Full text of "Papers presented at the ... Clinic on Library Applications of Data Processing, .."

See other formats

r~ 3,

PROCEEDINGS OF THE 1965 CLINIC
I ON LIBRARY APPLICATIONS
OF DATA PROCESSING

Graduate School of Library Science
University of Illinois

PROCEEDINGS OF THE 1965 CLINIC ON LIBRARY
APPLICATIONS OF DATA PROCESSING

Held at the Illini Union on the

Urbana Campus of the University

of Illinois, April 25-28, 1965

Edited by
FRANCES B. JENKINS

Distributed by

The Illini Union Bookstore

Champaign, Illinois

FOREWORD

The University of Illinois Graduate School of Library Science
in cooperation with the Division of University Extension sponsors an
annual Clinic on Library Applications of Data Processing. This
volume presents the papers given at the third of these clinics which
was held in the Illini Union Building, on the University of Illinois
campus, on April 25-28, 1965.

As Herbert Goldhor pointed out in the preface to the proceed-
ings of the first of these clinics, in helping to meet the challenging
problem posed by a mounting volume of publication combined with the
increasing demand for factual information:

"the Library School's role in this area is seen as half-way be-
tween the theoretical work being done by those at the frontier
and the needs of the practitioners in the field. It would be un-
fortunate if there were no one doing the difficult and necessary
exploration of new ideas and of possible solutions; it would be
equally unfortunate if there were no one next in line to inter-
pret this work and to pass on its results to those who will have
to apply them in practice. The enthusiastic response of those
who came to this Clinic would bear out the correctness of the
position.*

The attendance at the third Clinic continues to reflect the grow-
ing number of doers in library automation. Approximately sixty
colleges, universities, public, state, and special libraries were rep-
resented by one to four librarians at this Clinic. These libraries are
either already using a computer or are planning to use one in the
near future.

As in other Clinics, representative types of libraries were
selected which were thought to have automated sufficiently to make
their experiences of value to other libraries. An appropriate person
from each of the selected libraries was invited to present a report
of the library's activities in this area. In addition to the papers
recording the experiences of individual libraries there are three
general papers, one by C. Dake Gull on the present state of the art,
one by Arthur Brody on the application in a library-oriented industry,
and one by William Greiner on library applications of UNIVAC
equipment.

iii

Responsibility for planning the Clinic rested with a committee
of the Graduate School of Library Science faculty composed of Direc-
tor Herbert Goldhor, chairman, Professor Holland E. Stevens, and
Professor Frances B. Jenkins. This committee had the able assis-
tance of Mr. Hugh Davison, Clinic Supervisor, Division of University
Extension, and Mrs. Bonnie G. Noble, Graduate School of Library
Science. The progress and conduct of the Clinic was the work of
many hands. First a word of thanks from the committee to the other
members of the faculty of the Graduate School of Library Science
with special thanks to Mrs. Barbara Donagon and Mrs. Nancy Works
for the preparation of these papers for publication; and a word of grat-
itude to the staffs of the Division of Extension and the mini Union for
their cooperation. But most of all a word of appreciation to the
speakers and registrants, those librarians who are making library
automation a reality in their institutions, without whose eager and
enthusiastic participation the clinic could not have been a success.

Frances B. Jenkins,
Editor

OTHER VOLUMES IN THIS SERIES

Proceedings of the 1963 Clinic on Library Applications of Data Pro-
cessing. Pp. 176.

Proceedings of the 1964 Clinic on Library Applications of Data Pro-
cessing. Pp. 117.

Order from the Illini Union Bookstore, 715 S. Wright St., Champaign,
m. 61803. $2.00 in paper covers, $3.00 in cloth covers.

TABLE OF CONTENTS

Page
FOREWORD iii

THE PRESENT STATE OF LIBRARY AUTOMATION

C. Dake Gull 1

THE DATA PROCESSING PROGRAM IN OPERATION AT THE
SUFFOLK COOPERATIVE LIBRARY SYSTEM, PATCHOGUE,
NEW YORK

Walter W. Curley 15

870 DOCUMENT WRITING SYSTEM OF IBM CORPORATION
IN THE LIBRARY SECTION, LOS ANGELES CITY SCHOOLS

Mary Seely Dodendorf 43

BRO-DART INDUSTRIES' EXPERIENCE WITH ELECTRONIC
DATA PROCESSING

Arthur Brody 65

HOW TO DESIGN DATA PROCESSING INPUT RECORDS FOR
OPTIMUM RESULTS

John P. Kennedy 79

FLORIDA ATLANTIC UNIVERSITY LIBRARY

Edward Heiliger 92

THE ONULP BIBLIOGRAPHIC CONTROL SYSTEM. AN
EVALUATION

Ritvars Bregzis 112

COMPUTER INDEXING OF THE PRESIDENTIAL PAPERS IN
THE LIBRARY OF CONGRESS

George R. Perreault 141

LIBRARY APPLICATIONS OF DATA PROCESSING
EQUIPMENT AT ABBOTT LABORATORIES

Walter A. Southern 156

DATA PROCESSING EQUIPMENT AND THE LIBRARY

William E. Greiner 175

LIBRARY APPLICATIONS OF DATA PROCESSING:
A SELECTED BIBLIOGRAPHY, 1963-64

James Krikelas 193

THE PRESENT STATE OF LIBRARY AUTOMATION
A Study in Reluctant Leadership

C. Dake Gull

This discussion will cover a rapid survey of the various un-
dertakings in library automation during 1964-65, a brief sketch of
how automation reached our national libraries, and the uses to which
it has been put, including this author's views concerning the role
these libraries must play in the future of automation in all libraries,
and some actions worth considering to insure that national libraries
fulfill their central and crucial role in the nationwide system as well
as in the process of automating libraries.

Library automation in April 1965 affects only a very small part
of the activities of libraries in the United States, for attempts to
automate have been undertaken in only a very small number of librar-
ies in comparison to the total, and only for a limited number of op-
erations in those libraries. In spite of this circumstance, the amount
of activity directed towards library automation is already so large
that accurate information is not available on how many libraries
have operative programs, and even less information is available on
how many libraries may be considering the automation or mechani-
zation of some of their operations. There is not even an organized
and continuing effort to gather information about library automation
and to publish it for the profession.

These conclusions, consequently, are based almost entirely
upon secondary sources because it was not possible to visit most of
the operations which will be described and summarized. The neces-
sity of relying upon written descriptions and conversations and the
failure to visit most of the installations should lead to suspicion that
there are inaccuracies and misinterpretations in this presentation.
Time and careful studies will uncover them. One lesson learned
thoroughly from efforts to introduce automation and to provide ob-
jective, balanced, and satisfactory consulting and engineering ser-
vices to libraries is that the reality of mechanization and automation

C. Dake Gull is Professor in the Division of Library Science, Indiana
University, Bloomington, Indiana.

in a library, as a direct experience, is likely to be quite different
from the impressions gained from conversations and written de-
scriptions. In spite of this, and in addition to the warnings about the
unreliabilities on which the conclusions may be founded, an attempt
will be made to present what the author believes to be a reasonable
selection and interpretation of the library automation which exists
at the present time. The author is very much indebted to many
librarians for the up-to-date information they have sent, because the
accounts in our literature are too often out-of-date on publication.

There are a few doers who are making library automation a
reality in their libraries. A larger number of librarians are de-
signing and planning, and an even larger number are interested or
are watchfully waiting to see what other libraries, especially our
national libraries, will do. It is fortunate that so many of the doers
in library automation have prepared papers for this Clinic.

The summaries the author has seen and has roughed out him-
self show that information retrieval (IR) is the least common of
library automation activities. Information retrieval is more common
in documentation centers than in libraries, a distinction which ac-
curately reflects their greater needs for IR.

The operating examples of library automation are found in
acquisitions and order work; cataloging, in the production of book
catalogs and card catalogs; serials work, in most of its aspects^ and
circulation records work. The introduction of automation for ac-
quisitions and order work improves those procedures internally and
offers better financial control. The ready availability of new catalogs
in book form, as yet of less than 100,000 titles, in cumulated series
or in new editions, serves our library users better than before,
because of their currency and easy distribution in published forms.
The use of automation for circulation and serials records has im-
proved service to library users, in an area where there is much work
to do. Daily circulation lists organized in two parts, one by book
call numbers and the other by borrowers' identification numbers,
and lists of reserved or otherwise displaced books are invaluable
aids, especially when they are in multiple copies and can also be
used for immediate inventory control. Weekly lists of newly-arrived
serials, showing all the pertinent facts about individual issues, are
a great help in serving users, especially when supported by updated
holdings lists of all serial titles.

The following equipment is currently being used:

1. Most of the standard units of punched card equipment.

2. Paper tape typewriters as peripheral computer components,
or as independent units for catalog card production.

3. A very representative sample, as to models and manufac-
turers, of small, medium, and large electronic digital computers.

4. Magnetic tape units— these are much more common than
random access units for storage, recall, and memory operations.

5. A small number of terminals connected to computers; even
fewer at remote locations.

6. Computer line printers in all capitals are in the majority;
a few upper-lower case computer printers chains are used;
computer -driven composers for photo -off set work are coming
in.

Operations research and engineering personnel are being used
at Purdue University, Johns Hopkins University, and Massachusetts
Institute of Technology to help solve their library problems. This
shift to reliance on professional persons other than librarians and
computer manufacturers is very significant.

One of the very interesting applications is the effort under way
at the Medical Libraries at Harvard, Yale, and Columbia Univer-
sities, as described by Frederick G. Kilgour, Librarian of the Yale
Medical Library. * The objective of these three libraries in their
computerization project is eventually the rapid and complete retrieval
of bibliographic information. However, they are starting with the
catalog, entering information for books in their collections published
in 1960 or later, and to this monographic record they hope to be able
to add the MEDLAR'S indexing which cares for the indexing of
journals supplying upwards of 75 percent of the recorded use in their
libraries. They intend to have each library connected to the com-
puter by telephone lines and to be able to ask through remote infor-
mation terminals for references on various subjects, with specific
limitations on the questions. Hopefully, the references will be
supplied to the inquirer almost immediately. The effective speed of
response will be limited only by the ability of the terminals to print
out references.

They are starting with preparation of catalog cards by machine
operations now and will go to information retrieval later ; use is
being made of the medical subject headings used by the National
Library of Medicine. The depth of their subject cataloging has al-
ready increased from approximately 1.5 subject headings per title
to 10.4. The production of catalog cards is designed around the IBM
1401 computer, because this is a very common computer at the
present time. The four major computer programs are used to allow
the production of punch cards to be placed in the IBM 870 Document
Writer to produce catalog cards from continuous forms. All of the
different entry cards are prepared for the libraries in this way, in-
cluding their contributions to the National Union Catalog. When the
machinery is operating perfectly, the cataloging completed one day
is punched the next day and the mechanized catalog cards are pro-
duced for filing the following day.

Once a month at Yale the accessions list is added to the Bulle-
tin of the Yale Medical Library. This process has been reduced
from one week of staff time to an hour of time in the computer center,
of which only a very small fraction is computer time. The average
production cost is now in the vicinity of $20 a month, and the saving
in the human cost has been far greater, for staff members disliked
the re -editing required. The accession list is now approximately
50 percent larger than it was formerly and costs less to prepare.

The project directors have endeavored to ascertain the cost of
producing cards, and it appears at the present time that mechanized
catalog card production does not cost more and probably costs less
than other conventional card production techniques. When an addi-
tional use is made, as in the case of accession lists, the costs drop
far below those for conventional procedures. The project is only at
a beginning stage. They intend to produce the cards on a computer
using upper and lower case characters, and this change will ma-
terially speed the card production to approximately one card every
two seconds.

The National Agricultural Library (NAL) has been actively
surveying its needs for automation in recent years, and a report on
its efforts, Project ABLE, is to be published very soon. The NAL
has already made a modest start. The August 1964 issue (and sub-
sequent issues) of the Bibliography of Agriculture contains an author
index produced with the aid of a special typewriter font (capitals and
numerals), a Farrington optical scanner located in New Orleans, and
IBM 7074 and 1401 computers in Washington. Input errors made in
typing the authors in random order can be lined out or indicated by
this symbol: I-. The optical scanner is used to produce magnetic
tape records, the alphabetic sorting is done on the IBM 7074, and the
printing is done on the IBM 1401.

The NAL has already advanced towards the preparation of the
subject index for the Bibliography of Agriculture, and its computer
production is planned for publication in the January 1966 issue. Con-
currently, the NAL is preparing for the retrieval use of the subject
index.

