College and Research Libraries


which is bound to aggravate continuously 
their catalogers and readers and to hinder 
them in their operations and services. But 
if Dr. Fuchs' commentary is not to stop the 
course of the revision in progress, it will con-
tribute to a sound revision. For aside from 
interpreting the instructions, Dr. Fuchs dis-
cusses throughout the commentary many 
cataloging issues which every code must con-
sider and resolve, and these discussions will 
benefit all who are interested in the prob-
lem of cataloging generally and particularly 
all who are engaged in the preparation or 
revision of cataloging codes. 

In his chapter on cataloging in Milkau's 
Handbuch der Bibliothekswissenschaft 
(1933) Rudolf Kaiser deprecates the 
thought that a catalog once well made can 
forever be continued, and says that "here, 
too, the saying is valid: every time has its 
catalogs." This is a time of widespread re-
examination of cataloging thoughts and 
methods, and in Germany as in other coun-
tries the time for a new catalog seems to 
have arrived.—Seymour Lubetzky, Library 
of Congress. 

Machine Literature Searching 
Machine Literature Searching. By James W. 

Perry, Allen Kent, and Madeline M. Berry. 
New York: Interscience Publishers, 1956. 
162p. |3.75. 
This volume reprints ten essays that origi-

nally appeared in American Documentation 
during the last two years and adds five new 
chapters. 

The first chapter deals with the general 
background material and outlines machines 
searching requirements; the second covers 
the intellectual problems involved in prepar-
ing material for machine searching and the 
types of indexing; the third, the conversion 
of indexes to make them more suitable for 
machines handling; the fourth, the methods 
used by the authors in collecting terms; the 
fifth, the method developed by the authors 
for attempting to systematize terminology for 
code development so that the specific terms 
will all be joined with generic broader terms; 
the sixth, the method for constructing a code 
to increase the effectiveness of machine 
searching; the seventh, the need for deter-
mining uses to be made of the information 

so that the level of headings can be made 
suitable; the eighth, definition of the opera-
tional criteria for determining whether a re-
trieval system is efficient and evaluation of the 
retrieval system; the ninth, the operational 
characteristics of searching machines includ-
ing such characteristics as ability to identify 
one or more patterns, interpretation of cer-
tain patterns denoting the beginning and 
end of an organized sequence, ability to han-
dle generic relationships as well as specifics, 
detection of logical relationships between 
criteria, etc.; the tenth and final essay in the 
reprinted group is a discussion of the factors 
underlying development of machine lan-
guage. 

Chapters 11-15 (pages 72-134) are the new 
materials in the book. Chapter 11 discusses 
the purpose of machine language, analytical 
and synthetical relations, machine language 
and machine capabilities, relationship of ma-
chine language to codes, methods of showing 
relationships by codes, etc. Chapter 12 covers 
the problem of coding of diagrams, of geo-
graphic areas, and of chemical structural for-
mulas. Chapter 13 handles the encoding of 
abstracts by reducing redundancy of words 
so that they may be searched directly by ma-
chines. Chapter 14 deals with searching strat-
egy and association-trails. 

The final chapter, entitled "A Look Into 
the Future," is quite general and concludes 
that there are "a number of theoretical and 
practical problems [that] will require care-
ful investigation. Even when utmost care is 
devoted to perfecting information processing 
methods, major investments of effort and 
money will be required to process research 
and professional publications so that they 
may be used most effectively. Such invest-
ment is apparently justified particularly in 
the field of science and technology." 

The basic assumption that underlies this 
series of studies is that we have machines 
capable of doing literature searches. The au-
thors state: "Modern automatic equipment is 
able to scan and recognize index entries . . . 
Scanning can be directed to one or simul-
taneously, several entries. Speeds of opera-
tion are such as to permit scanning and cor-
relating of generic and specific aspects of 
indexes in reasonable time" (page 1). As a 
matter of fact there are no machines in pro-
duction that will do even a fraction of what 

450 COLLEGE AND RESEARCH LIBRARIES 



is claimed here. The machine on which most 
of the authors' work was based was an ex-
perimental model of the IBM machine de-
veloped by Mr. Luhn, on which develop-
ment work was discontinued more than a 
year ago. The only others that might be con-
sidered as remotely approaching these claims 
are Eastman's Minicard, of which the first 
experimental model has not yet been com-
pleted, and the Rapid Selector, of which 
only a prototype has been built. 

Similarly, a second common thesis in pa-
pers such as these is that conventional classifi-
cation has become less and less effective in 
coping with the problem of a mounting mass 
of materials, because, the authors say, "Con-
ventional classification is characterized by the 
following feature. Criteria are combined in 
a rigid array as a basis for defining the sub-
divisions. Thus, for example, dyestuffs may 
be classified as to features of their chemical 
constitution." However this is not avoided in 
the authors' classification scheme. It is dif-
ferent from others, and the notation scheme 
is different from that of other classification 
schemes. But we have a broad concept fol-
lowed by a more specific concept and then by 
the next more specific concept in fixed array 
in their classification just as in what they 
term conventional classification schemes. 

