) 
 
 honorum optimum fu \ isse virum virorum, 
 
 Lueivmi Scipionem. | Filius Bariati 
 
 consul, censor, aedilis \ hicfuit apud vos. 
 
 Sic cepit Corsicam Aleri \ amque urhem pugnando ; 
 
 dedit tempestatibus | aedem merito votam. 
 
 3. On P. Cornelius Scipio, perhaps son of Sc. AfricEinus 
 Major, augur 180 B.C. (Inscription about 160 B.C.) 
 
 quel . apiceinsigne . dial aminis . gesistei | 
 mors . perfe tua . ut . assent . omnia | 
 breuia . hon os . fama . uirtusque | 
 
 gloria . atque . ingenium . quibus . sei | 
 in . longa . licui set . tibe utier . uita [ 
 
 fa . cile . factei superases . gloriam ] 
 maiorum qua . re . lubens . te . ingremiu J 
 scipio . recip t terra . publi | 
 
 prognatum . publio . corneli 
 
 Qui apicem insignem dialis \ Jlaminis gessisti, 
 
 mors perfecit tua ut | essent omnia hrevia ; 
 
 honosfama virtusque \ gloria atque ingenium, 
 
 quibus si in longa licuis | set tibi utier (i, e. vii) vita 
 facile factis super asses \ gloriam, majorum. 
 
 Quare lubens te in gremium, \ Scipio, recipit 
 
 Terra, Publi, prognatum \ PubUo Corneli (i. e. Cornelia.) 
 
Apjiendise I. 239 
 
 4. On L. Cornelius Scipio, (uncertain who is meant). The 
 inscription dates about 150 B.C. 
 
 I. Cornelius , en . f. en . n. scipio . magna . sapientia [ 
 multasque . uirtutes . aetate . quom . pa;rua | 
 posidet . hoc . saxsum . quoiei . uita . defecit . non | 
 honos . honore . is . hie . situs . quei . nunquam | 
 uictus . est . uirtutei . annos . gnatus . xx . is | 
 
 d ei . s datus . ne . quairatis . honore | 
 
 queiminus . sit . mand u 
 
 Lucius Cornelius, Cnaei filius, Cnaei nepos, Scipio 
 
 magnam sapientiam mul | tasque virtutes 
 
 aetate cuim pa/rva | possidet hoe saxum. 
 
 quoiei (i. e. cm') vita defecit \ non honos, honorem. 
 
 Is hie situs. Qui nunquam | victus est virtute 
 
 annos natus viginti \ is Dili est mandatus : 
 
 ne quaeratis honorem \ quiminus sit mandatus. 
 
 II. From the Senatusconsultum de Bacchahalibus, (or Epis- 
 tola Consulum ad Teuranos,) 186 B.C. ; cp. Livy xxxix. 8-18. 
 
 censuere | homines . pious . u . oinuorsei . uirei . atque 
 
 . mulieres . sacra . ne . quis quam | fecise . uelet . neue . inter . 
 ihei . uirei . pWus . duobus . mulieribus . ploustribus | ar/iiise 
 . uelent . nisei . de . pr . urbani . senatuosque . sententiad . utei . 
 suprad | soriptumest. haioe . utei . in . couentinoid . exdei- 
 catis . ne . minus . trinum | noun dinum . senatuosque . 
 sententiam . utei . scientes . esetis , eorum . | sententia . ita 
 . fuit . sei . ques . esent . quei . aruorsum . ead . fecisent quam . 
 suprad [ scriptum . est . eeis . remcaputalem . faciendam . cen- 
 suere atque . utei | hoce . in . tabolam . ahenam . incei- 
 deretis . ita . senatus . aiquom . cen suit | uteique . earn . 
 figier . ioubeatis . ubei . facilumed . gnoscierpotisit . atque 
 I utei . ea . bacanalia . sei . qua . sunt . exstrad . quam . sei . 
 quid . ibei . sacri . est J ita . utei . suprad . scriptum . est . in . 
 diebus . x . quibus . uobeis . tabelai . datai erunt . faciatis . utei 
 dismota . sient in . agro . teurano . 
 
 Censuere homines plus quinque wiiversi, viri atque mulieres, 
 sacra ne quisquam fecisse vellet, neve interibi viri plus duobus. 
 
340 Appendix I. 
 
 mulierihus plus trihus, adfuisse vellent, nisi de praetoris urbani 
 senatusque senientia, uii supra scriptv/m est. 
 
 Haec uti in contione exdicatis ne minus trinum mmdinwm ; 
 senatusque sententiam uti scientes esseiis. Eorum sententia ita 
 fuit : siqui essent qui advorsum ea fecissent, qyMm supra 
 seriptum est, eis rem capitalem facienddm censuere. 
 
 Atque uti hoc in tabulam ahenam incideretis ; ita senatus 
 aequum censuit ; 
 
 Utique earn figi juleatis, uHfacillume nosc} possit ; 
 
 Atque uti ea Bacchanalia, siqua sunt, extra quam si quid ihi 
 sacri est, ita uii scriptv)m est, in diebus decern, quibus vobis 
 tabellae datae erunt, faciatis uti dimota sint. In agro Teurano. 
 
 III. Inscription of the same age as S. C. de Bacch., but less 
 antique in spelling, (the former being of a more formal legal 
 nature). The earliest example of doubled letters. 
 
 laimilius . 1 . f . inpeirator . decreiuit | utei . quel . hasten sium 
 
 . seruei | in . turri . lascutana . habitarent | leiberei . es 
 
 sent . agrum . oppidumqu | quod . ea . tempestate . posedisent 
 
 I item . possidere . habereque | iousit . dum . poplus . senatus 
 
 que I romanus . uellet . act incastreis | ad . xii . k . febr 
 
 IV. From an inscription at PoUa in Lucania, recording works 
 executed by P. Popillius Laenas, consul 132 B.C. 
 
 uiam . feci . ab . regio . ad . capuam . et | in . ea . uia . ponteis 
 . omneis . miliarios | tabelarios que . poseiuei . hince . sunt | 
 nouceriam . meilia . >H . capuam . xxciiii. | muranum . S'Xxnii 
 . cosentiam . cxxni | ualentiam . c^t'xxx . ad . fretum . ad sta- 
 tuam . ccxxxi | et . . eidem . praetor . in | sicilia . fugiteiuos . 
 italicorum | conquaesiuei . redideique | homines . Dccccxvii . 
 eidemque | primus . fecei .ut.de. agro . poplico | aratoribus 
 . cederent . paastores | forum . aedisque . poplicas . heic . feci. 
 
 [Note in this inscription the fluctuation between e, ei, I 
 (homines, ponteis, feci, fecei) and one example of doubled vowel 
 (paastores)^ 
 
 V. Two inscriptions found at Capua and Aeclanum, and 
 assigned by Ritschl (Pr. Lat. Monumenta, lxiii A, lxiii 0) 
 
Appendix I. 241 
 
 to the years 108 and (about) 90 B.C. ; but exhibiting in some 
 forms (e. g. venerus, loidos, moiros) the spelling of a considerably 
 earlier period. 
 
 I. heisce . magistreis . venerus . iouiae . muru | aedificandum 
 . coirauerunt . ped ccvl/xx et | loidos . fecerunt . ser. sulpicio . 
 m . aurelio . cof. 
 
 Hi magistri Veneris Joviae murum aedificomdum cura/oerunt 
 "pedes CCLXX et hidos fecerunt, Servio Sulpicio Marco Aurelio 
 consulibus. 
 . [cof. is a stone-cutter's mistake for cos.] 
 
 2.0. quinctius . c . f . ualg . patron . munic | m . magi . min 
 . f . surus . a . patlacius . ^ . f | mi uir . d . s . sportas . turreis . 
 moiros | turreisque . a . equas . qum . moiro | faciundum . coi- 
 rauerunt. 
 
 C. Quinctius, Caii Alius, Valgus patronus municipii, M. 
 Magius, Minucii filius, Surus, A. Patlacius, Quinti filius, quat- 
 tuor viri de senatus sententia portas, turres, muros, turresque 
 aequas cum mwro faciwndum (error i<ys: facixvnda^ curaverunt. 
 
 VI. From Sulla's Law 'de xx quaestoribus,' B.C. 80; cp. 
 Tac. Ann. xi. 22. 
 
 q . urb . quel . aerarium . prouinciam . optinebit . earn | mer- 
 cedem . deferto . quaestorque . quel . aerarium . prouin ciam | 
 optinebit . earn . pequniam . ei . scribae . scribeisque . herediue 
 I eius . soluito . idque . ei . sine . fraude . sua . facere . li . ceto 
 . quod 1 sine . malo . pequlatuu . fiat . olleis que . hominibus . 
 earn | pequniam . capere . liceto. 
 
 VII. From Lex Julia Municipalis, B.C. 45. 
 
 quae . uiae . in . urbem . rom . propiusue . . r . p . q . ubei . 
 con tinente . habitabitur . sunt . erfint . quoius . ante . aedificium 
 . earum . quae- [ uiae . erunt . is . earn . uiam . arbitratu . eius . 
 aed . quoi . ea . pars . nrbis . h . 1 . ob . uenerit . tueatur . isque . 
 aed . curato . uti . quorum | ante . aedificium . erit . quamque 
 . uiam . h . 1 . quemque . tueri . oportebit . ei . omnes . eamuiam 
 
 B 
 
342 Appendix I. 
 
 . arbitratu . eius . tueantur . neue eo | loco . ao . consistat . quomi 
 nus . conmode . populus . ea . uia . utatur. 
 