The NAL has also started, with Vol. 1, No. 1, March 19, 1965,
the biweekly Pesticides Documentation Bulletin. The first issue,
computer produced, contains 1,289 entries, an author index, and a
permuted keyword-out-of-context (KWOC) subject index, in which
each index term is in the left margin as well as in the body of the
title. On the same date, the NAL invited potential bidders to submit
proposals for a computer based systems design and programming
for a Pesticides Information Center.

The Library of Congress (LC) has responded to the need for
automation within the last year by establishing the position of In-
formation Systems Specialist, to which Samuel S. Snyder has been

appointed. He has had much experience with computers. His staff
includes Barbara Evans Markuson, his assistant; Donald M. Ricker-
son, electronics engineer, and Henriette D. Avram, an experienced
programmer.

Their plans are under way. They are beginning a systems
analysis of the LC and are gathering data using six of LC's latest
interns. They are working on the machine readable record problem
as related to catalog cards, the subject heading authority list, National
Union Catalog entries, and a possible subscription service corre-
sponding to current catalog card sales. Samuel Snyder hopes that
his office can become an information exchange on library automation
and that some internal memoranda can be published later in a pro-
fessional journal, but he wrote that he could not release the amount
of money requested for the 1966 fiscal year budget.

The preceding survey shows that none of the libraries which
have adopted mechanization or automation to any extent has attempted
seriously to consider a whole library as a total information system
which is to be integrated so far as possible in its concepts, design,
provision of equipment, and daily operations. That kind of integra-
tion is certainly in the minds of the top staff members at the National
Library of Medicine, the National Agricultural Library, and the
Library of Congress, but as yet only as intellectual aspirations. The
King Report2 and the Proceedings of the Airlie Conference, 3 re-
markable reports in themselves, provide us with an adequate real-
ization that an integrated, electronic library system is a monumental
undertaking which should not be embarked on lightly, but one for
which there is every prospect of eventual success.

The foregoing examples in various libraries show diversity
in operations, equipment, programming languages, philosophic ap-
proaches to the problem, and methods of implementation. The amount
of compatibility from one system to another is unfortunately low. In
this lack of compatibility, a repetition of the divisive events in
library history can be seen. Unless there is a change, it is reason-
able to predict that librarianship and library users will continue to
suffer from the failure to standardize, cooperate, and centralize.
Evidently, the spirit of independence, of seeing one's own problems
and solving them one's own way, still prevails to the detriment of
librarianship as a whole. This condition demonstrates that there are
too few librarians in this country who are willing to rise to the op-
portunity of leadership presented now and to work for the benefit
of libraries, librarianship, and library users. Many librarians pre-
fer to work on their own restricted problems and solutions. It is
this prevailing situation which led to the adoption of the subtitle,
*A Study in Reluctant Leadership," for this paper.

For twenty years it has been clear to a number of librarians,
most of whom are still active in this work, that first mechanization,

next computers, and more recently the restricted field of informa-
tion systems engineering, offer the only real hope that libraries will
be able to cope with the ever-growing problems of acquiring, analyz-
ing, controlling, manipulating, and distributing information. In re-
retrospect, that period seems much too long for what has been
accomplished.

The first stimulus to action came from outside the library
profession. The Office of Naval Research at the close of World War
II saw the need to do something with the mass of research and
development reports which were left over from the war effort. The
Office placed a contract with the Library of Congress to undertake
research and development for the better handling of this material,
and turned over the mass of reports. That backlog and the growing
bulk of material produced by our enormous defense program side-
tracked the research and development aspects, and the problems
became almost wholly operations at LC, rather than research, as can
be seen in the name changes which occurred. The Science and
Technology Project at the Library of Congress became the Navy
Research Section; in turn it became the Reference Center of the
Armed Services Technical Information Agency (ASTIA) when that
agency was established under the order of the Secretary of Defense
in 1951.

In 1952, ASTIA placed a contract with Documentation Inc., a
new commercial research and development organization, seeking an
improved method of analyzing the content of the report literature
which would be amenable to manual and mechanical or electronic
operation. The result was the Uniterm System of Coordinate Index-
ing. There are a number of manual and electronic installations now
using this system, attesting to the success in meeting the operational
requirements of the contract. Controversy arose immediately over
fragmenting index terms in this manner; the controversy was once
violent and is perhaps only quiescent now. There were other research
and development efforts during the early 1950's, as well as ever
widening study by people concerned with the problems.

Another federal government agency embarked upon some con-
tracts which eventually led to a markedly different development at
the Battelle Memorial Institute and subsequently in the newly estab-
lished Center for Documentation and Communication Research within
the School of Library Science at Western Reserve University. The
result of these efforts was the even more controversial telegraphic
abstracting style of subject analysis based on some traditional ideas
of classification and a new concept of semantic factoring. Telegraphic
abstracting proved to be one of the most complicated methods ever
developed for the control of information. It was based on some firm
belief, and weakly supported by experimentation, and is used today
for only one sizable effort, the control of metallurgical information

by the American Society for Metals. One cautious report suggests
to me that the development has turned out to be metallurgy's greatest
obstacle in the past decade.

The relative measure of success of these methods, coordinate
indexing (extremely simple) and telegraphic abstracting (extremely
complex) remains unresolved at this time. The simplest and the
most complex methods are both being implemented with computers
at the present time. Perhaps the most useful conclusion is that
computer is larger than either of the systems, and we are fortunate
that this is so, because we have the opportunity to develop other
useful methods within the extremes.

During the period that the Library of Congress administered
part of ASTIA, it participated in the developments which were leading
toward automation in libraries. However, the ever growing require-
ments of the Library of Congress for space and other considerations
led eventually to the removal in February 1958 of ASTIA activities to
Arlington Hall across the Potomac in Virginia, where ASTIA installed
a computer in 1960 for its operations. This change removed the
Library of Congress from the mainstream of developments leading
toward the adoption of computers. Computers were adopted earliest
and most willingly by documentation and information centers.

Within a year the Librarian of Congress, L. Quincy Mumford,
let it be known that small surveys about the possibility of automation
for the Library of Congress would be welcome and that the Library
of Congress would consider unsolicited proposals for work. Richard
S. Angell, Chief of the Subject Cataloging Division, was chosen as the
official contact and soon demonstrated that he was the LC's most
knowledgeable member in this area.

This act of leadership produced mixed results. Three com-
panies, General Electric, International Business Machines, and
Thompson- Ramo-Wooldridge, made brief surveys and in October
1959 presented proposals for initial work, thus preparing their per-
sonnel for later work elsewhere. This preparation was the principal
benefit.

The proposals made clear to the top administrators at the LC
that automation would eventually be feasible and successful, that an
integrated electronic library system would be complex and difficult
to achieve as well as costly ($20,000,000 over a twenty year period
as a minimum), that the catalog records could be computerized to
provide a new level of bibliographic apparatus, that the Card Division
operations were particularly suitable for early automation, that
inquiry stations should be developed, and that LC needs exceeded the
capabilities of the existing equipment in some areas. The proposals
made clear that the LC would have to undertake its pioneering effort
to solve its problems with the aid of a branch of engineering which
was also still in its initial pioneering stages.

Modest sums were quoted by the companies for the initial
studies, about 1 percent of the Library's annual expenditures. This
figure is a low percentage for innovation in contrast to figures of
3 to 5 percent for this country's leading innovating corporations and
10 percent for research and development in the Federal Government.

The proposals were politely and gratefully accepted, but no
funds were obtained from the Congress and no contracts were let.
Because of their proprietary nature, the survey proposals were never
published, and the information given to the library profession was
limited to a brief account in the Library of Congress Information
Bulletin.4

The unfortunate result of the proposals was to paralyze the
spirit of innovation at the Library of Congress just when encourage-
ment and action were required. Libraries are still paying the penalty
and probably will continue to pay for years to come, because of in-
ordinate delays in attacking the problems of our national bibliographic
apparatus.

In contrast, the history of the National Library of Medicine in
1958-1965 provides a remarkable demonstration that the spirit of
innovation, decisive administrative action, thorough homework,
an outside contractor, and available technology can be combined to
accomplish a great deal towards automating a library.

The Library of Congress chose a reasonable but regrettably
inadequate course of action in 1960-61. The course was to seek
high level advice without other action. Even the offical contact was
changed and became Henry J. Dubester, Chief of the General Refer-
ence and Bibliography Division. He soon became LS's other expert
in library automation; he was later to leave the Library for the
National Science Foundation in 1964.

Again the result was delay. The blue ribbon study committee
produced sound advice, but it took too long to publish the King Report
(January 1964)2 and the Proceedings of the Airlie Conference (Oc-
tober 1964) ;3 and no implementation of automation has been accom-
plished as yet at LC.

No one should argue against the desirability of securing full
and competent advice for the automation of the Library of Congress.
The point is that LC could have obtained action as well as advice
during the past five years. While a Congress accustomed to large
appropriations for research and development in all areas of federal
activity is unlikely to reject a request for R&D funds for its own
Library, it is unlikely to appropriate funds without being asked by the
Librarian or without being urged to appropriate funds by the library
profession. We can assume very reasonably that if LC had asked
for funds five years ago, at least some significant operations would
be computerized at LC now.

In support of this study, a summary of the accomplishments of
the NLM for this same period should be noted. With a grant from the
Council on Library Resources in the period 1958-1960, Seymour
Taine and Frank B. Rogers, then Director of the NLM, studied the
problems of compiling, indexing, and publishing Index Medic us, and
converted its production from a manual card shingling operation to
a paper-tape typewriter, punched card, automatic camera production
method. They also defined what they wanted to accomplish in a very
complete fashion, formulated a public request for proposals, secured
twenty-six proposals, selected a contractor— the General Electric
Company— secured the financing of a $3.4 million effort from the U.S.
Public Health Service, monitored the contractor's work, and con-
verted Index Medicus to computer production in January 1964. In the
same period, they also expanded the coverage of indexed journals,
more than tripled the depth of indexing, revised the list of subject
headings, reduced the productions time of Index Medicus, and secured
by a development contract the most sophisticated computer-driven
photo-composer yet in use. They also kept their secondary objectives
always in mind, some of which are being worked on now.

The present Director, Martin M. Cummings, summarized the
accomplishments of MEDLARS up to March, 1965, as follows: Since
January 1, 1964, with MEDLARS, they have:

1. Organized and composed 15 monthly issues of Index Medicus
(10,476 pages) , and the 1964 annual volume of Cumulated Index
Medicus (5,698 pages).

2. Produced 300 demand retrieval searches for experimental
systems testing.

3. Produced more than 1,000 demand bibliographies for
physicians and scientists.

4. Produced the following recurring published bibliographies:

Cerebrovascular Bibliography (semi-annual)

Index of Rheumatology (bi-weekly)

Index to Dental Literature (quarterly, cumulative).

5. Produced magnetic tapes for use by others and in other
locations; tapes are being made available in IBM as well as
Honeywell format to increase compatibility for nationwide use. 5

NLM is now engaged in a computer programming to accomodate
monographic entries. Scott Adams, Deputy Director, has also written:

". . . there are two major strategic considerations underlying
NLM planning:

1. NLM intends to share with the medical library community
the power to search for and repackage citations in response to
interdisciplinary needs we have developed through MEDLARS.

2. NLM intends to establish, through Federal grants and con-
tracts, a medical library technical assistance program, to
strengthen the medical library network so that its resources
and services . . . (will become more nearly equal) to the com-
plexity and volume of new interdisciplinary needs, and to the
power of retrieval provided by MEDLARS. "6

While library automation in its present state is only poorly
advanced and is still in its pioneering stages, this development has
brought forward again a number of the problems which are central
to librarianship in this country. In the activities which are presently
being carried on, there exists a diversity of philosophies, methods,
techniques, and systems, and a variety of equipment already in use.
Examples of individual leadership in individual libraries are evident.
We recognize that the status of our national bibliographies is still
central to the problems; witness the efforts of Verner Clapp to
develop the report by Larry Buckland.^ But there still remains a
lack of standards for this bibliography and for other library prac-
tices. There is a considerable degree of incompatibility among the
new systems being introduced within libraries. There is a very
large lack of understanding of the capabilities and potentialities of
electronic information systems within the library profession. The
major lacks which have been brought to the surface by the arrival of
library automation are our lack of national leadership in librarian-
ship and the lack of a plan for automation.

The present situation must be translated into needs for some
kind of equipment and for increased finances to accomplish the work
that must be done. The engineers claim that equipment can be built
for almost any requirement if the requirement can be adequately
defined. It is known that increased finances can be obtained for
worthy causes if they are presented carefully and persuasively.