Having based the argument largely upon 
the need for multi-dimensional searches 
which can not be made by a rigid classifica-
tions scheme, the authors go on to say: "An-
other question that must also be decided re-
lates to the degree of detail to be accorded 
consideration during indexing . . . In a 
strictly logical sense all these details must be 
included in our indexing if it is to be com-
plete. Practical considerations based on ex-
perience and common sense, however, must 
guide us in determining what degree of in-
dexing is to be carried through . . . It is 
obviously useless to index something to 
which a search will never be directed . . . 
There are a number of consequences that 
flow immediately from the fact that the pur-
pose to be served by the index is a determin-
ing factor in deciding how the indexing step 
is to be accomplished . . ." (page 8). Thus, 
starting with the basic argument that we 
need to have much more intimate indexing 
for multi-dimensional searching, we wind up 
with the argument that we must still use 

judgment in determining the level of index-
ing. It would appear, therefore, that the au-
thors do not subscribe to the theory that in-
dexing can be made detailed enough so that 
multi-dimensional searches, in all combina-
tions, can be done automatically by machine. 

Despite the fact that this whole series of 
studies is based on the theory that the 
mounting mass of recorded information is 
responsible for the lengthening time required 
for searching larger and larger indexes, etc., 
no evidence is given as to what is considered 
large, or large enough to justify machines. 

Another intellectual base for these studies, 
which is common in studies of this type, con-
fuses library book classification with classi-
fication of knowledge and points to the ob-
vious inadequacies of a system designed to 
put a physical object in place in the files as 
the reason for doing something different and 
radical. This passes over all the work that 
has been done, particularly in Europe, on 
classified indexes using conventional classifi-
cation schemes. 

It may very well be that the code or nota-
tion system developed by the authors is a 
better notation system for either machine 
or manual searching than are many other 
notation schemes. This has not been estab-
lished in practice and, in fact, there are no 
examples of successful experience with this 
scheme in large-scale operation. 

The combination of semantic factors and 
analytic relations, plus arbitrary assigned dif-
ferentiating numerals, described in detail on 
pages 81 to 90 is fascinating. But it is difficult 
for this reviewer to see how much space is 
saved by using MUSRMACHTWMP 03 for 
the word thermometer. It requires 40 spaces 
to write Springfield, Illinois, and 44 to write 
Chicago, so this notation would overflow 
from a punched card with only Springfield 
and Chicago needed on one card. The use-
fulness of a notation this long on a medium 
capable of storing a total of 80 characters 
only is open to question. 

This is a field in which this reviewer nor-
mally is made to feel exceedingly stupid by 
practically everything he reads, and that it 
is just as true of this book as of other ma-
terials he has seen. As in most cases in this 
general subject area there appear to this re-
viewer to be many things that are given as 
obvious conclusions which are not particu-

SEPTEMBER, 1956 451 



larly obvious. Assumptions as to the state of 
development of machines for information 
handling which are contrary to fact form 
the basis of this study (and many similar 
studies), and the conclusions frequently ap-
pear, in this as well as in other treatments 
of this subject, to represent unfamiliar and 
confused terminology rather than a revolu-
tion in information handling. 

Nevertheless the authors represent one of 
the major schools of thought in this part of 
the field of documentation and it is, there-
fore, probably a good thing to have the 
quintessence of their studies over the last ten 
years brought together in one volume.— 
Ralph R. Shaw, Rutgers University. 

C l a s s i f i e d 
A d v e r t i s e m e n t s 

Rate: $1 per line; 3-line minimum. Closes 
first of month preceding date of issue. 

BOOKS not obtainable from publish-
ers may be available immediately from 
our stock of over a million volumes or 
may soon be found by our Search 
Service. Send lists to Dept. CR, Barnes 
& Nobel, Inc., 105 Fifth Ave., New 
York 2, N.Y. 

COLONIAL BOOK SERVICE—Spe-
cialists in supplying the out-of-print 
books as listed in all library indices 
(Granger Poetry; Essay and General 
Literature; Shaw; Standard; Fiction; 
Biography; Lamont; Speech; etc.) 
Want lists invited. 23 East 4th St., 
New York 3, N.Y. 

BACK D A T E MAGAZINES 
M I D T O W N MAGAZINE CENTER 

1105 6th Ave., New York 36, N. Y. 

NEW! 
MICROFILM ABSTRACTS 

A U T H O R INDEX 
Covers vols. 1-11 of current title Dis-
sertation Abstracts. Order from Mrs. 
Martha Jane Zachert, 1760 Harbin 
Road, S.W., Atlanta 11, Ga. $2.00. 

Expert Service on 

MAGAZINE SUBSCRIPTIONS 
for 

ALL LIBRARIES 
* 

Faxon's Librarians Guide 
free on request 

* 

For the very best subscription 
service at competitive prices—ask 
about our Till Forbidden plan. 

* 

F. W . F A X O N CO., INC. 
83-91 Francis Street Boston 15, Mass. 

Continuous Library Service 
Since 1886 

W O R L D ' S L E A D I N G 
P R O D U C E R O F A U T H E N T I C 
F O L K M U S I C O N R E C O R D S 

Music of over 200 peoples, including: 

American Indian • Arabia • Argentine 
Armenia • Australia • Bahama • Burma 
Canada • China • C o n g o • C u b a 
England • Eskimo • France • Greece 
Haiti • Hindu • India • Ireland • Italy 
Japan • Jewish • K o r e a • L i b e r i a 
Mexico • Negro • Nigeria • Pakistan 
Peru • Rumania • R u s s i a • S p a i n 
Switzerland • Yugoslavia 
Most of the issues are original record-
ings on High Fidelity and are guaran-
teed for quality of reproduction & 
content. 

For complete free catalog write to: 

F O L K W A Y S R E C O R D S 
1 1 7 West 4 6 t h St. • New Y o r k 3 6 , N. Y . 

452 COLLEGE AND RESEARCH LIBRARIES