 [p . q . is a mistake ior p . m .=pas8us mille.^ 
 
 VIII. A monumental inscription upon a young actress, dated 
 by Mom.msen (C. I. R. 1009) and Ritsohl {Pr. Lat. Mon. 
 ixxxi.) at the end of the EepubUc, and exhibiting in the main 
 the orthography fainiliar to us in classical Latin : most of the 
 peculiarities being due to carelessness of the mason who cut 
 the inscription (e. g. propiravit, line 7 ; denecavit for denegavit, 
 1. 9 ; deposierunt for deposuerunt, and infistae for infestae, 1. 14). 
 The ' apex' (see above, p. 48) is found on two words only, leti 
 (1. i) and h6ra (1. 7). 
 
 Eucharis . liciniae . I 
 docta . erodita . omnes . artes . uirgo . vdxit . am, . xiiii 
 
 heus . oculo . errante . quei . aspicis . l^ti . domMS 
 morare . gressum . et . titulum . nostrum . perlege 
 amor . parenteis . quem . dedit . natae . suae 
 ubei . se . reliquiae . conlocarent . corporis 
 heic . uiridis . aetas . cum . floreret . artibus 
 crescente . etaeuo . gloriam . conscenderet 
 propirauit . hdra . tristis . fatalis . mea 
 et . denecauit . ultra . uitae . spiritum 
 docta . erodita . paene . musarum . manu 
 quae . modo . nobilium . ludos . decorauichoro 
 et . graeca . inscaena . prima . populo . apparui 
 en . hoc . in . tumulo . cineremnostri . corporis 
 infistae . parcae . deposierunt . carmine 
 studium . patronae . cura . amor . laudes . decus 
 silent . ambusto . corpore . et . leto . tacent 
 reliqui . fletum . nata . genitori . meo 
 et . antecessi . genita . post . leti . diem 
 bis . hie . septeni . mecum . natales . dies 
 tenebris . tenentur . ditis aeterna domw 
 rogo . ut . discedena . terram . mihi . dicas kiiem. 
 
APPENDIX II. 
 
 A. — Adverbial Terminations in Latin. 
 
 The following list of the various terminations of Adverbs, 
 Prepositions, or Conjunctions in Latin will supply more or less 
 obvious illustrations of the statement on p. 93, that they are all 
 in their origin case-forms of Nouns. A few examples only of 
 each are given; and for a more complete list the student is 
 referred to Roby's Latin Grammar, i. §§ 508-547, from which 
 this arrangement is borrowed. 
 
 I. Vowel-endings : — 
 
 -a : ea, hoc, circa, juxtd, su^rd, contra [ablative sing, feminine 
 from a- or 0- stems] : anted, proptered, postilld, etc., seem to be 
 accus. plur. neut. (see p. 155). 
 
 -d : ltd, quid [apparently accus. plur. neuter]. 
 
 -ae : jprae [probably locative -ai, or dat. with locative force]. 
 
 -6 : a common adverbial termination, originally ablative 
 singular of -0 stems =: older -od ; cp. Greek ovra, oirms. 
 
 (i) From substantives : ergo {^pyf), extemplo {=:=extempulo, 
 diminutive of ex tempore), illico {in loco), modo, etc. 
 
 (2) From adjectives and participles : cito, denuo (de novo), 
 omnino, suhito, vera; improviso, necopinafo, etc. 
 
 (3) Numeral adverbs; primo, secundo, etc. ; so immo {=imo, 
 ' at the lowest or least '), postremo, etc. [ablatives with local 
 signification, — in such a place.] 
 
 (4) Pronominal, chiefly of direction to a place (perhaps 
 originally adverbs of manner or circumstance) : eo, quo-ad 
 (' whither to '), illoc (Plautus, later illuc), intra, porro (ird/jprn). 
 
 B 2 
 
244 Appendix 11. 
 
 i^f'^f' Under this head come the adverbs of direction ending in 
 
 in Latin. -orsus, -orsum, i. e. -o-vorsus, -o-vorsum, ' turned towards ; ' 
 a combination analogous to quo-ad, ad-eo. 
 
 (5) Adverbs formed by suffix -do : qua/ndo {quam-do), endo, 
 or indu, an old form of in, seen in the forms endoperator, 
 induperator (Ennius), and indigeo (indu, egeo). 
 
 -u : diu, noctu, du-dum {=:diu-dum ?). Ablatives. 
 
 -e : the common adverbial termination from positive and 
 superlative adjectives with -0 stems [apparently ablative sing. ; 
 see above, p. 124] ; le. g. valde, vere, fere, ferme. Apprime (prae- 
 Ciceronian) is probably =ac? prima, an adverbial expression 
 analogous to imprimis; the e being due to the analogy of 
 other adverbs. 
 
 -e : generally speaking a shortened form of the preceding (-e) 
 termination, like abl. sing, in -e of 3rd decl. (p. 125). 
 
 (i) From -o stems (as -e above) : hene {bono-), male, su- 
 perne, etc. ; and perhaps macte in the phrase made virtute 
 esto. . 
 
 (2) From other stems : dhunde, facile, i/mpune, mage, sponte 
 (abl. of spans.) 
 
 -pS : a form of que, e. g. nem^-pe {=nam-pe, cp. namque). 
 
 -ve : perhaps ='«e?, thus mev«=' or not.' 
 
 -ce : ecce=en-ce, sic, etc.; and appended to locative adverbs, 
 Mc, ilUc, etc. (see p. 153). 
 
 -que : a case-form of qm, appended to pronouns and pro- 
 nominal adverbs (perhaps, as Eoby suggests, a kind of redupli- 
 cation) : quis-que, quando-que, etc. ; so with prepositions and 
 conjunctions, absque, atque (ac-.que), na/mque, etc. 
 
 -pte : suopte, etc. ; possibly =j3cJfe, as in vi-pote. 
 
 -de: in-de, unde, etc. Perhaps =(ie (preposition) shortened 
 by loss of accent. 
 
 -ne : sine, pone {=pos-ne). Probably the same as the inter- 
 rogative particle we, which again may=»ie, 'verily' (wrongly 
 written nae from analogy of Greek vai, vfj). 
 
 -I (occasionally shortened to i) : — 
 
 (i) Ablative cases of manner : qui, qmn (qui-ne), si (abl. of 
 a pronoun =' in which case'), sm (=«i-ee, 'in this way'), utt 
 
Appendix! II. 345 
 
 (probably =g'Mo-<j, 'in which way,' ti being=s? above, Srew, Adverbial 
 
 J. , , terminations 
 
 prockm, etc. in Latin. 
 
 (2) Locative (or Dative) cases ; UK, isti (Plaut., Ter.), more 
 usually illlc, isUc ; heri, temperi, etc. : and in the termination 
 -bi of pronominal adverbs, ibi, ubi, etc. 
 
 II. Consonant-endings (Labial and Dental) : — 
 
 -6; ah, Sb, m6=; Greek ano, vrrb, Sanskrit ara, upa, from 
 pronominal stem -pa (p. 97, note). The form virai suggests 
 a locative case. 
 
 -am : probably accusative cases, e. g. quam, obviam, etc., and 
 thence by analogy to such forms as coram {cum,, os), perperam 
 (per-per-am), 'thoroughly;' in termination /anam. of fctfariam, 
 etc. ; and in suffix -dam appended to pronouns and pronominal 
 adverbs — quidam, quondam, etc. (see p. 156). 
 
 -om (-um). Accusative cases : donicum (Plaut.), later doiiec, 
 dum, quom {cum); actutum, circvm, (stem eirco-), clanculu/m 
 {clam, with diminutive suffix), emtremum-, parum, etc. ; rursum, 
 adversum, etc. 
 
 -em : accusative cases : propediem ; -tern in autem, etc. (from 
 prominal stem to-, see p. 156 ; -dem in quidem, etc. (cp. -dam 
 above). 
 
 -im : (i) in pronominal adverbs, with the meaning of 'at' 
 or ' from ' a place ; and so apparently a locative termination 
 (see above, p. 126), e.g. illim, hinc {him-ce), enim {=zin-im, 
 im being locat. of is). 
 
 (2) In termination -tim, {-sim) of adverbs from or similar to 
 past participles : raptim, conjunctim, mixtim, strictim, etc. ; 
 sta-tim, {sta-re=' a,s you stand'), confestim (perhaps from con- 
 ferire), etc. ; pa/rtim,, ubertim, mcissim,, etc. These may also 
 be locative cases, with the meaning of ' circumstance under 
 which ;' but such a form as partim is suggestive of the accusa- 
 tive case, which (as we have seen, p. 115) is thus formed with 
 certain nouns of the consonant or i- declension. From verbs or 
 nouns with a-, u-, or i- stems we have adverbs in -a-tim, 
 -u-tim, -i-tim : e. g. certatim, nominatim, minutim ; gregatim, 
 turmatim, generatim, tributim, viritim, etc. 
 
 -d in ad sed, haud, apud is perhaps the old ablative suffix 
 
346 Appendix II. 
 
 Adverbial (p. 124): and quod in guodd, quod utinatn, etc., is by some 
 
 terminations „ , , , . „ 71 
 
 in Latin. called an ablative. But quod=. ' because, is most naturally 
 
 accus. of respect (cp. Greek on and the use of 6=' whereas'); 
 
 and the same explanation must hold in th^ other phrases with 
 
 which it is compounded. 
 
 -t in ast, et, at, may be only a phonetic variety of -d {homt 
 and set are given as varieties of havd, sed); ut seems to be 
 short for uti. It is true that Greek <a£=a)r, the ablative of or 
 (see p. 124) rather points to the t as final; but then uti would 
 be a different word : and as this is improbable, it seems better 
 to give up the attempt to show that as and ut are identical. 
 
 -n : aw = Greek hv, which by some is connected with ava; en, 
 in {endo); quin=quine. 
 
 -I : procul, simul (older semol= simile), accusative neuter : 
 vel, perhaps imperative of volo. 
 
 -er : the suffix of the comparative degree (p. 12^) in sup-er 
 (comparative of suh=' up ') and its compounds. 
 
 -per : same word as per (preposition), connected with jrapa, 
 nep-av, pierce, etc., always of time ; nuper (novumper), semper, 
 €tc. 
 