Most of the conditions described immediately above prevailed
at the time Herbert Putnam became Librarian of Congress in 1899.
He overcame many of the problems of his period by a series of
actions within the Library of Congress, by his effective relations
with the Congress, and by his leadership for the profession in this
country. He was not faced with problems of the same quantity as
we are facing today, and his library was relatively more the national
library than the Library of Congress is today. In the sixty-five
years which have elapsed, the Agricultural and Medical libraries of
the Federal Government have grown tremendously and have both
achieved legal status as National Libraries, while it has been im-
possible for the Librarians of Congress to achieve the same formal
elevation for the Library of Congress. Herbert Putnam did not face
the technology and its capabilities which we face today, but he did
face the absence of national leadership and the lack of a plan, just
as we face these today.

If the profession were at this time in the situation of having a
newly appointed Librarian of Congress, it might be reasonable for it
to decide to let the newly appointed Librarian endeavor to provide
national leadership and work towards a national plan for automating
the libraries in this country; but this situation does not prevail. The
Librarian of Congress, L. Quincy Mumford, has been in office since
1954, and his administration has been marked by the steady growth
of the library, a corresponding but less rapid growth in its financial
support, and a very cautious approach to changes of all kinds. All
of the circumstances that have been considered suggest that Mr.
Mumford truly needs our help at this time, and even more, help from
outside the library profession.

What are some steps which could be considered to offer the
type of help necessary to provide national leadership and to develop
a plan for library automation ?

A Federal Library Committee has been appointed recently,
with Mr. Mumford as chairman, and has held its first meeting. The
establishment of this committee is clearly a step in the right direc-
tion, but the question must be asked as to whether this committee
can properly be concerned with the problems of library automation,
nationwide and for the library profession, as it should be. The
cautious record of the chairman and the restricted phraseology of
the name of the committee suggest that too much hope for the solu-
tion of library problems should not be placed in the creation and
future activities of this committee.

The American Library Association (ALA) is the most broadly
based of the library professional societies and the largest in its
membership. It has demonstrated a continuing interest in the
Library of Congress, in all its national activities over the past eighty
odd years. The history of those years also shows that the American
Library Association has had relatively little effect upon the develop-
ment of the Library of Congress and upon the development of our
national library system in general. Only in recent years has the
ALA started a program of lobbying before the Congress, and this
program is now beginning to show results.

This author inquired by correspondence of the chairmen of
several committees in the ALA who are concerned with the automa-
tion of libraries in several aspects, trying to learn whether ALA is
presently exerting leadership in this area, and if not, how it should
go about developing leadership. From the replies I conclude that
ALA has no effective leadership in this area, and these committee
chairmen offer no concrete suggestions that will get ALA started in
this direction. Consequently, there do not appear to be any particu-
larly optimistic reasons for belief that the ALA is the direction to
which the profession should turn for assistance in this matter.

One solution, sometimes used in this country to establish
national leadership and to develop a plan, is the establishment of an
independent non-profit corporation, designed to accomplish the ob-
jectives of the incorporators. Perhaps a title such as "Americans
for the National Libraries of the United States" would be a suitable
name for this type of organization. The creation of an instrument
of this kind is so familiar that it seems unnecessary to describe it
further except to say that its objectives should be to determine re-
quirements, obtain financial support, influence the Congress, and
persuade or lead librarians toward library automation.

The choice of the plural for "the National Libraries" is deliber-
ate here. We can no longer afford to allow the historical accident
that there are three national libraries, one in the legislative branch
and two in the executive branch of our Federal Government, to hold
up further progress toward a national library system making full use
of library automation.

There are several organizations already in existance in our
society which might be able to serve the purposes we have in mind.
These are non-profit corporations concerned with one or more of the
intellectual disciplines and the application of those disciplines to the
general welfare of the population. These organizations include the
National Academy of Sciences— National Research Council, the newly
created National Academy of Engineering, the Social Science Research
Council, and the American Council of Learned Societies, for example.
With slightly differing charters, these organizations are in position
to work together for advisory committees to debate the requirements,
determine the needs for funds, lay out the requirements for national
leadership, and bring information to the attention of the Congress.
They are also capable of receiving funds for the purposes of con-
ducting surveys and developing plans. All of the organizations named
as examples have been and are being used for this purpose.

The problems of library automation are of such magnitude and
so pervasive in their national effect that the possibility of the estab-
lishment of a Federal Commission for the National Libraries ought
to be considered. Federal commissions have been most effective in
many areas in accomplishing the objectives of informing the Con-
gress, developing plans, and changing the direction of events in this
country. Such a body can be an ad hoc or permanent group, em-
powered by statute, resolution, or executive order to investigate a
problem in the public interest. It can secure information basic to the
framing of new legislation. It informs public opinion on matters
under inquiry; it can gain public support for new legislation.

A federal commission can be a legislative body or an admin-
istrative committee; it may hold oral hearings in Washington or
throughout the country. It can be empowered to compel testimony.
The body can be directed to report at any time and to make a final

report at a stated date. Majority and minority reports are always
possible to reflect differences of opinion. There need be no limits
to the subjects under investigation. Expenses of the federal com-
mission can be met by special appropriation or from the contingent
funds of the House and Senate.

George B. Galloway, at one time a member of the Legislative
Reference Service of the Library of Congress, has written that the
results of federal commissions have justified their use.8 They have
secured exposure of officials and set cautionary examples for others.
With their aid inefficient officals may be removed or forced to resign.
He also observes that the Congress may transfer neglected duties to
another department or create a new agency.

James W. Fesler has expressed the hope that the establishment
of a federal commission is not a concession to heavy pressure for
action while at the same time its appointment postpones a decision.^
One of the real advantages in the establishment of a Federal Com-
mission for the National Libraries is that its membership would be
composed of a minority of librarians and a majority of members of
other occupations in this country. The problems are so great here
that we need some unusually competent people drawn from the ranks
of politicians, financiers, industrialists, scientists, and executives,
as well as representatives from the academic world to take care of
the humanities and the arts.

REFERENCES

1. Kilgour, Frederick G. "Development of Computerization of
Card Catalogs in Medical and Scientific Libraries." In Herbert
Goldhor, ed. Proceedings of the 1964 Clinic on Library Applications
of Data Processing. Urbana, University of Illinois Graduate School
of Library Science, 1965, pp. 25-35.

2. King, Gilbert W. Automation and the Library of Congress.
Washington, Library of Congress, 1963.

3. Markuson, Barbara Evans, ed. Libraries and Automation
(Proceedings of the Conference on Libraries and Automation held
at Airlie Foundation, Warrenton, Virginia, May 26-30, 1963, under
the sponsorship of the Library of Congress, National Science Foun-
dation, and Council on Library Resources, Inc.) . Washington,
Library of Congress, 1964.

4. U.S. Library of Congress. Information Bulletin, 23:33-36,
January 27, 1964.

6. Adams, Scott. "The Scientific Revolution and the Research
Library," Library Resources and Technical Services, 9:140, Spring
1965.

7. Buckland, Lawrence F. The Recording of Library of Con-
gress Bibliographical Data in Machine Form. (A report prepared
for the Council on Library Resources, Inc.,) . Rev. ed. Washington,
Council on Library Resources, 1965.

8. Galloway, George B. 'Investigations, Governmental,"
Encyclopaedia of the Social Sciences. New York, The Macmillan
Company, 1937, VoL 4, pp. 251-259.

9. Fesler, James W. "Commission." In A Dictionary of the
Social Sciences. Ed. Julius Gould and William L. Kolb. New York,
Free Press of Glencoe, 1964, pp. 105-106.

THE DATA PROCESSING PROGRAM IN OPERATION AT THE

SUFFOLK COOPERATIVE LIBRARY SYSTEM

PATCHOGUE, NEW YORK

Walter W. Curley

February 10, 1963 will always be remembered by the staff of
the Suffolk Cooperative Library System as the day the button was
pushed. Prior to that time, all of the reasons why mechanized data
processing might be applicable had been carefully explored; all were
aware that a manual operation might keep the system alive and would
have the added advantage that, if it failed, the disaster would make
less of a noise from one end of the state to the other. However, the
reasons for mechanization were so compelling that the exciting step
was taken.

The creation of the Suffolk Cooperative Library System, the
second to the last one to be authorized in New York, had been
achieved only after many years of effort, and came hard to many of
the conservative but conscientious people who were running the
libraries of Suffolk County. In February 1963, only twenty-four of
the forty-five public libraries in the County were members, and
perhaps some non-members were waiting on the sidelines to watch
the System fail: ( l) because the libraries would be forced to develop
rapidly in order to cope with the demands of the fastest growing
county in the United States and (2) because it would be necessary to
keep operations confined to 5,000 square feet of space for at least
several years. Because the System could be spared the problems of
switching from an on-going manual operation and because it would
be necessary to show speed and accuracy while keeping operating
costs at a minimum, it soon became evident that the best procedure
would be to start mechanizing.

When the word got out that the System was going to travel this
rather revolutionary road, there was no trouble making contact with
representatives of the various companies which make and lease or

Walter W. Curley is Director, Suffolk Cooperative Library System,
Patchoque, New York.

sell the kind of machinery which would be needed. What was being
sought was the best possible equipment which the System could afford
to:

1. Handle the entire book ordering procedure.

2. Make labels for the spines, cards, and pockets of books to be
processed.

3. Allow plenty of down time from these operations for the
introduction of circulation control for the bookmobile and for
any member libraries desiring the service.

4. Provide further down time for inauguration of tighter con-
trols on existing operations, and for the introduction of new
services now still in the early planning stage.

5. Produce good catalog cards in quantity.

6. Produce catalog cards in a satisfactory manner.

7. Prepare accounts receivable and payable, as well as monthly
statements.

8. Provide vendor's invoices mechanically coping with the
problems of discount.

For various mechanical, financial, and personal reasons, the
following bank of equipment was leased under the following arrange-
ment with IBM: one 026 Printing Card Punch; one 082 S50 Sorter,
one 085 Collator, one 403 S50 Accounting Machine with Multiplying
device, one 514 Reproducing Punch with Mark Sense. For the pres-
ent, it was decided to produce catalog cards from a combination of
Ekatalith and the A. B. Dick Offset process printing cards, 4 up from
masters.

Although the establishment of procedures for the equipment had
been taking every waking moment of the staff for the previous
several months, it was noticed that no matter how fool-proof a pro-
cedure was thought to be, someone could always find a flaw in it
during those endless brainstorming sessions designed for this very
purpose.

When the equipment eventually was installed and the Boards
were finally wired as desired, a whistle was blown, pictures were
taken, a large sigh of relief was breathed, and promptly it all proved
to be premature! Some of the staff will always remember the happy
day when the data processing supervisor fed a large batch of cards
for the wrong member libraries into the IBM 403 during the billing
operation with the happy result that everybody was mad for at least
two weeks afterwards, as books arrived at some libraries without
bills, and, worse yet, bills arrived at other libraries for books they
had not ordered ! Lesson number one to be learned is that the human
element is always present and should not be underestimated even
in a mechanized operation. About all the machines can do in such a
case is insure bigger and better mistakes ! And then there was the

glorious week when order cards were fed into the machine by the
hundreds without noticing that they had not been properly coded for
the member libraries ordering the books in question. Some of the
staff worked into the very late hours for several nights after that
repositioning the code number for suitable mechanical handling. And
here was learned Lesson number two: No procedure is workable
until it has worked. After reading countless articles in library
journals, and listening to countless speeches at library conferences
about what the machines can do without a word of actual experience
or proof to back up the rosy speculation of the writer or speaker, it
seems that there is no other field of library science in which theory
is so easily substituted for practice.

Being able to start slowly, the System had the invaluable op-
portunity of erasing "bugs" while they were in a controllable stage.
At the start, fifteen of the member libraries ordered books from
the System; then others were added. Today the System orders and
processes books for all but two of the forty member libraries.

Over the past two years, changes in procedures have been
made from time to time to correct problems which, in the beginning,
seemed almost insurmountable. For example, producing clean
catalog cards, cutting edges, storing catalog cards, and making re-
finements in data processing procedures.

At the time that the bank of IBM equipment was installed, it
was known that Remington Rand was soon to come out with the Univac
1004. IBM had no comparable piece of equipment which would give
the speed and capacity of the Univac 1004, but the company was aware
that in this highly competitive area, they would soon have to provide
a "poor man's computer." Thus, the IBM 403 was installed with the
realization that its capacity would do very well for a time but that
eventually it would need to be replaced either with the Univac 1004
or with a similar IBM computer if that company should market one
in time. In 1964, it was determined that the IBM 403 was too slow,
the IBM 407 was fast enough but lacked the storage memory that
the staff believed was needed, and that the IBM 1400 series was a
very substantial leap to make from the IBM 403. The Univac 1004
uniquely fitted into the gap; for it was about as fast as the more
sophisticated IBM equipment, had storage, and fitted the System's
pocketbook at $1,150 per month.