 -ter : a common adverbial suffix from nouns of 0- and i- 
 stems, by some identified with -tus, and regarded as an exten- 
 sion of the ablative suffix -d or -t, by others connected with the 
 comparative sufiix seen in Greek -repo-s, Latin de-ter-ior, etc. 
 (seep. 133). 
 
 -tus: appears to = Sanskrit -tas of adverbs, and perhaps 
 Greek -6cv, as -mus of ist pers. plur.= Greek -fiev (see p. 170): 
 but -6ev more probably = another suffix -dhas. 
 
 -as : alias, foras, etc., accusative plural. 
 
 -us ; secus, tenus, etc. ; perhaps neuter accusatives. 
 
 -is: {i)=ios {ius), neut. ace. of the comparative suffix (p. 
 133): magis=-magios, nimis-=nimios. Perhaps this is the 
 origin of is in pa/ullisper, tant-is-per, etc. (2) Ablative plur. 
 in ybm, gratis {gratiis), etc. 
 
 -iens or -ies the suffix for numeral adverbs, connected perhaps 
 with the comparative suffix -yams (Sanskrit, iyan(s), iyaa ; 
 Greek -uov, -wv; Lat. -ids, -ius : see p. 132). 
 
Appendix II. 247 
 
 Table of Cases used Adverbially in Xiatiu. 
 
 Cases.used. 
 
 Examples of Adverbs formed. 
 
 I. Accusative — 
 
 
 Sing. -m 
 
 quom (cum), circum, quam, obviam, prope- 
 
 
 diem. 
 
 -d, -s 
 
 quod, magis (-ios), secus. 
 
 Plur. -as 
 
 alias, foras, etc. 
 
 „ -a 
 
 quia, ita. 
 
 
 Also neut. sing, and plur. of all adjectives. 
 
 2. Ablative — 
 
 
 Sing. -0 (-6), -a, -u 
 
 ergo, subito, primo, illoo, modo, cito, ea, juxta. 
 
 
 interea, diu, nootu. 
 
 „ -e (-e), -i (-i) 
 
 faoillime, bene, qui, brevi, uti. 
 
 „ -d, -t, -tus, -ter 
 
 faoiUimed, baud, apud, ast, et (?), oaeli-tus. 
 
 
 firmi-ter. 
 
 Plur. -is 
 
 gratis. 
 
 3. Locative — 
 
 I 
 
 -ai, -ae, -i 
 
 prae, illi-c, isti-c, heri. 
 
 -bi 
 
 ubi, ibi. 
 
 -itn 
 
 iUim, istim, illin-c. 
 
 B. — Table of Greek Case-forms used Adverbially. 
 
 (Those which only exist as adverbs are called specially ' Adverbial Cases,' or, 
 ■tniitsta fSim as distinguished from Titiiaas Kotval.) 
 
 Cases employed. 
 
 Adverbs formed. 
 
 Substantival. 
 
 Adjectival. 
 
 Pronominal. 
 
 1. Accusative 
 
 2. Genitive (Local) 
 
 3. Ablative — 
 
 (i) Modal -01s 
 -01 
 
 dtmiv, xipty, etc. 
 
 Compar. neut. sing. 
 Superlat. neut. plur. 
 liaxpSiv (sc. oSov) 
 
 -as ( = -aJT, Lat. -od) 
 -a (Lat. -0) in Kira, 
 KaToiTipa, etc. 
 
 o5, oirow, etc. 
 
 Sis, 6iras 
 Si-Se 
 
 • 
 
348 
 
 Appendioa II. 
 
 Cases employed. 
 
 
 Adverbs formed. 
 
 
 Substantival. 
 
 Adjectival. ' 
 
 Pronominal. 
 
 (2) Local -0a> 
 
 oipav6-9ev {caelirtus) 
 
 
 Seec, etc. 
 
 4. Locative — 
 
 (i) -o(, -m 
 (2) -9. 
 
 OlKOt, x"/"^ 
 
 v-f6-ei 
 
 of, iroi, etc. 
 
 5. Dative (Modal) -j? 
 
 
 
 5. ?"«?. etc. 
 
 6. Instrumental — 
 
 
 
 
 (I) -V 
 (or ? Dative) 
 (2) 
 (or ? Ace. plur.) 
 
 ffxoX?, ffianri} 
 crr7a 
 
 Koiv^, TTffp, etc. 
 Tr&VTr] (Dor, iravTo) 
 Tiixa. «/""> etc. 
 
 Tt'a 
 
 N.B. — The adverbial forms in -rj, -a, classed here as lastrumental cases, may pos- 
 sibly be Dative Sing, and Accus. Plur. respectively. But the absence of 'i sub- 
 scriptum ' in one case, and the occurrence of parallel forms in -a in the other, seem to 
 point (cp. p. 131) to the -a of Indo-European instrumental case as a common origin. 
 The pronominal adverbs y, p»r«p> etc. appear to be dative cases with locative meaning 
 arising from an eUipse of iSy, as in the Latin expressions reeta {via), qua, si qua 
 (ratione) 
 
 The adverbial suffixes -6a (local), -ko, -re (temporal), and 
 -o-e (local, of direction towards) can hardly be assigned in their 
 present form to any case-termination, but may be assumed to 
 have had a similar origin to others which have been so 
 assigned. 
 
 C. — Adverbs in Sanskrit. 
 
 Simple adverbs : — 
 
 I. From cases of nouns and obsolete words. 
 
 (i) From nom. or ace. neuter of any adjective, and of 
 
 certain pronouns and obsolete words. 
 (2) From instrum. case (rarely dative) of nouns, pronouns, 
 and obsolete words. 
 
Ajppendipe II. 
 
 349 
 
 (3) From ablat. case of nouns, pronouns, and obsolete 
 
 words. • 
 
 (4) From locat. case of nouns and obsolete words. 
 
 2. Adverbial affixes: — e.g. -tas (cp. -6tv, Lat. -tui) with 
 sense of 'with,' 'from;' -tra forming adverbs of place; -di 
 (Lat. -de) forming adverbs of time. 
 
 3. Adverbial prefixes : — e. g. a-, ' privative ' (cp. Crk. a-, Lat. 
 in-, Engl, in-, mw-); dus-, dur- (Grk. Suo--) implying 'badly,' 
 •with difficulty;' su-, 'well,' 'easily' (Gk. ^i). 
 
 D.— Comparative Table of Prepositions in Sanskrit, Greek, 
 and Iiatin. 
 
 Sanskrit. 
 
 Greek. 
 
 Latin. 
 
 a, 'to," near to' 
 
 
 ad 
 
 
 apupi, Ep. adv. &iMpls 
 
 (ami- in compounds) Germ. 
 
 
 avA. (adv. ajia) 
 
 ((Mi- in an-helare, p. 149) 
 
 
 avrl 
 
 ante 
 
 apa 
 
 &jrb 
 
 ab, ab-s, ob 
 
 (dS, adverbial afiEx) 
 
 did. 
 
 de 
 
 
 fis ( = €i'-s, p. 149), Is 
 
 in 
 
 
 Iv, Ep. hi 
 
 in 
 
 
 l«, ii (€«-s) 
 
 ex, e 
 
 api 
 
 ii,l 
 
 
 autar 
 
 Kwrd. (adv. «iiT<u) 
 
 inter, intra 
 
 
 liircL 
 
 Germ, mit 
 
 para, ' back,' ' backwards ' 
 
 mxpcL 
 
 
 pari 
 
 mpi (Ep. adv. -ntpl) 
 
 per (per- in per-magnua) 
 
 pra 
 
 ■npb 
 
 pro, prae (pra-i) 
 
 prati 
 
 irpos, Ep. vporl 
 
 
 sam 
 
 ovv, (iiv {Kai)v, p. 47) 
 
 cum 
 
 
 virip 
 
 super, supra 
 
 upa 
 
 imb, Ep. imal 
 
 sub 
 
 
 — 
 
 
Appendix II. 
 
 Sanskrit. 
 
 Greek. 
 
 Latin, 
 
 abhi,' to," towards' 
 
 'Spurious Preposi- 
 
 Other Prepositions of ad- 
 
 adhi, ' above ' 
 
 tions.' 
 
 verbial form, many of 
 
 anu, 'after' 
 
 Sa/iv 
 
 them as adverbs. 
 
 atl, 'across' 
 
 "X/«. M^XP' 
 
 apud 
 
 ava, 'down,' 'off' 
 
 lUToiii 
 
 ciroum, circa, circiter 
 
 ud (ut), ' up,' ' upwards ' 
 
 ivfKtt 
 
 cis, citra 
 
 nl, 'in,' 'on,' 'down' 
 
 irKiiv 
 
 clam, coram 
 
 nis, nir, 'oilt' 
 
 
 contra; erga, extra, in&a 
 
 vi, 'apart' 
 
 
 juxta, penes 
 
 N.B.— But few of these 
 are used as prepositions 
 
 
 pone (posne), post 
 praeter, prope, propter 
 secimdum 
 
 with case of nouns ; 
 
 
 trans, tra- 
 
 they are more com- 
 monly prefixes to verbs 
 or verbal derivatives. 
 
 
 v& mil n, UA a^ 
 
 ultra (cp. ultro) 
 versus, versum 
 
GENEEAL INDEX. 
 
 Ablative, inflections of, in Latin, 
 
 124. 
 Accent, effect of, upon decay of 
 
 vowel sounds in Latin, 60. 
 Accusative, inflections of, in Greek, 
 
 Latin, and Sanskrit, 115, sqq. 
 Adverbs originally case forms, 95. 
 
 — list of terminations of (Appendix 
 II), 243 sqq. 
 
 Agglutinative stage of language, 4. 
 
 Alliteration, illustrative of the ten- 
 dency to reduplication, 52. 
 
 Alphabet, relationship of Greek to 
 Phoenician, 40. 
 
 — Phoenician, Greek, and Koman, 
 comparative table of, 42. 
 
 — Greek, changes and modifica- 
 tions in, 43. 
 
 — Boman, changes and modifica- 
 tions in, 46. 
 
 Analogy, influence of, upon the 
 forms of language : seen in dis- 
 tinctions of gender, 107; -esnom. 
 plur. of consonant-stems, 113; 
 declension of ms, grus, and gen. 
 sing, -is of consonant-stems, and 
 -t of M- stems, 120; dative dual 
 in Greek, 130; gen. sing, isti, 
 152 ; -a$a of 2 sing., according to 
 Bopp, 172; amaminor, 168; con- 
 junctive of verbs in -fu, 1 84. 
 
 Analytic languages, difference of, 
 from synthetic, 159. 
 
 Aorist, 'simple' or 'strong,' 196; 
 reduplicated, 197. 
 
 — traces of in Latin, 198. 
 
 — ' weak' or compound, relation of, 
 to strong Aorist, 210. 
 
 — passive in Greek, 221. 
 
 Aryan, use and meaning of the 
 
 term, 10. 
 Aspirate sounds, and modifications 
 
 of (spirants), 34. 
 
 — changes of, in Latin, 69. 
 Aspiration of uuaspirated letters, 
 
 77- 
 Assimilation of vowels, 6 1 . 
 
 — of consonants, 73-79. 
 Augment, in Greek and Sanskrit, 
 
 theories of its origin, 163. 
 Auxiliary verbs, traces of, in Latin, 
 26. 
 
 — (prosthetic) vowels and con- 
 sonants, 83. 
 
 Cases, original number of, 105. 
 
 Chinese language, the, 4, 7, 8. 
 
 Chordae vocales, their part in the 
 formation of sound, 30. 
 
 Classification of languages, morpho- 
 logical, 4 sqq. 
 
 — genealogical, 8 sqq. 
 
 — of nouns, 108. 
 
 — of pronouns, 142.. 
 
 — of verbs, 168. 
 
 Claudius, the Emperor, his attempt 
 to introduce new letters, 47. 
 
 Comparative philology, questions 
 treated by, I . 
 
 Comparison of adjectives, 132 ; com- 
 parative and superlative suffixes, 
 ib. 
 
 'Conjugations' of Greek verbs, 168; 
 of Sanskrit, 199. 
 
 Conjunctive, suffix of, 184. 
 
 Consonants, meaning of the term, 
 
 30- 
 
 — classification of, 31-3S. 
 
 — changes of, 63 ; general ten- 
 
252 
 
 General Indece. 
 
 denoies, ib. ; substitution, 64 ; 
 loss, 70 i assimilation, 73 ; dissi- 
 milation, 79. 
 .Consonants, insertion of auxiliary, 
 80, 83. 
 
 Grimm's Law, original process of 
 
 changes expressed by, 87. 
 — illustrations of, 86, gi. 
 Guttural sounds, 33. 
 
 D 
 
 Dative case, inflections of the, 127- 
 130. 
 
 Declensions of nouns, 108 ; division 
 of, into vowel- and consonant- 
 declension, ib. 
 
 Definite article, development of, in 
 Komance languages, 26. 
 
 Dental sounds, 33. 
 
 'Dentalism,' 51. 
 
 Derivative sufBxes used in forma- 
 tion of noun-stems, 102. 
 
 Derivative verbs, formation of, by 
 suffix ya (ja), 103. 
 
 ' Digamma Aeolicum,' the, 43, 67. 
 
 Diminutives in -ellus, -ollus, -illus, 
 -ullus, 59. 
 
 Diphthongs, formation of, 36. 
 
 — weakening of, to simple sounds 
 in Latin, 56. 
 
 Double consonants, when first 
 written in Latin, 47. 
 
 Dynamic change, 51 sqq. 
 
 E 
 
 Etymology, general principles to be 
 
 observed in, 49. 
 ' Explosive' or 'momentary' sounds, 
 
 32- 
 
 F 
 
 Final sounds, loss of, etc., in Greek 
 
 and Latin, 72, 73. 
 ' Fricative ' or ' protracted ' sounds, 
 
 32. 
 Future Active (Greek) m-aai, 213; 
 
 in -ffl (Attic), 215; Latin in -bo, 
 
 216; in -so, -sim, -sere, 217. 
 Future Passive (Greek), 223. 
 'Futurum exactum' in Greek and 
 
 Latin, 220. 
 
 G 
 
 Gender, expression of, in Indo- 
 European languages, 106. 
 
 Genitive Case, inflections of, 118- 
 123.^ 
 
 Grimm's Law, formulae of, 89, 90 
 
 Imperfect Tense, forms of, in Greek, 
 209 ; Latin {-bam), 220. 
 
 — Subjunctive (-rem), 223; of ei^i 
 and svim, compared, 196. 
 
 Inchoative verbs, 208. 
 
 Indistinct articulation, changes due 
 to, 79-84. 
 
 Indo-European family of languages, 
 meaning of the term, 9; subdi- 
 visions of, II, 12; comparative 
 antiquity of, 12, 13 ; table of, 13 ; 
 divergence of, 16. 
 
 Infinitive, not a mood but a verbal 
 noun, 224. 
 
 — forms of in Greek, 225-228 ; in 
 Latin, 228-231. 
 
 Inflectional (or terminational) stage 
 
 of language, 5. 
 Inflections of nouns, meamng of the 
 
 term illustrated, 98. 
 
 — of verbs, variety of, 159. 
 
 — effects of phonetic change upon, 
 22 sqq. 
 
 Inscriptions, specimens of Latin, 
 
 250-45 B.C., 237 sqq. 
 Instrumental Case, inflections of, 
 
 131- 
 Intensification of vowel-aound, 53. 
 Iranio subdivision of Indo-European 
 
 family of language, 11. 
 Isolating stage ; see ' Kadical.' 
 
 K 
 
 Keltic group of languages, 12, 15. 
 
 Kelts, their migrations, 14. 
 
 Koppa, the letter, in Phoenician, 
 Greek, and Boman alphabets, 40, 
 45. 46. 
 
 L 
 
 Labial sounds, 33. 
 
 Labialism, 51. 
 
 ' LautverscMebung' of Grimm's Law, 
 87. 
 
 Liquid sounds, 34. 
 
 Locative Case, inflections of, 1 26. 
 
 Long vowel sound, attempts to ex- 
 press in Roman character, 47. 
 
General Index. 
 
 '253 
 
 M 
 
 ' Mediae,' meaning of term and va- 
 rious names for, 32. 
 
 Medial sounds, loss of, in Greek and 
 Latin, 71. 
 
 Middle Voice, a prior development 
 to Passive, 177. 
 
 Middle or Passive (Medio-Passive) 
 inflections, 178-183. 
 
 Moods, number of possible, and 
 meaning of term, 161. 
 
 Mood-signa, position and function, 
 159, 161. 
 
 — forms of (Conjunctive and Opta- 
 tive), 183 sqq. 
 
 Mutes, meaning of term, and other 
 names for, 30. 
 
 N 
 Nasal sounds, formation of, ex- 
 plained and illustrated, 33. 
 
 — influence of, upon preceding 
 sounds, 77- 
 
 — insertion or addition of (Nasalisa- 
 tion), 55 ; employed in formation 
 of present-stem, 304. 
 
 National peculiarities of utterance. 
 
 Nominative Case, inflection of, no. 
 Numeral signs, Phoenician and 
 Greek, 42 ; Roman, 46. 
 
 
 
 Optative, suffix of, in Greek, 185 ; 
 
 forms of, traceable in Latin, 186 ; 
 
 ' Aeolic' in -<xeia, 313. 
 Orthograply, fluctuations of, in 
 
 Latin, 82. 
 
 P 
 
 Palatal sounds, 33. 
 
 Paradigms of notm-inflection, 135- 
 
 141- 
 — of pronominal inflection, 147- 
 
 149. 
 Participles, Perfect Active (Greek), 
 
 231 ; Present and Aorist, 232 ; 
 
 Middle and Passive in -/ievos, 
 
 ib. ; Passive in -tm, 233 ; Fut. 
 
 Active in -turus, 235. 
 Passive Voice, developed from 
 
 Middle, 177. 
 Perfect Active (Greek), strong and 
 
 weak forms of, 191 ; aspirated 
 
 form, ib. 
 
 Perfect Active (Greek), Middle and 
 ' Passive, ib. 
 Perfect-stem, formation of, in Greek, 
 
 188 sqq.; in Latin, 192 sqq. 
 Person-endings of Greek and Latin 
 
 verb, 161, 169 sqq. 
 Phonetic change, influence of, upon 
 
 grammatical structure, 22 sqq. 
 
 — general principles of, 29. 
 
 — limited sense of, as involuntary 
 change, 51, 56. 
 
 Plautine prosody, illustrative of de- 
 cay of vowel-sound, 60. 
 
 Pluperfect Indicative (Greek and 
 Latin), 219; Subjunctive (Latin), 
 223. 
 
 Present-stem, its relation to the 
 pure verbal-stem, 199. 
 
 — different formations of, classified, 
 200 sqq. 
 
 Pronouns, classification of, 142 ; in- 
 flection of, 143 sqq. 
 'Prosthetic' (auxiliary) vowel, 83. 
 
 R 
 
 Radical (or isolating) stage of lan- 
 guage, 4. 
 
 Reduplication, origin and general 
 use of, 52. 
 , — employment of in Greek and 
 Latui Perfect, 187, 192 ; Present- 
 stem, 201. 
 
 — 'Attic' in Greek, 189. 
 Relationship between languages, 
 
 evidences of, 18. 
 
 — illustrations of, for Indo-Euro- 
 pean family, 23 ; for Romanic 
 languages, 24, 25. 
 