The Suffolk System's approach to data processing has been
conditioned by the following considerations. In a cooperative sys-
tem which contains tiny little libraries as well as fairly large ones,
it is healthy to remember that a System Director has, in addition to
a nine member Board of Trustees, as many bosses as there are
member libraries. It seemed obvious then that the data processing
program would have to please the extremely varied needs of many
people. Providing them with sufficiently detailed information on

time and frequently enough would have been a formidable task if
handled manually. Also, a high rate of inaccuracy might have been
expected if more staff had been crowded into the cramped
quarters and pushed hard to produce or else. So mechanization
seemed most likely to insure the ability to please the "customers"
in a highly sensitive area.

Taking the long view of things, concern arose when it was noted
that the vast majority of centralized processing programs with which
the staff was acquainted depended on consolidated book ordering as
the primary basis for service. In a manual operation, this is almost
compulsory. It is believed that a few years of this will have an
alarming and debilitating effect on the book collections of the librar-
ies involved. In an area like Suffolk County where broad subject
coverage is essential, this procedure would not serve the purpose.
Furthermore, book selection should be regarded as one of the most
important and professional services which the member library
director performs at the local level. To subjugate this "sacred
trust" to the convenience of a centralized processing program seems
very much like the tail wagging the dog. With Suffolk's mechanized
program, member libraries have been able to order whatever books
they want, whenever they want them. In fact, the more variety the
better, because it is improving the scope and probably the quality of
the book resources of the System. The System has received a lot
of abuse because it does not and will not consolidate book orders.
Fellow librarians have said that the System will come to it "as sure
as God makes little green apples"; a major jobber said that unless
we did come to it like other systems, they would have to cut the
discount. To the first group, it can be reported that the machines
could not care less whether twenty copies of a book are ordered over
a four -week span, or twenty all at once; to the jobber the following
necessary reply was made: "Alright for you, or Suffolk will take its
trade elsewhere!" This was done not so long ago, and the change
has been an improvement all around. Incidentally, the new jobber
had seen the handwriting on the wall and had mechanized, allowing
his operation to dovetail with the System's in a mutually satisfactory
way. It is our belief that if the quality of service to the library
patron must suffer as a result of the trend toward centralization of
services, then a long hard look should be taken at the real reasons
why such centralization seemed necessary in the first place.

In working with the equipment, it was discovered that pro-
cedures which were developed for them should not hew exactly to
the traditional operations which are usually handled manually. Those
libraries which must make a switch-over from a manual to a
mechanized operation would have to be especially careful here. Many
of the steps in manual ordering, processing, circulation control, etc.

are not necessary or even advisable if machinery is going to do the
job. Also, procedures which are established for the machinery
should make use of the equipment's particular capacities. Other-
wise, it is interesting to note that machine production of a parallel
manual operation will be more expensive than the manual operation.

Another interesting thing learned in working with the data
processing equipment was that those button-eyed monsters tend to
intimidate people. Often in programming the operation, the staff
would find themselves enveloped in the most complicated maze of
extraneous procedure because there was the impression that the
"407 (or the 1004) could not do that but could do this!" After a few
rounds of such nonsense, it became clear that man should be the
master of the machine! It was also realized that a data processing
supervisor is trained to know his machines, but he does not know
library procedures and could not be expected to grasp the need to
produce certain results with a minimum of programming. Once this
was understood, operations began in a new direction. The program-
ming was done primarily in the administrative office, and once the
data processing supervisor knew what had to be achieved, he found
a way to make the machines achieve it.

The final decision which has guided the System has been a
resolution, widely announced and often repeated to the clientele, that
the System intends to make haste slowly, to expand services only
after it has been proved that the machinery can handle it. Each
phase of the program has begun with a few member libraries and
expanded to others as soon as the "bugs" were removed. No
promises have been made except to try to stay out of trouble. There
has not yet been a move into the final phase of the cataloging pro-
gram, that of cataloging older non-fiction. Hopefully that will begin
at the end of the year. For now, the member libraries and the pay-
ing non-member libraries which also utilize the service are very
understanding and expect full cataloging and processing only for
juvenile books, adult fiction, and adult non-fiction of the current
and past two years. Operations continue in 5,000 square feet of
space with a staff of forty-five. It is desirable that full processing
and cataloging of all books ordered not be undertaken until a move
is made into the new building, hopefully at the end of 1965.

Review of Current Procedures

A look at the actual procedures which have been developed for
the combination IBM- Remington Rand installation will give a clearer
picture of current operations of the Suffolk Cooperative Library
System. (See Appendix l)

The book ordering, processing, and billing program is as
follows:

The System's "customers" are supplied with a quantity of four-
part, tab-set order forms. The top or original copy is made of card
stock, and the three duplicate copies of vari- colored paper. For
each book to be ordered, one form is filled out at the member library.
The card-stock original comes to the System office bearing author,
title, publisher, the source in which the title can be verified, and the
library's code number. Where necessary, special information about
edition, library binding, etc., is also included. If the library desires
some limitation of the usual full cataloging and processing of all
books except older adult non-fiction, this limitation (for example
*DO NOT CATALOG") is stamped across the face of the order card.
Member libraries have been supplied with special stamps for this
purpose.

Once received at the System office, the order cards are sent
to the data processing department where the key punch operator
punches up pertinent information about author, title, vendor, library
code number, and date of order. Incidentally, the key punch operator
handles the order card for this particular book for this particular
library only once, and the key punch procedure is set up so that the
operator works mechanically, (it was learned the hard way that key
punch operators are trained to work in a particular way and should
not be expected to sort, separate, or handle in different ways order
cards contained in any one batch.) This information which the key
punch operator punches into a blank beige data card is reproduced
back into the original order card. This is done quickly and mechan-
ically by the IBM Reproducer. For each book ordered, the System
has two documents (or cards) with which to work.

Twice a week, all of the accumulated order cards are placed
in the Univac 1004 where orders to designated vendors are then
printed out on sheets. The names and addresses of the vendors are
contained on a tape file which is played into the Univac 1004 as the
order cards are run through. The orders are then mailed to the
vendors, and the original order cards are then sent to the Catalog
Department where they remain for five days, during which time the
Catalog Department is expected to make full use of them for the
first and last time, if possible.

Here the order card is checked against the authority file. If
the item has been cataloged before, all is well. A batch number is
placed on the order card which will tell the processing line where
a set of catalog cards for the book may be quickly found. ( These
card sets had been overrun in limited quantity the first time the
Catalog Department had handled this particular edition of this par-
ticular title.) If the Catalog Department has not cataloged the book

before, they will endeavor to do so during this five-day period of
grace. K they fail to do so, back go the cards at the appointed time,
and the Catalog Department can expect to receive the cards and the
books when they arrive from the jobber. At that point, it becomes
a top priority item in the Catalog Department, which cannot survive
any kind of log jam because of the existing space problem. To date,
they have stayed alive because they are a really top-notch cataloging
staff!

Order cards are placed in a tub file, arranged by order number.
To simplify matters, the order number is actually the date on which
the order is placed with the vendor. Within order number the cards
are usually filed by author. However experimentation is now in
process with one order arranged by title, and another filed by pub-
lisher, and then by title within publisher. At this point, it is believed
that speed and accuracy are best served by filing by title within order
numbers.

When the books arrive from the vendors, normally in seven or
eight days, clerks hand pull order cards for books listed on the
vendor's invoice, and sense-mark into the cards the list price and
discount for each. These order cards are then sent along to the Data
Processing Department along with the invoice. Here they are mark-
sensed by the IBM Reproducer, and the list price and discount are
punched into the card. The order cards are then placed in the Univac
1004 where a balance sheet is prepared for each order. The 1004
will now figure the discount against the list price, arrive at the net
price for each item, and will total the sum of the net prices, giving
the total net price for the bill owed to each vendor.

This net price on the balance sheet is compared against the
vendor's invoice. If the bill is correct, it is approved and placed
in a file for payment two months hence. Because of the "breaks" in
discount for each item, it is ever possible that a small adjustment
will be necessary. As is known, vendors deal in whole pennies and
will take for themselves one-third, one-half, or one-quarter of a
cent on items where there is less than an even break. When the
adjustment is insignificant, as it usually is, the System absorbs the
loss. After such an adjustment has been made in the bill, if neces-
sary, it is then approved for payment.

There are rare occasions when the adjustment figure is con-
siderably out of line. In these cases, it is necessary to backtrack
through the invoice and through the order cards eventually to dis-
cover where the error has occurred. Mark- sensing has proved to
be extremely accurate and to have cut operating time greatly. Any
other corrections which may be needed have been indicated by the
Order Department clerks previously mentioned and are made at this
time by the Data Processing Department.

At this point, all of the order cards for books received from
the various vendors are sorted alphabetically by author, placed in
the Univac 1004 which now reproduces an additional deck of cards
which are then interfiled into the Financial File. This file includes
cards for all books which have actually been received from the
vendors and have been approved for payment but for which the Sys-
tem has not as yet billed the libraries. This Financial File (or
deck) also serves as an inventory, just about, of the books which are
somewhere in process within the office. To some extent, it also can
serve as an Accounts Receivable file.

The order cards are next sorted by jobber and by title within
jobber, and both the order cards and the source documents are re-
turned to the Order Department. The order cards are inserted into
the books which they represent. If there are any books listed on
the invoice which have not been received in this shipment from the
vendor, a debit memo is made out and forwarded to the Data Process-
ing Department. For each book which has been received, the Order
Department's check- in clerk will look at the order card to see
whether the book is to be cataloged or just processed. A batch
number written on the card indicates that the book goes straight to
the Processing line.

As the Processing Department also handles final placement of
books on the shelves designated for each 'customer library," all
books actually go there if cataloging is not indicated. It may be
recalled that libraries sometimes request that a book not be pro-
cessed or cataloged; they stamp this message on the order card when
they send it into the System office.

The System, now in its second year of processing, is process-
ing more than 150,000 volumes per year. The Catalog Department is
handling about 25,000 titles a year. Thus, it should be noted that for
every book which goes to the Catalog Department, more than five
go dire Jtly to the processing line without stopping at the Catalog
Department.

Books which are sent to the Cataloging Department are, as
previously mentioned, top priority items because of the space prob-
lem, as much as for any other reason. Once they arrive there,
however, each is cataloged as quickly as possible and is assigned
the batch number for catalog cards which are now to be made in the
Machine Room. A Dewey Decimal classification number will also
be assigned, of course, if the item is non-fiction and will be mark-
sensed into the card. The cards for these "just-cataloged" books
will now be run through the IBM Reproducer, and the sense-marked
information will be punched into the card later allowing the cards
to be used for the production of the labels which are to be placed on
the spine of the book and on the book card and pocket .

Now, let us follow the books which are in the Processing De-
partment, arriving there either directly from the Order Department
or from the Catalog Department. On the day before the books are to
go down the processing line, one card is pulled from each book.
(Remember, there are two in the book.) The cards are kept in the
same sequence as are the books which are sitting on the Awaiting
Processing shelves. These cards are sent to the Data Processing
Department early the next morning, where labels are quickly re-
produced by the Univac 1004. When the cards and labels come back
from the Data Processing Department, the books go down the Pro-
cessing line, where labels are placed on the spine and on the book
card and pocket; the book pocket is affixed to the inside of the book's
back cover, and a plastic cover is placed on the dust jacket, if there
is a dust jacket. It should be mentioned that the label for the spine
of the book will give the Dewey Decimal classification number, if
there is one, plus the author's first two initials. The Univac 1004
automatically picks out the initials to be printed out. For Biography,
the labels must be hand processed. For Fiction, the label for the
spine shows only the first two letters of the author's last name. It
is possible if the library indicates a wish for it, to provide any two
symbols which may be needed. The labels for the book cards and
pockets list author, title, and call number.

Just prior to placing the processed books on the shelves to
await delivery, catalog card sets are pulled from the batches which
had been prepared the first time the Catalog Department handled
the titles in question. Each set is placed in a plastic container and
inserted in the pocket of the appropriate book.

The fully processed volumes, plus any others not to be pro-
cessed, now have been arranged on shelves by library. Each day
order cards are pulled from the books awaiting delivery on the
following day. (Daily delivery is maintained to most member
libraries.) For each library, there is a name, address, and accounts
receivable file. The order cards which have just been pulled are
sent to the Data Processing room where they are placed in the
Univac 1004, and invoices are run off for each library about to re-
ceive a shipment of books. At the same time, a packing slip is made,
and the accounts receivable file is updated for each of the libraries
involved.