 Romanic or 'Romance' languages, 
 
 their descent from Latin, 19. 
 Roots, definition of,.94 sqq. 
 
 — list of pronominal, 97 note. 
 
 Sanskrit, relationship of, to Greek 
 and Latin, 1 7. 
 
 — alphabet, and value of to philo- 
 logists, 38 : Table of. Preface, ad 
 fin. 
 
 Scipios, epitaphs of the, 237-239. 
 Sclavonic group of languages, 12, 15. 
 Semivowels (fricative consonants), 
 as distinguished from mutes, 32. 
 
 — in limited sense, consonantal 
 sound of i, u, 35 note. 
 
254 
 
 SenatusoonsuKnra de Bacchanali- 
 buB, extract from, 239. 
 
 Sounds, relative strength and phy- 
 sical conditions of, 29. 
 
 Sounds, relation of, to letters of the 
 alphabet, 37. 
 
 Spirants (s, 0,/, »), their connection 
 with spintus asper and lenis, 35. 
 
 — changes of, 65-69. 
 
 'Spiritns' 'asper,' and 'lenis,' 34. 
 
 Stems, how distinguished from roots 
 and inflected words, 98. 
 
 Suffixes, different kinds of, 100. 
 
 Tense-stems, 160; classification of, 
 162. 
 
 Tenues, meaning of term, and va- 
 rious names for, 32. 
 
 Terminational stage of language ; 
 see 'Inflectional.' 
 
 Teutonic group of languages, 1 2. 
 
 Thematic (connecting) Vowel, 165. 
 
 Turanian family of languages, 8. 
 
 Verb, how distinguished from noun, 
 
 158. 
 Verb, forms of, more complex than 
 
 those of nouns, ib. 
 
 — inflection, elements o^ i6r. 
 Vocative, not a case, 105 ; forms of, 
 
 in Greek and Latin, 117. 
 
 Voice, distinctions of, in Sanskrit 
 and Greek, 177. 
 
 Vowels, meaning of term, 30 ; enu- 
 meration of, 35, 36 ; changes of 
 56; substitution, ib. ; loss, 59; 
 assimilation, 61; dissimilation, 62. 
 
 — auxiliary, prefixed or inserted, 
 
 83- 
 Vowel-scales, 54, 55. 
 
 W 
 
 Word-formation, processes of, 99. 
 Words, analysis of, into radical and 
 formative elements, 92. 
 
 — division of into noun and verb 
 exhaustive, 93. 
 
INDEX OF SOUNDS AND FOEMS 
 EXPLAINED. 
 
 [N.B. — In order to keep Greek and Latin forms together in this Index, 
 the different order of the respective alphabets has been thus adjusted : 
 17, <v ( = e, 0) appear under e, 0; y imder g ; 9 (th) under t ; f under s ; 
 i under x: <J>, x> ^> appear in their usual place, after «.] 
 
 A, the vowel, 35. 
 
 ■ breaking up of, into E, 0, 36, 
 
 54, 95- 
 -a, accus. sing., 115, 
 -d, instrumental sing., 131. 
 -a, -tea, perf. termination, 191, 
 •a, -a, adverbial termination (Latin), 
 
 243- 
 • -d, neuter plural, 155. 
 -o, adverbial termination (Gr.), 248. 
 -a, thematic, 167. 
 -d (orig. a), I. E. conjunctive suffix, 
 
 1 84. 
 -a, weakening of, in Latin to i, 58. 
 ab-sem, prae-sens, 232. 
 accestie, 224. 
 
 acer, acris, 112; aeerrimus, 134. 
 aSf\<pe (voc), aS€\(pds, ll'J. 
 adeo, 154. 
 
 -ae, nom. plur., 113; gen. sing., 121. 
 deiSev, infin. (Doric), 227. 
 dyayeiv, 160. 
 5710s, 66. 
 
 ago, actus, dfou, 77. 
 ago, in compounds -igo, 58. 
 
 -ai (archaic), nom. plur., 113; gen. 
 
 sing., 121. 
 -aiva, -eiva, verbs in, 104, 206. 
 aUl, alfs (Doric), 126. 
 aXyjnjT^, alx/J-fJTTjs, 1 10. 
 i\i\riiiai, l8p. 
 alicuti, alityimde, 71. 
 alis, alid, 153. 
 alius, a\Xo5, 207. 
 dWofuu (saZio), 206. 
 alterae, dative, 153. 
 am (Eng.), 169. 
 Him, 131. 
 
 dfi0pOTOS, 84. 
 
 &fie\yes, 2 sing., 170. 
 
 &filies, dfiite, 66, 74, 82, 144. 
 
 ApmivaXiiv, 198. 
 
 dvd, hv, 157. 
 
 avaffaa, 75. 
 
 avS&vai, iaSov, rjSoiMi, etc., 203. 
 
 ivSpbs, 83. 
 
 amte-hac, 124. 
 
 -ao, -a, gen. sing., 119. 
 
 -aw, -ofiu, derivative verbs in, 103. 
 
 Apolones (old genitive), 62. 
 
 apprime, adverb, 244. 
 
 ar-biter, ar-cesso, etc. {ad), 65.. 
 
 arcesso, 218. 
 
 dpripoTai, 189. 
 
 'Aryam,' meaning of the term, 10. 
 
 ■as (archaic), nom. plur., 113; aco. 
 
 plur., 117; gen. sing, (archaic), 
 
 121. 
 ■as (I. E.), gen. sing., 118. 
 asellus, 59. 
 
 dariip (star, steUa, etc.), 17, 83. 
 -oToi, -OTO, 3 plur. (Epic), 182. 
 'ASfirqai (locative), 127. 
 oLTp^KTjs, 95. 
 
 ' Atmane-pada' (Sanskrit), 177. 
 au, diphthong, in Latin, 36, 56. 
 aucfips, aucup-is, 61, 109. 
 ausim, 66, 217. 
 ay a (aja), verbal suffix, 103. 
 
 B, formation of sound, 33. 
 
 &, confusion of, with v sound in 
 Latin, 68. 
 
 0, use of, by Greek writers to repre- 
 sent Latin v, it. 
 
 b, seldom initial in Sanskrit, 90. 
 
 0, parasitic, 84. 
 
%^6 Index of Sounds and Forms Explained. 
 
 h, d, pEirasitio, in modem languages, 
 84. 
 
 ■bam, Lat. iniperf., 210, 220; quan- 
 tity of d in, 221. 
 
 PaaiXea (iSatriAcv-s), 115 ; 0aal\fas, 
 118 ; PaaiXfjos, 165. 
 
 len^, maU, 124. 
 
 -bM (-^i)> instrumental sing., 131. 
 
 -bi, dat. sing., 143; as adverbial 
 termination, 245. 
 
 -hhyams, dat. sing, (pronominal), 
 143 ; ■ hhyams, dat. plur., 129. 
 
 hiho, 201. 
 
 bifariam, adv., 245. 
 
 bin (German), 169. 
 
 Ms = Sis, 50. 
 
 -bo, Latin future in, 216; excep- 
 tional forms in, of 3rd conjuga- 
 tion, ib. 
 
 bobus, bubuSt^^g, 
 
 0obs (Poi-s), 118. 
 
 bovermn, etc., gen. plur., 123. 
 
 brother, /rater, etc., 20, 91. 
 
 -bus {-bos, -bios), dat. plur., 1 29. 
 
 C, in Latin, 31, 46. 
 
 — pronunciation of, 37. 
 
 capesso, 218. 
 
 capio, in compounds -cipio, 5S. 
 
 -ce, enclitic termination, 157, 244. 
 
 cecidi (cado), 59. 
 
 ceteri, posteri, etc., 133. 
 
 ci, pronunciation of, in Italian, 78. 
 
 ci and ti, interchange of, in Latin, 
 
 78. 
 dto, adv., 243. 
 clamor, clarus, 59. 
 Claudius, Clodius, 56. 
 coerare, coirare (old fonns of curare), 
 
 56. 
 cogo, 59. 
 
 condtoio, orthography of, 78. 
 confiuxet, 223. 
 consumpse, 224. 
 
 corolla, 69. . 
 
 corpus, corpor-is, 58. 
 coram, 245. 
 
 credere, etc, (root dhd), 88, 222. 
 cuculus, 52. 
 cucwrri [cwrro), 59. 
 cum (aiv), preposition, 50. 
 cum {quom, quvm), conjunction, 61, 
 
 J), formation of sound, 33. 
 d, change of, to I, r, 65. 
 d, parasitic before y, 80. 
 
 d, 'old' ablative termination in 
 Latin, 124. 
 
 -d, neut. sing, termination (pro- 
 nominal), 151. 
 
 dS, dha, distinction between roots, 
 88. 
 
 SatSA\tos, 53, 
 
 S&ms, 64. 
 
 de, preposition, 156. 
 
 -de, adverbial termination, 244. 
 
 SiSae, 198. 
 
 dedro, dedrot, old toims=dederwnt, 
 
 195- , . ^ 
 
 Siinvv/u, StMvviuv, quantity of, 1 90. 
 SeiSlaffofiai, 53. 
 SAiplv, d(\(pls. III. 
 Sfl, 131- 
 817(010, 65. 
 
 -dem, -do, -dam, etc., 156, 245. 
 denuo, adv., 243. 
 deus, 156. 
 
 dico (irirdic-are, Stictj), 55, 203. 
 diewndo {jure), 57, 61. 
 SiSrj, StSivTOJV (Siu), 202. 
 SiSafu, S3. 
 die, gen. sing., 121. 
 dies, Diespiter, 156. 
 -Sto-s, adjectival termination, So. 
 diu, 244. 
 JDius FidMis, 156. 
 dixti, 59, 224. 
 Si^i;ai, 2 sing., 178. 
 domimus, formation of, 98. 
 dono.dedit, 73. 
 dds (imper.), 171. 
 Brireipo, 107, 236. 
 Sovpis (doph), 119. 
 duellum (bellum), duonorum (Jono- 
 
 rum), 51. 
 hvybv (Boeotian) — ^v/hv, 80. 
 dudum, 244. 
 duim {daim, So^ijv), 187. 
 -dwm, 156. 
 