Remember that financial deck which was quickly reproduced
some time ago to serve as a record of books received from the
vendor but not yet billed to the libraries ? It is imperative, of
course, that the System receives payment for each book which has
been purchased for member libraries. The ideal arrangement is to
have, at year's end, a payment received for each book ordered.
Again using the Univac 1004, the order cards for books just billed
are played to the receiving libraries against this financial deck.

This purges the financial deck of all cards for items just billed. In
effect, a balance for these items has been achieved, and the trans-
action is completed.

The order cards have now served their major purpose. At
this point, they are sorted by author and placed in a file which is
sent, every two weeks, to the Reference Department which uses the
cards to update the Union Catalog. In the near future, they will also
be used to update the authority file which will be a finding list as
well.

The books awaiting delivery are now checked against the pack-
ing slip, and both packing slip and books are placed in canvas library
delivery bags, formerly known as "coal bags," and off they go on the
next day's delivery run. Early in the System's history, the driver
one day noted that the bags contained unequal numbers of books , and
to ease his aching back, he reshuffled books from one bag to another
until he had achieved a nice balance. The phones had to be taken off
the hooks for a day or two after that. The human element again! It
does turn up!

Bills for delivered books go out the day after delivery. Each
shipment is invoiced. It is quite possible for any one library to
receive six, seven, or eight shipments a month plus as many bills.
It is our opinion that it is extremely important to handle it this way
rather than to accumulate bills. For one thing, when mistakes do
occur, it is far better to clear them up quickly than to try to do so
weeks after. Besides, the member libraries much prefer this modus
operandi because it keeps them absolutely up-to-date in their busi-
ness relationship with the System. A nice bonus is that under this
arrangement the System is paid much more promptly than if bills
were rendered only once a month.

And that describes the System's ordering and billing procedure.
It sounds complicated, but it works smoothly. Of course, there must
be many ways in which the procedures could be streamlined further.
Constant changes, which it is hoped are improvements, are made.
The System counts on further constructive ideas from those who can
view it with the perceptive eye of complete objectivity. If the in-
tention had been to develop such a procedure for manual implementa-
tion, the System would certainly have courted disaster. But the
beauty of a mechanized operation is that the equipment can produce
records, duplicate decks, check and cross-check information, in less
time than it takes to describe exactly what is happening.

The Latest Development— Magnetic Tape

In the fall of 1964, it was decided to expand the Data Processing
Department to include magnetic tape. With this additional equipment,

it was felt that catalog cards would be produced directly out of storage
by utilizing the Univac 1004. Also, magnetic tape would make it pos-
sible to improve existing operations by taking advantage of the vastly
increased storage of information which would become available. It
should now be possible, for example, to conduct bibliographic searches
of the authority file which will be placed on tape. Two Univac 1004
magnetic tape units have been purchased and arrived last month.

The primary reason for moving into tape was the urgency of the
need to produce those catalog cards directly on the Univac 1004. As
matters have stood up to this time, sets of catalog cards have been
stock-piled, using more and more of the 5,000 square feet of office
space. Also the production of the cards takes most of the running
time of the offset equipment and its operator who could be performing
other valuable services for the member libraries. A large amount
of time is lost to the Processing Department staff, because of the
necessity of searching out these catalog card sets from their storage
place in the System office. The idea of producing catalog cards from
the Univac 1004 as they are needed and in the same sequence in which
the books are to flow down the processing line has great appeal. A
special drum has been prepared by the Univac manufacturers which
will allow printings in both upper and lower case, an improvement
which will warm the hearts of the many traditionalists in the library
profession who have disliked catalog cards on which the print is all
one size. This drum is also to be used by another System, and will
become a standard item for purchase.

The card stock for the Univac -produced catalog cards will be
pin-feed, die cut, fan fold, continuous form and will be fed through
the Univac 1004 at the rate of 600 lines a minute. The machine will
read a main entry (key punched) authority card, print a determined
number of author cards, plus one card for each of the tracings.
These, of course, will be printed in sequence and will come from the
machine in collated sets, thereby eliminating the the need to overrun
and stockpile catalog card sets.

At the present time, under a special grant from the state of
New York, a key punch operator has been hired who is hard at work
placing the authority file on tape. It will take about a year for the
task to be completed. In about three months, current cataloging
information will begin to be placed on tape as well. Since close to
125 titles a day must be cataloged, it seems essential that the input
operation be streamlined as much as possible. Cataloging staff can-
not be spared to block out catalog card information for the key punch
operator. Neither can a key punch operator be expected to be expert
in dealing with transfer of information in a suitable manner from
catalog cards to tape. The answer to this apparent dilemma seems
to lie in making the Univac 1004 mainly responsible for the format.

The procedure being carried out by the key punch operator is
as follows:

1. A code number, which happens to be the batch number for
location of catalog cards, is punched into the author section of
the catalog card. This is automatically repeated on every card
in the set. This batch number indicates the sequence in which
the title in question was cataloged. There are incidentally
sixty-seven spaces in each card which may be used for key
punching.

2. Cards for title entry are punched, and the entire title is
punched in. If more than one card is required (this will be the
case for all but short titles) , the key punch operator runs off
the end of the card. Then she starts a second card for title re-
punching the word that is not completed on the preceding card.
The machine (Univac 1004) will automatically read out any
word not completed at the end of a card simply by reading
column seventy-nine and automatically rejecting any word
which spills over into that column.

3. Collation will have a separate card for each section.

4. -8. will be reserved for notes— each beginning with a dif-
ferent number.

9. Tracings will be numbered "nine" with a sub number for
each library to establish location. There will be a separate
card for each tracing.

The Univac 1004 will automatically figure identations, spacing,
etc., because it is all built into the program. Cards will be printed
out in complete sets printed at the rate of 600 lines a minute with no
collation problem.

The big problem has been input. Normally, it will take three
minutes to type a main entry which can be used for filming. Then of
course there is the overtyping of headings on the masters in addition.
It is believed that an operator should be able to punch an authority
card with tracings in five or six minutes, and this should make it
economically feasible.

The board already has been programmed. This operation will
begin in a limited way in May and by year's end if it is successful,
it should be a full blown effort. The authority file should be almost
completely on tape by that time.

One drawback encountered with the System's machine is that it
has only four special characters: the apostrophe, period, comma,
and hyphen. This is necessary because the drum, even though it is a
special drum, is limited to sixty-one characters. This means cer-
tain types of punctuation will be eliminated for the present. The
System chose upper and lower case over complete punctuation
principally because it could not sell the idea of catalog cards in all

upper case. However, the key punch operator will punch into the card
all punctuation, and it will be so recorded on tape. The Univac 1004
will read out (reject) the punctuation it cannot print, but as the equip-
ment becomes larger and more sophisticated, it will be possible to
print cards with all punctuation desired. As far as upper and lower
case is concerned, the key punch operator will be expected to strike
a key as a typist now does to signify capital— otherwise, the print out
will be in lower case. A standard key punch machine will be used for
this purpose.

Other benefits occur when the machine tape drive units are
asked to search the tracings for the subject approach. The equipment
will print out this information listing, for example, all of the items
listed under Marketing- Statistics. Plans are to make this a finding
listing as well, by playing the order cards into the authority file,
searching by batch number, and listing location of library purchasing
the title. Hence a subject listing will be created which gives the
location of a book in a library.

Another use of the authority file on tape will be to feed order
cards, not coordinated, (about 3,000 plus a week are received) into
the file to determine if the title has been previously cataloged. If it
has, there will be a print out giving batch number, if the author and
title correspond. The estimate is that this may reduce the manual
searching by 30 to 40 percent.

Plans are also being made to print a book catalog of fiction
from the authority file. For the present, fiction is not in the Union
Catalog, and this inhibits effective interlibrary loan of fiction ma-
terial. This catalog can be automatically kept up-to-date by feeding
order cards into the storage unit once the books have been processed.

The tapes also can be used for printing out overdue notices for
member libraries by utilizing registration files on storage. This will
be added slowly and has a low priority at the moment.

Tape will be used immediately to store information concerning
the accounts receivable. The financial deck (receivables) will be
filed by a number (described under ordering) . Formerly this was done
without tape by sorting alphabetically on the sorter, and this took
hours every week. As vendor's bills are approved they are added to
the deck; as member libraries are billed subtractions are made
from the deck. All bills in and out are also recorded by customer
on tape. At month's end, a "print out" will provide a statement
indicating "ins" and "outs" with a net figure owed the System.

There are many other possible applications for magnetic tape,
but at the present, attention is being focused on those already men-
tioned. There is a gamble in what is being attempted, but to stand
still in such an operation is not possible.

In reference to the capacity and speed of the equipment, the
Univac 1004 with two magnetic tape drive units will be able to ac-
complish the following:

1. Automatically print out the entire master authority file once
on tape in thirty minutes /20, 000 titles, three across, three lines
per title.

2. Allow for 25,000 titles representing approximately 250,000
punched cards to be contained on one reel of Univac 1004
magnetic tape.

3. Provide print out at the rate of about 600 lines, or 34,000
characters, per minute.

4. The Univac 1004-in Model A has 31 two address program
steps. The System is externally programmed. That is, boards
are wired much as the IBM 403 was wired for the System's
other operations. Memory is 961 characters of core storage
and is addressable, variable word length. Memory cycle line
is 6.5 micro-seconds. Tape, of course, adds to the memory
just outlined.

5. Produce rewind speed at less than three minutes per 2,400
foot reel.

6. Permit tape to be read at 42.7 inches per second.

Miscellaneous

At the present time, book catalogs are being prepared and used
for specialized services. The bookmobile, for example, has a book
catalog as does a System sponsored reading center, and shortly many
of the reference tools in the two Central Libraries will also be in-
cluded in a special book catalog. At the moment, the catalog is
produced from order cards for books that have been processed for
the respective departments. It is an unsophisticated listing giving
the last name of author, title, publisher, date published, and Dewey
number. It is, in short, a quick finding list.

With the authority file on tape, it should be possible to have a
book catalog which will provide full main entry catalog card listings.
This will mean better book catalogs produced inexpensively, and
they should be able to be updated with a minimum of expense.

The card files now maintained for the book catalog not only
provide a bare minimum of information but take considerable effort
to keep current. Strangely enough, it is the two least sophisticated
pieces of equipment— the sorter and the keypunch— that are the
weakest links in the production chain. This is true for this as well
as most of the other programs of the System.

Circulation control is now being offered to four member
libraries and the bookmobile program. The four member libraries
and the bookmobile circulate 750,000 volumes a year. Adding the
fifth library will boost circulation to approximately 900,000 volumes.

The other member libraries circulate about three million additional
volumes and will be added a few a year as they indicate interest.
There are basically two plans with some variations now in use:

1. A filming charge-out procedure utilizes a pre-numbered
punched stub transaction card. When a book is circulated the
stub is taken from a consecutively numbered deck and filmed
with the borrower's card and the book card. The stub (which
is end printed) is placed in the pocket, and when the book is
returned the stub is pulled. Each day all stubs returned are
sent to the System office. The stubs are reproduced into con-
ventional sized data processing cards (on the reproducer) and
filed numerically after having been sorted. One day each week,
cards for books charged out six weeks earlier (four weeks
loan and two weeks for a margin) are fed into the Univac 1004.
The machine reads the cards printing out only missing numbers,
and these are sent to the library to be checked against the films
for overdues.

2. Another procedure is a borrower participation plan calling
for patrons to fill in author, title, and borrower's number on
unnumbered data processing cards. These are brought to the
charging desk by the patron, and a pre-numbered stub (end
printed and punched) is placed in the book pocket. A sequential
numbering stamp keyed in to the number on the stub is used to
stamp the transaction number on the patron's slip. This slip

is placed face down as in many transaction systems, and the
day's circulation is sent to the System the following day. There
the cards are end punched and printed using the reproducer
and matching the cards against a pre-numbered deck. These
cards are checked to see that the stamp and the end printing
are similar and then sorted and filed in the books outstanding
deck for that library. When the book is returned by the patron,
the stub is removed from the book and sent along with other
stubs to the System office. The stubs are reproduced into
regular sized data processing cards by the reproducer. They
are then sorted and routed through the collator to purge their
numbered counterpart. At the end of a specified period of
time, which varies by library, the remaining cards are overdue
and are sent to the library involved for overdue notices.

This is a very simple operation. There are other ways to
operate these procedures. One entails transmitting information over
a telephone to a slave keypunch or to a computer. The Suffolk Co-
operative Library System has chosen the simple and the inexpensive.
Next the tape drive units will eliminate the need for sorting, collating,
etc. in circulation procedures. It is expected that tape may be utilized
to streamline this operation and enable the System to enlist more
libraries into the program.

A Union List of Serials has just been produced for the holdings
of member libraries. The program will now call for updating the
publication annually, but the bulk of the work is now over.

Labels for mailing and a myriad of special purposes are, of
course, easily produced by the Univac 1004. This is a fringe benefit
and probably the most easily established program.

Soon the System expects to add an additional core unit and a
translator to the Univac 1004. There are many reasons for this, but
one which should pay off is that the eight channel tape produced by a
teletype machine may be used to mechanize the Interlibrary Loan
procedure. Daily the System now copes with about 100 requests for
books by member libraries. It is expected that this will double with-
in eighteen months. Utilizing the teletype enables materials not in
the County to be located. At present, of course, records are main-
tained manually and record statistics the same way. Plans for 1965
call for mechanizing this program by typing all requests on the tele-
type machine feeding the eight channel tape through the Univac 1004
into the magnetic tape. This way overdues, statistics, and other
record keeping can be a responsibility of Univac.

Summing up, the System's data processing operation costs
personnel-wise $17,500 annually for a tab machine operator, two
keypunch operators, and one part-time machine operator. The equip-
ment costs about $18,000 a year not including the outright purchase
of the two tape drive units for $32,000. The program is growing,
soon better than 200,000 volumes for three Systems and many school
libraries will be processed. Accounts now exceed 100. Taking into
account the control annually of the flow of about $1,500,000 and the
provision of services previously described, it can be said with cer-
tainty that without data processing equipment this System would be
out of business. It is also equally certain that without an entire
System concept, the equipment would not have produced adequately.

Many of the procedures described in this paper are about to be
phased out with the coming of more sophisticated equipment. Changes
will be taking place for the foreseeable future. Sample forms and
printouts presently in use appear in Appendix n.

Man is limited by his intelligence; thereby he limits the
equipment. The System attempts never to box itself in and hence
has retained IBM peripheral equipment with the operation of the
Univac 1004. Eight columns allows the necessary flexibility. There
are many other instances where flexibility now, allowing for de-
cisions to be made later, based on knowledge learned and earned
has paid off. It is hoped that the potential of the equipment can con-
tinue to be challenged, because by so doing, new service possibilities
for the Suffolk Cooperative Library System are opened.

Appendix I

ORDERING PROCEDURE

ORDER DEFT.

REPRO- \ PUNCH INFORMATION INTO
SOURCE DOCUMENT

DATA PROCESSING DEPT.

• J PRE 'CATALOG ING

B A PRE "CATALOG ING

SORTER \ BY AUTHOR BY TITLE

DATA PROCESSING DCPT.

CATALOGING DEFT.

DATA PROCESSING DEPT.

ORDER C£PT.

C 1 INVOICE BALANCING AND INVEHTQg

ORDER C£PT.

VENDOR'S
INVOICE

PULL RELATED ORDER

NARK PRICE AND DISC
ON THIS CARD

OUNT /"^

f ORDER CARDS

DATA PROCESSING DEFT.

ORDER DEFT.

DATA PROCESSING DEFT.

BOOK OCCK-IN

BOOK CHECK- IN

SORT \ BY JOBBER BY TITLE

INVOICING. ACCOUNTS RECEIVABLE.
INVENTORY UPDATING. UNION CATALOG.

INVOICING. ACCOUMTS RECEIVABLE. INVENTORY UPDATING. UNION CATALOG

f r J BOOK LABELS

PUNCHES DEVIEY DECIMAL NO.
FROM MARK SENSED CARD

PROCESSING DEFT.

LABELS AND JACKETS

Af F IXED
CATALOG CARDS

INSERTED

Appendix II

SUFFOLK COOPERATIVE LIBRARY SYSTEM
PATCHOGUE, NEW YORK

SAMPLE FORMS AND PRINTOUTS

BOOK ORDER FORMS

CARBON INTERLEAVED BOOK ORDER CARD SETS

1OOK TITLE

if 11(1 m mi 111

1 1 1 1 1 1 1 1 1 1 1

tiiiiiiiiiiiiiiiiiitiiiiiiiiiiimiiii]
mmmimmmmimiimmmti

41 4444444444444

iimimm

I 1 I 1 1 1 1 I I 1 I 1

4H444444444444444444I44

III III III III I Mil III

iimmimm

1 1 1 1 1 ii in M

mmmmmtimmmmmimm
1 41 i|i 1 1 1 m in in in i in in m m m in

if J 1 ! 7 1 1 1 1 j|l 1 1 7 1 1 1 1 1 M T 1 1 1 T 1 1 ? 1

I I II I I I I ! I II I II I I I

mmmmmmmm

C1=>CI=JC13

C 1 =C 1 =>C I =lc 1 3d 3C 1 3c 1 =

MARK SENSE ORDER CARD

VENDOR PURCHASE ORDER

* THE Sumu

COOPERATIVE LIUARY SYSTEM

VDOX5R-S COPT ', •

•

CAMPBELL-HALL INC 21*1

'Is^iTl*

PUICHASC ORDEII

191 CLARENDON STREET

BOSTON IT. MASSACHUSETTS

MO.

loo. mu

~~m

OMAKS

MILNE

BALANCE OF NATURE

KNP

•

LEAF

BEIN* AN AMERICAN CAN BE FUN

LIP

BRONO.SKI

BI06RAPHY OF AN ATOM

HPR

**! •

JONES

BIRTH OF LIBERTT

75|

d»i

BOB TRA BOE

CALIFORNIA AND THE KCST

"i*

DUNCAN

CAMERONS AT THE CASTLE

STM

•5

* THE Sumu COOKMTIVI LIMARV SVSTIM

VDfDOR'S COPY

KI •

'«•""•"•—- ., .SET" """""•"""

KI -

•••• pg-1

• i

• o.d..«j Fro* CAMPBELL-HALL INC 2

161

1 «; Taj t j [j

" J

PUICHASE C«De»

-• •

•

« !

«,.«»

.00. T.UI

U«H

•MAM

-. •

TRESSELT

HIDE AND SEEK F0«

LLS

•

"*! -

FLETCHER

HIST OF ARCHITECT ONCOMI

•AR METHOD 17EO

SCR

•2

•

•OOLLEY

1 LIKE TRAINS

HPR

•

75 |

KOFFLER

I LL SHO. YOU CATS

HPR

•

»B

FARQ4JHAR

INDIAN CHILDREN OF AME

<ICA

HRK

•»i

.IBBERLEY

ISLAND OF THE ANBELS

•OR

SECHRIST

IT S TIME FOR STORY HOU

MAR

•

SECHRIST

IT S TINE FOR STORT HOUR

MAR

»5

•

SECHRIST

IT S TIME FOR STORT HOUI

MAC

•

7i •

SMITH

ITALY

UNP

••

.ISE

JANE AODAMS OF HULL-HOUSE BIOBRA

HBK

•

bO.ERS
COATSKOHTx

JEFFERSON t HAMILTON
JON THE UNLUCKY

MR.

JOSLIN

LA PETITE FAMILLE

-B.

•I

•

DAY

LANDSLIDE

MCN

CROSS

LIFE IN LINCOLN I AMD

IICA

RAN

SOBOL

LOCK. STXK AND BAR

•L

1ST

•

JONES

LORD S PRATER

•

ARISTOPHAhC

LYSISTRATA

UNC

•

CHASTAIN

"•SIC ISLAND

•

TRAVCKS

MARY POPPJNS

ADAMS

MR. PICKLEPAI S POPCORN

LLS

"••

MOBSON

MYSTERT HOUSE

LIP

PAN AMER

NE> HORIZONS .ORLO BUlOE LATEST EO

SIS

CONTINUOUS FORM BOOK LABELS

•

WODEHOUSE

•

BRINKMANSHIP OF

• wo

GALAHAD THREEPWOOD

GALAHAD THREEPWOoO

WO •

• 942.

HIBBERT

942.

HIBBERT

942 • •

COURT AT WINDSOR

• HI

HI •

• X

FITZGERALD

X •

ECHO ANSWERS

• FI

MURDER

FI •

• 641.

FARM

641.

FARM

641. •

631

JOURNALS COMPLETE

631

JOURNALS COMPLETE

631

• FA

PIE COOKBOOK

FA •

•

CHASE

•

JOURNEY TO BOSTON

• CH

CH •

• 613.

MYDEN

613.

MYDEN

613. •

OVERWEIGHT SOCIETY

• WY

WY •

• 973.

MORI SON

973.

MORI SON

973. •

OXFORD HIST OF THE

• MO

AMERICAN PEOPLE

MO •

• X

PAUL

X •

PULLED DOWN

• PA

PA •

MEMBER LIBRARY INVOICE

' THE SUFFOLK COOPERATIVE LIBRARY SYSTEM

• sold ,0 NORTHPORT PUBLIC LIB [74] |o»| lt| 49j |1|

119 MAIN STREET

NORTHPORT NY 73

"sssr

M?£ci

fITfNyOM

SINCLAIR AOVAILABLE MAN MAC

3 It 49

6.99

• .17

SINCLAIR AOVAILABLE MAN NAC

3 It 49

6.93

3.79

• .17

_ MEMBER UBRARY COPY l| m

THE SUFFOLK COOPERATIVE LIBRARY SYSTEM

1 It 44

9.99

•0

3.37

-» ""* At,.,i£"" ' •_

1 It 44

• .99

3.01

•

• sold 10 NORTHPORT PUBLIC LIB ITS] iMIttlMI [jj) •

S It 44

• •99

3.01

INVOICE

3 It 44

*.94

3.01

* *

3 It 49

• .43

3.01

• .u,-o.

.00. mu

«•««•

"SSK-

,,i!"Xc,

l«l«0.

•

4ENDICK

HORSE NAMED SUMMER

RANMT1

01 11 63

3.90

• 0

1.10

MARKER

I COME FROM THE STONE A4E

OUTTO

01 It 69

6.95

• .t*

01 It 49

9.00

3.11

•OOLLEY

I LIKE TRAINS

HPR

3 It 69

1.97

3 14 49

.3,90

l.tl

• j

SCCHRIST

IT S TIME FOR STORY HOUR

MAR

3 14 49

• .49

3.11

) 01 It 43

*.90

1.79

•

• CROSS

LIFE IN LINCOLN S AMERICA

RAN

3 t4 44

3.49

l.»4

3 14 49

1.49

1.44

SOUOL

LOCK, STOCK AND BARREL

•ST

3 16 69

3.43

t.»4

3 It 69

3.95

1.37

JONES

LORD S PRATER

R»C

3 16 69

1.00

1.26

3 It 49

3.49

•0

1.37

•

• OROTHERTON

ME N STEVE

M«C

3 14 49

1.43

1.77

•

« 01 It 69

3.43

t.*l

AOAMS

MR, PICKLEPAIT S POPCORN

LLS

3 16 49

1.49

1.44

v 01 It 44

3.49

l.»l

PAN AMER NE> HORIZONS WORLD 4UIOE LATEST ED

S4S

01 It 49

1.49

1.44

• j

9 It 44

4.49

•0

• .17

• .

• HUTCHINSON NED PENNSYLVANIA DUTCH COOKERY

» 01 49

3.49

l.«t

•

3 14 44

3.30

t.ll

CASTE! NISHTMARE RALLY

A1LEA

01 14 69

3.90

1.1*

3 16 64

1.44

2.46

SUB TOTAL _^>| 66.91

]•

.JUVENAL SATIRES

INDIA

Ot It 69

3.79

l.tl

• 4AUDOUY SECRET OF THE HIDDEN PAINTINt

H»«

3 14 44

2.95

1.44

•

KRAUS

SPRIN4FELLOH S PARADE

HPR

3 14 44

1.99

•

• 4LICK

STORY OF OUR FLA4

PUT

3 14 65

3.90

till

SNYDER VELVET ROOM

ATHEN

Ot 16 69

3.tB

1.37

•

• UBELL

•ORLO OF THE LIVING

ATHEN

01 16 69

3.14

•0

1.14

PAT TMI» »»OUNT^>) 119.11

•

THE SUFFOLK COQFEMTI« 'tajjtrji

-. ^ -W~— -: •-' • ' : -''-^ ~ •'

s •

•

15 tfEST AVENUE. PATCHOSUE. N.

" •

•

NANUET PUBLIC LIBRARY PA8E IT i ~

''' =

_j>

•

175 SO MIOOLETOWN RD

•

NANUET MY IIT* ift^g"T7

DATE C3/31/65

•

"- ' :--.:

1 •

DATE

DESCRIPTION . CHAMK

C_REBITS BALANCE

•

0 02/28/65

PREVIOUS BALANCE 795.W

79S.29

03/01/65

CREDIT MEMO

16.35 618. 94

0 03/04/65

BOOKS 92.11

7(1.05

03/05/65

BOOKS :.'.:-

866.39

0 03/09/65

BOOKS 8S. 62

975.01

03/10/65

BOOKS M'Tt .