 E, a phonetic variety of A, 36, 57, 
 95 ; position of, in scale of Latin 
 sounds, 57. 
 
 e, affinity of to r, 61. 
 
 -e, gen. sing., 121. 
 
 ? (orig, e), abl. sing., 125. 
 
 -e, -d, -i, dat. or loc, 128. 
 
 I, pronoun, 145 ; declension of, 147. 
 
 I, for reduplication, 189. 
 
 t, in conjug. of weak aorist (Epic), 
 212. 
 
 ■e, -i, Latin adverbs in, 244. 
 
Index of Sounds and' Forms Explained. 
 
 357 
 
 ^a=^v, 210. 
 
 taSov, 164, 
 
 -rai (Epic), -€(, -J), 2 sing, mid., 1 78. 
 
 ecce, 244. 
 
 ecus (equus), 62. 
 
 l7(J), €7011', etc., ego, 142; declension 
 
 of, 147. 
 edim, 187. 
 ^Sed (jSeiv), conjugation of, op. with 
 
 Latin videram, 219. 
 tfo/au (sed-cs), 76. 
 ei-, for reduplication, 189. 
 -«, 3 sing., 176. 
 rfijf, optat., 186; e1it]v ( = l(r-i?)i/, 
 
 syam, aiem), conjugation of, ib. 
 eiKoai, vigtnti, 78. 
 eZ/ti, ei^! (c<f/«J). 6, 99. 
 -eii/ (-effo, -fa), pluperf. I sing., 219. 
 -iiv, infinitive in, 227. 
 cfcai, efifzivm, 226. 
 flo, eo, eS, etc., 146. 
 diraifu, conjunctive, 169. 
 -et/ja, feminine, 107, 236. 
 ~fis, 2 sing.,. 1 70. 
 -CIS, -taaa, adject., 231. 
 -efs, plur. of 0- stems, 114; of i- 
 
 atems, 113. 
 -eis, dat. plur. of is, 63, 155. 
 eis=is, 154. 
 
 ess, nom. plur. of is, 114, 155. 
 ds, preposition, 157. 
 eiffT^KCiy, 66. 
 cff worTo^ii/, 131. 
 eicTova, perfect, 79. 
 kK&aaasv, 75, 306. 
 -171', aor. pass., 221. 
 -ey, Doric infinitive, 227. 
 -ellus, -Ulus, ullus (diminutives), 59. 
 ij\vSov, fi\'fi\ov8a, 54, 83. 
 -em, accus. sing, in Latin, 116. 
 -em, subj. (optative) forms in, 187. 
 ejjteio, kfievs, eftovs, gen. sing., 143. 
 ^/ifts, 66, 144. 
 kfdy, 143. 
 
 ifijA (Aeolic), 74; ^c, imp., 210. 
 -endo (m), 157. 
 -cvai, infinitive, 226. 
 evetlia, iveiijjia, itrreiKa, (<TTfWa, etc., 
 
 212. 
 enim, nam, 83, 245. 
 fvvviu, ves-tis, 74. 
 -ens, -ent-is, participial termination, 
 
 103, 232. 
 eov (imperf. elpil), 210. 
 -ew, -efo), derivative verbs in, 104, 
 
 305. 
 
 -eo;, gen. sing, of a- stems in Greek, 
 119. 
 
 iii6om>, iavoviiriv, kiipaiv, &c., 164. 
 
 f<pi 0iy4)i, 131. 
 
 liroimi, sequor, 50. 
 
 equester, 79. 
 
 eqwidem, qiiidem, 83. 
 
 equus (iirnos, asvas), 49, 50. 
 
 erajd, conjugation of, cp. with Skt. 
 Ssam, Gk. ija, ^v, 210 ; quantity 
 of a in, ib. 
 
 -ere, 3 plur. perf. indie, 58. 
 
 -Ire, 2 sing, pass., 58. 
 
 ergo, adv., 243. 
 
 epyov (work), 68, 71. 
 
 ero, 207. 
 
 Ippeov, 165. 
 
 erunt, 3 plur. perf. indie, 195 ; ten- 
 dency to shorten e of, ib. note. 
 
 Ss (8«m), es, (edo), 170. 
 
 -cs, nom. plur., 112 ; -es (Lat.), 113. 
 
 -ess, feminine termination, 108. 
 
 'iaaa (Jiwv/u), 212. 
 
 esse—edere, 74, 228; infin. of sum, 
 228. 
 
 ta-jxtv, eff-re, 17. 
 
 iaaarat, Doric fut., 214. 
 
 ia-ai {it), 17, 170. 
 
 fiaatiiv, 75, 206. 
 
 es* (edo), 79. 
 
 karaoTfs, 232. 
 
 fffTfXAa (Aeolic), i aor., 74. 
 
 ^a6a, 171. 
 
 iiBi}V, i0i/j.7jv, conjugation of, 197. 
 
 cvaSc (Epic), 164. 
 
 euntes, 61, 62, 232. 
 
 extemplo, 243. 
 
 eidet, 21 6. 
 
 F in Latin alphabet, 43, 69. 
 / sound, formation of, 35 ; pronun- 
 ciation, 69. 
 / representing aspirate sounds, 70. 
 facesso, 219. 
 facillimus, 75, 134. 
 fallo, trcfdXXa), 71. 
 fmrml, 67. 
 /on, 70. 
 /oOT (cp. egi,jeci, etc.), quantity of, 
 
 193- 
 faxim, 217. 
 femina, 10. 
 feriae {fet-tus), 66. 
 ferrem, 75, 223. 
 ^e, gen. sing., 121. 
 
a^S Index of Sounds and Forms Explamed. 
 
 fldea, f'tdo, foedm (foidos), 55, 203. 
 foedos, fardewn ( = hoedus, hor- 
 
 deum), 70. 
 fores (Sipa, door), 70. 
 fraB-tu8,frag-or, 49; frango, 204. 
 fui, 70, 194. 
 /undo, atj>fvS6vri, 71. 
 
 G, in Latin, 31, 46; pronunciation 
 
 of, 37- 
 gwudeo, gavistis, 59. 
 yevirstpa^ 65. 
 ycy~i~Trjs, 166. 
 7€i'os, jeMM, franas, 58. 
 generis, yivovs, 61, 66, 120. 
 genus, declension of, 137. 
 gigno, 201 ; yiyru, 302. 
 yvvat (voo.), 118. 
 
 H, the character, 43. 
 
 h, sound of, 34. 
 
 h, insertion of, after p, t, c, r, in 
 
 Latin, 82, 
 harena, harundo, haruspex, etc., 
 
 orthography of, 82. 
 "EKiPrj, Heovha {Hecoha), 61. 
 heri (x*^*), hesternus, 66, 126. 
 hie, pecu^rities in declension of, 
 
 .154- 
 hiemps, hiems, 50, 84. 
 hisce, nom. plur., 114. 
 ho3 = huc, 154. 
 honSr, honoris, 60. 
 
 I, vowel, 35 ; J (Y), semirowel, ib. ; 
 
 attempts to distinguish them, 47 ; 
 
 to represent i, ib. 
 I, scale of, 54, 55. 
 i sound, weakness of, 57, 59. 
 { representing a in Latin, 58. 
 i, loss of, in Latin, 59. 
 i, affinity of, for dental sounds, 62. 
 i becomes e in contact with a, 0, u, 
 
 62. 
 i (y, j), sound thrown back, 104. 
 -», nom. plur. of Latin 0- stems, 114 ; 
 
 gen. sing., 120. 
 -» (-ei), abl. sing., 125. 
 -t (Gk. 1), loc. sing. (orig. in), 1 26., ; 
 ■i, increase of pronominal stems by, 
 
 in Latin, 152 ; i, Greek suffix, ib, 
 i (perf. Bubj.) and I (2 fut. indie), 
 
 confusion of, 185. 
 
 i, characteristic of Latin perfect, 
 
 193- 
 jam, 156. 
 
 -If, -1)7, -I, 6k. optative suffix, 185. 
 tijfu, 202. 
 Uvai, i/tevai, 226. 
 iens, euntis, 232. 
 -ien, -i, pass, iniin., 229, 230 ; period 
 
 of transition between, 231. 
 Ulectum, cp. with elidttim, 235. 
 illico, adv., 243. 
 iUius, ipsius, etc., 62, 152. 
 Am., accus., nouns which retain. 115. 
 ■im {-in), locative, of pronouns, 153 ; 
 
 as adverbial termination, 245. 
 •im, subjunctive (optat.) forms in, 
 
 187. 
 imago, imitor, 202. 
 ifiey (d/i), 202 ; Imus (?o), ib. 
 '■>kn/m, I plur. perf. indie, quantity 
 
 of, 195. 
 vaclatus, 235. 
 mdmperator, 244. 
 -inis, genitive from Latin nouns in 
 
 -o{n), homo, etc., 62. 
 inqua/m, 169. 
 interdiu, 156. 
 -10), -io, verb forms in, 104, 205, 
 
 207. 
 iofiev, conjunctive, 184. 
 ■lojc, -ior, comparative suffixes, 132, 
 
 i.?3- 
 tiiv, ioivei, Uivya (dialectic for I7411), 
 
 ■143- 
 fffTTOs {equus, asvas), 49, 50, 74, 82, 
 
 140. 
 luir/na, imr^Tijs, no. 
 ipsus, ipse, 58, 67, 153. 
 •is, -it, of 3 conjug., quantity of, 207. 
 -is, dat. plur., 130. 
 iaaai, 176- 
 taiiiv, 77. 
 