1040.73

0* 03/12/65

BOOKS 116.20

1178.93

03/16/65

BOOKS 69.03

1247.96

0 03/16/65

PAYMENTS

705.11 542.85

03/16/65

CHARSE BACK ON LIBRA I*fi7_

544.42

0 03/19/65

BOOKS 63.04

627.46

03/23/65

BOOKS 79.6.9.

707.35

0 03/26/65

BOOKS 37.64

744.99

03/29/65

SUPPLIES fi»14 .

750.15

0 03/29/65

DEBIT MEMO 14.60

764.75

03/31/65

•

BOOKS ?4.4U

799.19

•

BALANCE DUE

799.19

ANALYSIS OF BOOK PROCESSING Tl

•

TOTAL

21 DAYS

3l DAYS OVER 31 *

AMITYVILLE FREE LIBRARY

60S

2 to*

• AMAGANSETT FREE LIBRARY

92*

18* •

BAYPORT PUBLIC LIBRARY

100*

« ^

BAY SHORE PUBLIC LIB

•0*

1 20*

• BRENTWOOD PUBLIC LIBRARY

88*

2 12* •

BRID6EHAMPTON PUB LIB

88*

1 13*

DEER PARK PUBLIC LIBRARY

9S»

1 2*

• FLOYD MEMORIAL LIBRARY

79*

3 21* •

HALF HOLLOW HILLS LIB

58*

11 t2X

HAMPTON BAYS PUBLIC LIB

75*

1 2SX

• NORTHPORT PUBLIC LIB

71X

It 29X •

THE PATCHOGUE LIBRARY

t 29*

86X

2 ItK

RIVERHEAD FREE LIBRARY

71X

5 29X

• SACHEM PUBLIC LIBRARY

80*

3 20* •

SOUTHOLD FREE LIBRARY

100X

NEW CITY FREE LIBRARY

100*

• NEWBURGH FREE LIBRARY

75*

3 2SX •

PEARL RIVER FREE LIB

100*

FINKELSTEIN MEMORIAL LIB

7 19X

72K

10 28X

•

351

11 3X

291

83*

60 17X •

•

870 DOCUMENT WRITING SYSTEM

OF INTERNATIONAL BUSINESS MACHINES CORPORATION
IN THE LIBRARY SECTION, LOS ANGELES CITY SCHOOLS

Mary Seely Dodendorf

The Los Angeles Unified School and Junior College Districts
are constantly growing. The present area of the city of Los Angeles
encompasses 458 square miles; the Los Angeles Unified School Dis-
trict, 710 square miles; and the Junior College District, 840 square
miles. Although there is little change in area from year to year,
there is considerable increase in the number of schools needed to
keep pace with the rapid population growth within the area. For the
purposes of this paper, discussion will be limited to services to the
District high schools during two recent periods: the fiscal period at
the beginning of use of the IBM 870 Document Writing System and the
most recent year for which statistics are available.

Increase in past
1953-54 1960-61 1963-64 ten years

(1953-54 to 1963-64)

Schools
Number

*K-14 484 578 589 105

7-12 83 114 122 39

Active enrollment March 31, 1960 March 31, 1963

*K-14 439,821 639,281 714,368

7-12 129,576 198,568 230,333

"Includes Adult Education (High School) in K-14 only.

The Catalog Department of the Los Angeles City Schools has the
responsibility of providing cataloging services for the school libraries
so that the librarians may have more time for their major responsi-
bility which is working with students and teachers. Since needs of
catalog users vary with grade levels, there are corresponding vari-
ations in the methods and results within the department. When the

Mary Seely Dodendorf is Librarian-in-charge of the Catalog Depart-
ment, Library Section, Los Angeles City Schools.

administrative decisions have been made as to services to be ren-
dered, the success with which they are carried out is dependent upon
the provision of adequate qualified personnel, quarters, equipment,
and supplies.

The aim of the Department is to provide useful, appropriate,
simple catalogs with pertinent cross references, and to provide them
quickly. The catalog format must be that best suited to the students.
Its arrangement should enable students to develop a skill in using the
catalog which will give them confidence in using various types of
public, college, and university libraries everywhere. The conventional
card catalog is still currently preferable to other forms for District
use. The collections are neither uniform nor static, and they are not
subject to interlibrary loans.

The school librarians are free to order any titles which are
needed in their individual schools and equally free to withdraw a title
which is no longer useful in the collection. They are helped in their
selection of books by lists of current titles which have been reviewed
and evaluated by committees of librarians and teachers within the
school system. These lists have been annotated, with maturity and
interest levels suggested. The titles are recommended as ones which
will be generally useful in the schools. The school needs vary as
population varies in a system which is geographically as large as
that of Los Angeles. This desirable freedom to select titles when
and as they are needed does present some complications for the
central office processes. Titles which appear on the recommended
lists are generally ordered in quantity, and such orders are placed
by machine data processing methods by the Evaluation and Order
Department. The cataloging of these titles is done from review
copies which have been used in the evaluation process and which
occasionally vary in edition from the books actually received. The
same title that appeared on the list may, and frequently does, appear
as a single copy ordered for the first time for an individual school
before that title appears on the list, as well as on orders placed
months, or even years, later.

Books ordered by the schools fall into three categories for
cataloging purposes. Titles which are new to a school require sets
of catalog cards. Titles which a school orders as duplicates, of
course, need no catalog cards and are not routed to the Catalog De-
partment. The third category is the title which is not new to the
school but for which cards have been requested to replace withdrawn
sets. In addition to the books which are received in the Order De-
partment and routed via the Catalog Department, there are books
presented as gifts to the school libraries; and for these the schools
may, and do, request sets of catalog cards. Two extensive "replace-
ment" lists annually result in the placement of orders for titles,

some of which may be new to several schools. All of this results
in considerable overlapping of titles handled at various times for
cataloging purposes.

When the Catalog Department was established in 1927 there
were fifty-five junior and senior high schools. The policy regarding
acceptance of central cataloging and classification was permissive
and although accepted quickly by most schools, had not been accepted
in all until 1950. By this time there were seventy-four secondary
schools. (Elementary cataloging did not become a part of the Catalog
Department until 1947.)

( High school only) 1960-61 1963-64

Central Catalog Department staff

Librarians 2 *3 + 1 = 4

Library clerks 1 2 + 2 = 4

Clerk typists 5 4

Duplicating machine operator 1/2 time 1/2 time

Duplicating equipment A. B. Dick 350 Offset Duplicator

A.B. Dick 350 Offset Duplicator IBM 870 Document Writing System

Standard Register Burster

New titles and editions cataloged 2,150 5,451

Sets of cards sent to schools 45,553 98,157

Shelf cards only

(Arrearages of sets) 10,152

(For "special" schools) 458 1,496

Library staff in schools— one in each high school with clerical help
ranging from 2 to 8 hours per day.

* Although there were again only two librarians cataloging for the
high schools in 1964-65, the overall increase continued because the
one librarian and two clerks assigned to the IBM 870 System were
working with only high schools records.

The Catalog Department has provided sets of catalog and shelf list
cards for secondary schools but no preparation of books for shelving
except for ten "special" high schools.

The initial reason for working with the 870 Document Writing
System of IBM was to clear the arrearages which were continuing to
build up in the Catalog Department. There were several related
causes for these arrearages. An increased number of books resulted
from increased book budgets. Basic collections of books for the
library had to be cataloged for each new school established by the
District. Each of these collections received approximately two

thousand titles during each of its first two years, and frequently four
new schools were opened during a year. It is easier to purchase
"service" than to add and reclassify staff. The staff was not sufficient
in quantity and job classification to handle the amount of work by the
methods then in use. Turnover of the already small staff as civil
service personnel moved into advanced positions outside the Depart-
ment put extra burdens on both professional and clerical persons.
The budget framework provided a handicap, for there is little "season-
able" employment. "Relief clerks helped, but inexperience and lack
of space handicapped them.

By the end of 1960-61, approximately 6,600 volumes ordered
during the past two years remained for clerical handling. An addi-
tional 2,450 volumes awaiting the first sorting included an undeter-
mined number of new titles to be cataloged. Unfilled requests
received from the schools for sets of catalog cards amounted to 7, 000.
The total number of sets of cards to be produced was probably in the
neighborhood of 10,000 for "single" orders. During 1960-61, 45,553
volumes had been sent with complete sets of cards to schools. An
additional 10,152 volumes had been sent with only shelf list cards,
and the corresponding 10,152 sets of catalog cards to be sent in-
creased arrearages.

While all these arrearages were accumulating, students were
reading in a District publication that one of the most important aids
in the use of the library is the card catalog as an index to the book
collection. "You will feel independent and at home in the library when
you know how to use this aid. Through the catalog you gain informa-
tion about the book collection." But the cards for books which the
student knew were in the library were not in the catalog. This was
not a good situation, for the student was likely to lose confidence in
a library tool and perhaps in the library itself.

Arrearages are not new. As they began to mount for the sec-
ondary schools, the "multiple" copies, which resulted from combining
the orders for several schools, received first priority because of the
physical bulk for which there was no storage space. It was possible,
generally, to prepare cards for these titles in advance of the receipt
of the books so that the complete sets of cards could be sent with the
books. "Singles" (titles ordered by schools individually) took second
place and remained on the shelves until the "multiples" were cleared.
In order to clear the "singles" and because of the manual typing in-
volved in preparation of cards, these books were sent to the schools
with only shelf list cards marked to show that catalog cards would be
sent later. The shelf list cards provided sufficient information so
that books could be processed for shelving and student use. This
device, of course, merely transferred the arrearages in the Catalog
Department from books to cards. The duplicating machine operator
was producing neither the quantity nor the quality of cards needed,
and this served to increase arrearages.

Previously the catalog cards of the H. W. Wilson Company had
been used in preference to typing our own cards; the annotations
were particularly appreciated. The cards were used without over-
printed classifications and headings because, although subjects might
lend themselves to change, the problems of classification changes
would be impractical to adjust in schools where the librarian had no
professional help. With regret this service had to be stopped when
the packaging became unwieldy and cumbersome for our procedures
as the number of collections grew.

Arrearages had been attacked at another point for several
years. In order to reduce the number of analytic cards prepared,
inquiries were sent to the schools in a form which the librarians
could check and return to the Catalog Department and which could be
returned to the school with the analytic cards. These forms could
then be used as a tracing record for the cards in the specific catalog.
Whether or not it saved work for the school librarian, it did not pre-
vent arrearages.

A partial moratorium on sending reference cards to the schools
was another attempt to find sufficient time to reduce arrearages.
Subject references were discontinued except in the case of new sub-
jects. The searching and recording involved in knowing which school
needed references was too time consuming. (Remember that collec-
tions in the schools are not uniform.) References were continued for
personal names. It was hoped that the school needing the reference
already had it or would order a title needing it when the moratorium
was lifted. "Requests" for catalog cards for books received at the
schools from sources other than the Library Section were discouraged
and were given bottom priority. Schools were encouraged to classify
these books in accord with books already in the school library. They
were directed to prepare their own temporary catalog records for
such titles.

Preparation of catalogs and shelf lists for basic collections for
new schools is very time consuming and delays regular, normal, day
to day work. Commercial photographic reproduction of the catalog
was not found successful a few years ago and required considerable
preparation time. It was a forward step, but the best forward step
for new schools was taken when the Department turned to a local
commercial service for cataloging and processing for new elementary
and secondary schools. Of course, the central Catalog Department
does the original cataloging. It provides unit card copy so that the
completed cards and books are compatible with those supplied from
the central department. Because consultation is easy with this local
company, they can be depended upon to recognize edition changes
which require further attention. The IBM 870 installation saves con-
siderable preparation time by producing the catalog copy in a form
that requires little revision time and one that the company can use as
its original document.

Prior to the installation of the IBM 870 System, there were a
number of production problems. The production of single sets of
cards by manual typing, even on an electric typewriter, was slow
and subject to error so that it was necessary to proofread each card.
The same conditions were true for manually typed offset masters.
The Department was, and still is, using an A.B. Dick 350 Offset
Duplicator to reproduce cards which are needed in quantities of six
or more. This break-off figure was selected because it is a practical
point in relation to book orders rather than for an economic advan-
tage. The offset masters are used only once because of lack of
storage space. There is no place to stock extra cards, no basis of
judging what titles would be needed in the future, and no assurance
that the same edition would be needed again. When another school
ordered the same title at a later time, it was necessary to retype
the offset masters or the cards and to revise the finished product.
All of this required considerable revision as well as typing time.