 -iese, perfect infin., 195, 228. 
 -issimns, superlative, 75, I34- 
 iste, declension of, 150. 
 larri (fara-Si), 171; Jirras (partic), 
 
 232 ; ia0t (ti/jJ and oTda, ft.) 
 -KTTo-s, Greek superlative, 134. 
 farap, 79. 
 -it, 3 sing. perf. indie, quantity of, 
 
 194 note, 
 ita, 156. 
 
 jugvm, ^vyllv, 80. 
 jure dicundo^dsAivel), 128. 
 T(ov, Epic aor. t«<u, 212. 
 l(a, 202. 
 
Index of Sounds and Forms Explained. 
 
 359 
 
 k, h, in Sanskrit, 33. 
 
 K, in Latin, 46. 
 
 KAL. (Kalendae), ih. 
 
 ■Ka, ' weak perfect,' 191 ; aorist, ib. 
 
 KaXaipaif/, 165. 
 
 kAitvos, vapor, 50. 
 
 Ke/caSHiiv, KeKKvSi, Kex&povTO, etc., 198. 
 
 Kevfos (Epic), 65. 
 
 KiKtiraa, 206. 
 
 H\avcro6iieBa, Doric future, 214. 
 
 -icoy, aorist (juxirXa-Kov, etc.), 191. 
 
 Kopiaaai {ic6pv6-os), 75, 206. 
 
 L, formation of sound, 34 ; origin 
 of its use as Koman numeral, 46. 
 
 I, affinity of, to «, 61. 
 
 I, interchange of, with r, 63, 64. 
 
 A.X, by assimilation from Ky, loi, 
 104, 206. 
 
 U, in Latin forms, 207. 
 
 -la, nom. suffix, 103. 
 
 lac, lact-is, 73. 
 
 lacesso, 217. 
 
 lacrima, 65. 
 
 Xaji^hvai, 204. 
 
 lauttts, lotus, 56. 
 
 X€(7r-£y, 'd-\Lir-ov, \e-\onr-a, 53, 54. 
 
 Xe\atfov, Ke\axov, 198. 
 
 AtXaieat, 178. 
 
 Xiaao^Lai {Kir-rj), 75, 206. 
 
 -AAoi, verbs in, 104, 206. 
 
 locassim, 207. 
 
 \vaeo, 212. 
 
 M, formation of sound, 33. 
 
 — history of the character, 44. 
 
 — origin of its use as Roman nu- 
 meral, 46. 
 
 -m, 1 sing, termination, Skt. and 
 
 Latin, 169. 
 -ma, -mo, -man, etc., noun suffix, 
 
 102. 
 mage (magis), 67. 
 magister, minister, 133. . 
 magistris, nom. plur., 114. 
 -fiai, -aat, -rat, middle and pass., 
 
 theories of their formation, 180. 
 malle, 228. 
 /lapimlptiv, 52. 
 
 -mas (Skt.), -lUS, mus, i plur., 170. 
 Itaxiaojim, imxiaaonai, irnxovfioi, 
 
 215- 
 medias, piiaaos, 207. 
 /ieAi, likkiT-os, 72. 
 melior, 133 ; form, as compared with 
 
 /idWov, 207. 
 
 memoria, 53. 
 
 -/iev, -nfs {-mius), i pers. plur., 170. 
 
 -Hevai, litv, infin. in, 225. 
 
 -lievo-s, participial suffix, 102, 232, 
 
 IxrjvTtv, quantity of, 205. 
 
 Iiipifxva, 70. * 
 
 -liiBa, -/ieaSa, 182 ; -neBov (dual), 
 
 183. 
 -pu, 1 pers. sing., 169. 
 miles, milit-is, 61. 
 lUfUo/tat, 52, 202. 
 •mini, 2 plur. pass., 1 79. 
 modo, adv., 243. 
 
 N, formation of sound, 33. 
 
 ii, a final sound in Greek, 72. 
 
 V e^€\Kv(XTiKdi/, 170, 171, 183. 
 
 -V, accus. sing, in Greek, 115. 
 
 -V, I sing, termination, 169. 
 
 -ya, -ve, -vrj, -yv, use of, to form the 
 
 present stem, 204. 
 -nam, termination, 156. 
 vavs (navis, naus), paradigm of, 
 
 J 39-. 
 
 vavy, vija, 115. 
 
 ne (»ae), val, vr), 157, 244. 
 
 -ne, adverbial term, 244. 
 
 neg-otiwm {nee), 64. 
 
 nempe, 244. 
 
 veds (navas, noims), 36. 
 
 venobes, 64. 
 
 nolle, 228. 
 
 nos, 144; v(b, vtti€, vaiL, vSiXv, ib. 
 
 ■ns, accus. plur. in Gothic, 116. 
 
 ■mt, final sound in Latin, 73 ; treat- 
 ment of stems in, no, in. 
 
 num, nunc, 157. 
 
 0, a phonetic variety of A, 36, 54, 
 
 57.95- 
 
 O lUKpbv and n ffya, 43. 
 
 -o, ablat. sing., 124; Latin adverbs 
 in, 243. 
 
 -a, I pers. sing., 99. 
 
 -q>, -a, dat. sing., 128. 
 
 -01, Si, classification of verbs in, i68. 
 
 Si {ta-a-iu), conjugation of, 184. 
 
 -3, contracted futures in, 214 ; ' At- 
 tic' future in, 215. 
 
 occurro (obc), 74. 
 
 oSovs, dens, 83, 86. 
 
 -01, -ai, nom. plur., 112. 
 
 otda {fiSuv, video), 68 ; conjugation 
 of, compared with Skt. veda, 
 Lat. vidi, 190. 
 
 S 2 
 
26o Index of Sounds and Forms Explained. 
 
 oTkos (yicua), otvos (yinum), 71. 
 
 'Otfjit, -otijv, optat.« 186. 
 
 -OIK, dat. dual, 1 30 ; -oiiV (Epic), ib. 
 
 oino^unum, 73, 238. 
 
 oTaSa, 171. 
 
 oUus, olim, J53. 
 
 S/i/M, 74. 
 
 6fi(j)a\os, 83. 
 
 -wv, gen. plur., 122 ; participle, 232. 
 
 ovoiM, nomen, 83. 
 
 -oi/T, participle stein, (Lat. -ent), 
 
 232. 
 -ovTi (-ovai), -unt, 3 pers. plur. 57, 
 
 176 ; -ocToi, -OKTO, 182, 183. 
 -dvTwv (Doric -6vtw), -wnto, 3 plur. 
 
 imperative, 176. 
 -ow, -0^, derivative verbs in, 104. 
 bipiiXtti, dtfi^Kkaij 206. 
 Spao, 181, 212. 
 -iis, -via, -ds (for-), perf. participle, 
 
 231. 
 -tus, Gk. adverbial termination, 124. 
 8» (ff/^os, suus), 66, r45 ; 8s and d, 
 
 151. 
 -oso, adjectival suffix, 231. 
 -ov, gen. sing,, 119. 
 -ov (-eiro, -co), 2 sing, mid., 180. 
 -ovaa, fem. participle, 232. 
 Si/*, «oa;, paradigm of, 135. 
 
 P, formation of sound, 33 ; the cha- 
 racter, 42. 
 
 p, seldom initial in Gothic and 
 Saxon, 90. 
 
 p, insertion of, between m and t, s, I, 
 84, 234. 
 
 paC'iscor, wfiy-vvfu, 64. 
 
 padas (Skt.)=iro5os, irdSes, trdSas, 
 
 57- 
 pagurit (xil Ta.h.) = panffunt, igS. 
 iranTa\6eis, 53. 
 ■nafKpaiviLVy 53. 
 
 ' Parasmai pada ' (Skt.), 177. 
 parentes, parientea, 198. 
 Parilia {Falilia), 79. 
 paterfamilias {gen.), 121. 
 ■pe, 157, 244. 
 vdSoj, neiroiSa, 54. 
 pejero, 71. 
 pepigi (pac-tum), 59. 
 treiretff/wii, 77. 
 TTftTiBov, 197, 
 pepuli, expuli, 192. 
 perperam, 245. 
 Tiiaov^i, future, 215. 
 petesso, 219. 
 
 irioimi, etc., pres. with fut. sense, 216. 
 Tr/iTTB), 53, 59. 
 plaudo, ex-plodo, 56. 
 7rA€U(7criT^at, iT\€Vffo6fie6a, Doric fut., 
 
 214. 
 jrX(x*'^» (irA.«K-ai), 50, 76. 
 irXoufftos, 78, 175- 
 7r(5\c(s, ir(5Ai€s (Ionic), 112; v6\€ts, 
 
 accus. .plur., 117. 
 irdKeais, •jt6\7]OS, 7rti\(OS (Ion.), gen. 
 
 sing., 119, 
 pone, adverb, 244. 
 pono, pomi, 71. 
 posse, 227. 
 possem, etc., 223. 
 porgo, porrigo, 59. 
 pote (potis), 67. 
 postridie, 126. 
 iroJs (jJes), 91. 
 rrpay/ja, Trpay/jar-os, 98. 
 itpa^iofits (Doric), fiiture, 65, 214. 
 praeseas, 103, 232. 
 prirfje, 156. 
 procus, precor, 55. 
 propediem, 156. 
 
 -yte, adverbial termination, 244. 
 iTToKefws, iTTdXis, etc., 205. 
 
 Q, Q (Koppa), 45. 
 
 jiiaesior, jMaesiioc, 59. 
 
 qiiaestuis, quaesti (gen. sing.), 120. 
 
 qna/ndo, 244. 
 
 quattuor, r^rrapes, 51. 
 
 gue, 244. 
 
 gues, nom. plur., 114. 
 
 qui, adverb, 117, 244. 
 
 quia, 243. 
 
 quine, quim, 244. 
 
 quis, tU, 51. 
 
 guoaiZ, 243. 
 
 quoiei, dative, 128, 152, 155. 
 
 quom, quum, cum, 62, 154. 
 
 K, formation of sound, 34 ; the cha- 
 racter, 42. 
 
 r, affinity of to c, 61. 
 
 -}■ ( = se), characteristic of Latin 
 passive terminations, 1 78. 
 
 -re (-8e), Latin infinitive, 228 ; 
 original quantity of, 229. 
 
 recepso, 217. 
 
 -rem, imperf. conjunctive, 223. 
 
 remits, 71. 
 
 ^iot, tppivaa, fppvrjv, etc., 7O1 203. 
 
 reppvli, repperi, rettuli, 102. 
 
 fii(a {Wurzel, wort), 71, 76. 
 
Index of Sounds and Forms Explained. 
 
 261 
 
 Moma, 71. 
 
 -rum, gen. plur., 122. 
 
 -ruri, 126. 
 
 5, S, different forms of, 43. 
 
 s, changed to r between vowels in 
 
 Latin, 64, 66. 
 s (17), changed to spiritua asper, 66. 
 <r between Towels lost in Gk., 66. 
 -«, final, loss of, 67. 
 -s, initial, loss of, 70. 
 <r<r (tt), 75, 306. 
 ss, in Latin, 75, 76. 
 -s, nominative sutfix, no. 
 -s {-as, -€s, -es), nom. plur., 112. 
 -s, 3 sing. (Eng.), 176. 
 s, representing root as, «s, various 
 
 use of in verb formation, 211 note. 
 -era, 'weak aorist,' 211. 
 -acu, 2 sing, mid., 181. 
 sal, &\s, 64. 
 aclr, the letter, 44. 
 ' Sanskrit,' meaning of the term, il. 
 aa<pr)s, 81. 
 cxis, 171. 
 -SCO, -aaco, verb forma in, 208 ; 
 
 -ffKov, ib, 
 scopulus, ffK6ve\os, 61. 
 seeundus = sequendus, 57. 
 sedes (?Sos), 35, 64, 66. 
 -aeia, ' Aeolio optative,' 313. 
 senatnos, senatuis, senati (gen. sing.), 
 
 120. 
 sepvMm (sepelio), 61. 
 sepositns, sepostus, 234. 
 sequer, iwoimi, 50. 
 sequor, socius, 55. 
 sero, 301. 
 
 -o-(, -s (Skt. -si, -s), 2 sing. term., 170. 
 -ai, -am, dat. (locat.) plur., 127. 
 si, sio, 155. 
 -0-1, 3 sing., 175. 
 -si, Latin perfect in, 193. 
 sldo, 201. 
 sUva, vXrj, 64. 
 
 «im (stem), conjugation of, 186. 
 sisto, 201. 
 seal (Doric), future termination, 
 
 214. 
 -ao, 2 sing. mid. (secondary form), 
 
 -so, -sim, -sere, Latin future forma m, 
 
 217. 
 -am, formation of future in, 213-215. 
 solUstmvwm (tripudium), 134. 
 awpia, aiiimT-m, 72. 
 
 . sont (French), 17. 
 spar-sus, 235. 
 -ssem, plup. conj., 223. 
 -aaai (-TTtu), verbs in, 75, 206. 
 ff^ii, a(pm, etc., 145. 
 -si, 2 sing. term. (Eng. and Grerm.), 
 
 172. 
 statim, 245, 
 steUa, 17, 59, 83. 
 -a9a, 2 sing., 1 71-174. 
 -a9ai, inf., 174, 227. 
 -fffle, 182. 
 aBhai, 81. 
 
 -fffloi, 174, 181 ; -fffloii', -ffflftKray, 182. 
 -sti, 2 sing., 171, 173, 196. 
 stuUus, stolidus, 61. 
 s«6 (uTTo), s«s (Ss), 35. 
 trill/, cum, 50. 
 skZcms (o\kos), 61. 
 summiis, 74. 
 gjimpsi (sijmo), 234. 
 avpiaSes (Doric), 170. 
 suus {apbs, &s), 66, 145. 
 -sya, gen. sing., 118. 
 
 T, change of, to a before i, 78, 175 ; 
 
 in Latin {-ti to -si), ib. ; -tus, 
 
 participle suffix to -sus), 235. 
 -t, nominative of stems in, iii. 
 to-, pronominal stem, declension of, 
 
 150. 
 -tar, the root, modifications of, 95. 
 -tar, noun suffix (agency), 7, 102, 
 
 236. 
 -tara (-repo-s), comparative suffix, 
 
 133- 
 Toeuc, gen. plur., 133. 
 tago = tango, 198. 
 -re, 2 plur., 175. 
 Tcofo (Epic), gen. sing., 145. 
 T€Taycbv, 198. 
 Terv^ffat, 228. 
 reTV(f>ws, TiT-v(p6T-os, no, 
 -TJ7P, -Torp, -Tpoi' (Lat. -tor, -trum), 
 
 noun-suffixes, 236. 
 -th, 3 sing. term. (Eng.), 176. . 
 6e\a, e8i\a, 83. 
 -Orjv, aorist pass., 222. 
 e^Xvs (femina), efjp (Jera), 35, 86, 
 
 91. 
 Sibs, vocative, 118. 
 eh, 171. 
 
 -0t, 2 sing, imper., 171. 
 Blv, Bis, 111. 
 
 -001 (pres.), -60V (aor.), 222. 
 Bvydr-rjp, daughter, 86. 
 
36a Index of Sounds and Forms Exjilained, 
 
 tMijo-i, jsing., 175. _ 
 
 -tim, adverbial termination, 245. 
 
 Tis, quis, 51. 
 
 -Us, 2 plur., 175. 
 
 -Ttti, -to, verbal suffix (present stem), 
 
 305. 
 -Toi', 2 dual, 175. 
 •raaav, 3 plur. imperat., 176. 
 TpeTs, tres, etc., 20, 86. 
 TV (oi), thou, etc., 20, 86, 145. 
 tugurium (tego), 62. 
 TiJi/ij (Boeotian), 145. 
 -taras, fut. participle, 235. 
 turtur, 52. 
 -iMS, participle suffix, 233 ; noun 
 
 suffix, 234 ; t of, softened to s 
 
 after dental stems, 235. 
 rvcji-Sds (riJir-Ttti), 50. 
 Tv<l>9rjTi, I aor. imper., 171. 
 T&lias, rdfco'T-ot, 1 10, 
 
 U, vowel sound of; 35. 
 
 — character (F) in Greek, 43. 
 
 V (=p), changes of, 68. 
 u, affinity of to I, 61, 192. 
 
 V (Latin), pronunciation of, 68. 
 vapor, nairvis, 50. 
 
 -vas (Skt.), dual, 170. 
 
 vayam (Skt.), nom. plur. i pron., 
 170. 
 
 S/3pis (uir^p), 64. 
 
 velle, 228. 
 
 Veda, Vedic, 53. 
 
 -ui (-vi), Latin perfect in, 193 ; 
 origin of, 194. 
 
 viciens, 79. 
 
 victrix, 107. 
 
 vidi, conjugation of, cp. vrith ofSa, 
 etc., 190. 
 
 videram, cp. with pSea, pSeii', 219. 
 
 vlginti, eixofft, 78. 
 
 virgo, virago, 59. 
 
 mxem, 223. 
 
 iilulo, 52. 
 
 -jjm, gen. plur., 122 ; adverbial ter- 
 mination, 245. 
 
 umerus, umor, orthography of, 82. 
 
 -ViTidue, gerundive termination, 57. 
 
 -unt {-ovTi, -ovai), 3 pers. plur., 57, 
 176. 
 
 -va, -ufeu, etc., derivative verbs in, 
 104. 
 
 volumus, 57. 
 
 uoic, vocis, vSco, 99. 
 
 -IMS, -uus, -uls, gen. sing, of u- 
 
 stems, 120. 
 inrd, eub, 35 ; viral, 245, 
 -Us, gen. sing, in Latin, 1 19, 152. 
 iiaiiivr}, 66. 
 uswa, 236. 
 liti, ut, 155, 144. 
 
 4, in Greek alphabet, 45 ; as a 
 numerical sign in Roman ditto, 
 46. 
 
 -<^, ' aspirated perfect, 191. 
 
 tp&aSa, 174. 
 
 (plpwv, ferens, paradigm of, 136. 
 
 ipevloOimi, future, 65, 214. 
 
 -<pi, -<piv (-6A»), instrumental, 131. 
 
 tpoploiai, 3 plur. (Aeolic), 176. 
 
 ^pdfw, 76. 
 
 X, history of the character, in Greek 
 
 and Latin, 45, 46. 
 XapUffffa, 206. 
 
 Xeo), ex^a, Ix^va, etc., 70, 203, 2 1 3, 
 X^ps (Aeolio) = xf 'P, IIX, 
 Xevia (Aeolic), 68. 
 XOiiiv, x^'Sj'os, 1 10. 
 XPl") imperfect, 164. 
 
 W, the character, introduction of, 45. 
 ipiKcL ypdi^fjuiTa, tf/i\Sis yp&cpuv, 32. 
 
 a ( = Ks), history of, 45. 
 
 Y, the character, 47. 
 
 y, the semivowel sound of «", 35. 
 
 y, how represented in Greek, 65. 
 
 y (^°, {), influence of in assimilation, 
 75, 206. 
 
 ■ya {ja, 10, is), noun suffix, 102. 
 
 •ya {trj, to, t, etc.), verbal suffix, 103 ; 
 185 ; (optative), 205 ; (present 
 stem) feminine termination, 107. 
 
 -yant (yans, ians, -taiv, -ios), compara- 
 tive suffix, 132. 
 
 yug, the root, 80. 
 
 Z, in Greek alphabet, 44; in Roman, 
 
 46. 47- 
 Zeis, 76. 
 
 -fo) ( = by. yy), verbs in, 104, 206. 
 fu-yjv (root yug), 80. 
 Iv/ti], 80.