The IBM 870 Document Writing System was originally developed
for the addition of individual headings for form letters and used tape
and cards to activate two typewriters.

The Department is not attempting anything dramatic— no new
or different kind of task— but a quicker, easier, and more economical
way of doing repetitive tasks. Relatively speaking, the IBM 870 is a
very simple, unsophisticated piece of machinery— a couple of type-
writers and a key punch— an electro -magnetic device with very little
logical ability. Setting up the program required a great deal of at-
tention to detail, and the more detail that was necessary the more
difficult the setting up became. Although when it was finished the
result was an unsophisticated machine, it has proved to be a valuable
one.

There was disappointment that carbon ribbon attachments have
not been possible for the typewriters in the installation, but the an-
ticipated breakage of such ribbons with consequent machine slow-
down as well as the conversion cost stopped us at that time. Because
carbon ribbons would not be used, it was necessary to experiment
with fabric offset ribbons which would be compatible with the offset
duplicating equipment and its supplies as well as suitable for direct
typing on catalog cards to be handled by students. The representatives
of the A.B. Dick Company, whose duplicating equipment the District
uses, were not only very cooperative during the experimental period
but have continued to insure fresh ribbons at all times. One's eyes
cannot tell when the safe period is past and the carbon necessary for
duplicating purposes is no longer being deposited on the paper
master. Experimentation has shown that during the first eleven hours
of typing time of a fresh ribbon (A.B. Dick Offset Nylon Ribbon,
Part No. 4-2510, direct image, IBM, 9/16n) carbon is deposited on
the master in sufficient quantity for good reproduction on the

duplicating machine. Consequently master stock is used for eleven
hours and then the ribbons are read onto card stock.

Problems concerning paper stock arose, and decisions had to
be made prior to the IBM 870 installation. The tufted edge which is
common on continuous stock was not wanted. After searching, a die
cut stock of Standard Register which leaves a smooth edge at the top
and bottom of the card and is broken only by three notches was found.

Another card stock problem which may be unique is the change
in color tone when there is a change in vendor. Natural manila color
is used for all of the cards with a color stripe. A blue stripe on the
Master Catalog Card and a green one on the School Catalog Record
Card are used. Since each title has both of these master cards stand-
ing together in the file, it was disconcerting when one vendor not only
produced a turquoise (almost green) line in place of the blue but
"corrected" the order with another turquoise. Perhaps there was an
unrecognized color blindness involved.

Filing took on a new look as the colored edges appeared in the
file. The School Catalog Record Card, which has a green line, neces-
sarily uses a shortened author-title combination because it is limited
to the length of a single line. This necessitated changing the arrange-
ment of cards in this file. The master file is now arranged by sur-
name and first initial. Emphasis has been added to the reference
cards for names by the use of a distinctive colored stock (red) in
preference to the use of a name reference for each title of an author
with a variant form of name. Added entry authorities are entered on
cards with a brown stripe, and temporary records are on solid yellow
stock.

The first year with the IBM Document Writing System was ex-
citing and at times frustrating. In October 1961, the librarian who
was to supervise the operation was sent to IBM school for an inten-
sive week's orientation course in IBM methods. From then until
August 1962, working closely with IBM programmers, she prepared
samples and diagrams for programming purposes. By February
1962, when the control unit and three typewriters were delivered, it
was thought that the programming was well along. The permanent
typewriters were not received until May, and by that time the librarian
had made adjustments to several new programs as previous ones
were scrapped. Major programming changes continued until August
1962 when production finally began.

It has been said that this installation is unique— essentially one
of a kind. The equipment consists of one 836 control unit and three
866 typewriters. The sum of these is one IBM 870 Document Writing
System. (No, this is not the New Mathematics!) Only cards are used-
no tape— to control the typewriters which have elite ( 12 to the inch)
type so that full advantage can be taken of the usable space on the con-
ventional (7-l/2cm x!2-l/2cm) catalog card. One typewriter is used
exclusively for shelf lists for which a special printed form is used.

No detail that was desired was impossible, or even very dif-
ficult; but the accumulation of details that varied from previous 870
installations for form letters made many problems— some of them
not yet satisfactorily solved. The solution of problems has been a
trial and error procedure. When luck was on our side, two problems
might be solved at a time. When it was not, one or the other problem
had to be resolved. It was the application that was unusual, not the
trial and error procedure. The first hurdle was one of semantics as
IBM systems engineers learned library terminology and needs, and
librarians learned to express those needs in layman's and machine
language. Changes in IBM personnel assigned to the project during
this period intensified this problem. Since school libraries are
learning centers preparing young people to use other libraries, a
responsibility was felt to maintain conventional catalog forms while
readying ourselves to change those forms in school libraries as
forms change in university and public libraries. The conventional
card catalog is still the preferable form for District use.

There was a need to produce these conventional catalog cards
by typing on either card stock or offset masters for later reproduc-
tion on card stock. There were a number of variables. There were
to be no fixed fields as in conventional key punching. Line spacing
was not constant. Vertical spacing had to be figured for each title
to assure proper placement of the typing on the continuous stock.
Since our right margin is not a set field but is irregular, an auto-
matic line finder and carriage return was not practical. Special
codes were developed to direct the typewriters by means of the
punched cards.

An insufficient number of impulses available in the control
unit has been a stone wall which has not yet been fully scaled. Im-
pulses are necessary to direct the typewriters to move in unison or
individually as well as to activate the keys for individual lower case
letters and numbers. Controls to activate impulses were needed for
turning on each typewriter or any combination of them and for start-
ing the carriage at the correct indention with either upper or lower
case. These controls were coded into set fields (first two columns
of the card) , but intermediary controls must also be coded to provide
for carriage return, tabulation, a single upper case character, or to
end a "mode." The need for both upper and lower case decreased
the number of impulses available for directing typewriter keys.

Cards have been divided into "modes" for punching purposes.
These "modes" (see Fig. 2) are used for materials typed by all or
by only one or two typewriters. A "mode" might consist of the
author, title through imprint and, generally, collation; or it might
consist of tracings or other information.

The high school catalogers prepare the original documents.
These master catalog records (see Figs. 1,4) are on IBM cards
which are lined to indicate the area of a catalog card that the

r«i»wa»"i

HI 1 i M&2I£E_^IA!^£_£A5£

817 Downs, Robert Bingham, ed.
D Bear went over the mountain, tall tales

_j of American animals . Macmillan, 196^.

;

358p. illus.

„

3 Humor I Title

Figure 1
Master Catalog Card as Prepared by Cataloger

ij>«t»i»i»it»iiHB»'mni--ra»-an'»iiii)i:i!)4 »!»» «tsi>x- »»»••••*

R&Ji

7i » n »pwn •

T~I (11 MASTER CMAL06 CARD

8lT Downs, Robert Bingham, ed.
D Bear went over the mountain, tall tales

Hii

of American animals. Macmillan, 1964.

n/j

•' !/. •

358p. illuB.

«!//

?;/

"/*

i;^

r ;/<•

;

/•'<

1 Animals . Stories S Folklore .U.S.

3 Humor I Title

2_j

I !

i ;

Figure 2

Master Catalog Card Including Vertical Spacing Code
and Punched Set Number

,«.

Figure 3
School Catalog Record Card Including Set Number and Bubbled Holdings

M " III MASTER CATALOG CARD

L-LL

979 A Dawson: California (2 of 2)
Anals: Petroleum. California p. 175-181

ill

jjjj

Spaniards in California p. 75-117

U.S. History. War with Mexico, 18U5-1848 4-

2;,,

P.129-1M

Water supply. California p.182-191*

i .

'• ^

';,

: '••

71 T - \ •* \ 1 MASTER CATALOG CARD

1 979 .k Davson: California (l of 2)
i Analytics
I California. Description and travel 4-
i p. 195-202
« California. Gold discoveries p.lU5-l!>2

Pii

|iij

•

3 i

i
I

1 California. Missions p.51-lo6
i Conmunication. California p. 158-166
Explorers p.25-1*^
• Indians of North America. California
P. 3-22
* Overland Journeys to the Pacific
• p. 118- 128

H / — N

•j/5

h- -5

' i'P

2J/^

i :t;

r 3

2 : ;

/ j /

' 1/.5

.. ..LI

»T

1 979 A Davson, Grace (Strlckler)1'
- D California, the story of our southwest
corner. Macmillan, 1939.
212p. illus., map.

1 California. History
Analytic tracing filed in shelf list

4 / — N

iiii

< :-•••

2 1//

5' /

3|y^.

>!/

(S CITY SCMOOU-Chv

i :, -

. g

*'4

; :4

;

t :

*-> O

i ;

Figure 4

Master Catalog Card with Analytic Tracing for
Shelf List Including Vertical Spacing Codes

offset duplicator will reproduce. This permits the key punch
operator to recognize instantly when codes must be punched into
the cards for typewriter operation. In addition to the original
documents, the catalogers prepare any needed "format" cards
(see Figs. 5,6). These cards are expendable documents used to
display card formats which are not clearly shown on the original
documents; e.g., a shortened form for shelf list, an unusual or dif-
ficult added entry form or, an extension card.

When the Master Catalog Card reaches the Document Writing
System process, it is reviewed, and coded information is recorded

MASTER CATALOG CARD

Harris, Leon A

Fine art of political wit.
196*.

Dutton,

MASTER CATALOG CARD

808.7 Harris, Leon A

H Fine art of political wit, being a live-
ly guide to the artistic invective, elegant
epithet, and polished Impromptus as well as
the gallant and graceful wordly wit of
various British & American politicians ...
Dutton, 1964.
288p . illus .

1 Great Britain. Politics and government
2 Huaor 3 U.S. Politics and government

I Title

CO HOT rrrx at fans i

Figure 5

Master Catalog Card and Format Card
For Shortened Shelf List

921

GASKELL, ELIZABETH CLEGHOKN (SIEVEHSON)

LIFE OF CHARLOTTE BBONfE
Lane, Margaret
Bronte . • .

MASTER CA1AL06 CARD

921 Lane, Margaret

B Bronte story, a reconsideration of Mrs.

Gaskell's Life of Charlotte Bronte.

Duell, 1953-
368p. illus.

1 Bronte, Charlotte 2 Bronte family
3 Gaskell, Elizabeth Cleghorn (Stevenson)
Life of Charlotte Bronte I Title

oo nor rrff Krotto Di/rrro LINE -

Figure 6

Master Catalog Card and Format Card
for Heading in Unusual Arrangement

in the boxes printed in the space remaining on the master card (see
Figs. 2,4) . This information indicates the number of lines necessary
after the last typing on one card to the starting position on the next
card. This information is then translated into codes so that direc-
tives will be punched resulting in maximum efficiency of operation
in turning on typewriters singly or in tandem. Operators who do the
coding must understand tracings in order to recognize the number
and kinds of catalog cards required for each title because the aligning
and sequencing of the punched cards is determined from these codes.
The coding is revised before punching begins.

TPLrSJJwr"»«CRIC«<"«NIH«.!s.

IBM 870 DOCUMENT WRITING SYSTEM CAHD

Figure 7
Punched Set of 870 Document Writing System Cards

Shelf list

817 Downs, Robert Bingham, ed.
D Bear vent over the mountain, tall tale*
of American animals. Macmillan, 196U.
illus.

Title

Bear went over the mountain
617 Downs, Robert Bingham, ed.
D Bear went over the mountain, tall tales
of American animals • Macmillan,

Viftrw HI TUB.

Subjects

HUMOR

817 Downs, Robert Bingham, ed.
D Bear went over the mountain, tall tales
of American animals. Macmillan,
358t>. illus.

FOLKLORE. U.S.
817 Downs, Robert Bingham, ed.

Bear went over the niountain, tall tales
of American animals* Macmillan, 196^.

ANIMALS. STORIES

817 Downs, Robert Bingham, ed.
D Bear went over the mountain, tall tales
of American animals. Macmillan, 1964.

Author

617 Downs, Robert Bingham, ed.

D Bear went over the mountain, tall tales

of American animals. Macmillan,
358p. illus.

1 Animals. Stories 2 Folklore. U.S.
5 Humor I Title

Figure 8
Set of Catalog Cards Read by 870 Document Writing System

A "set" number is assigned in consecutive order to each Master
Catalog Card. This number is punched into the Master Catalog Card
(see Fig. 2), the School Catalog Record Card (see Fig. 3) which was
forwarded with the Master Catalog Card, and each of the punched
cards (see Fig. 7) . (The alphabetically filed master records serve
as an index to the numerically filed punched cards.) The punched
cards include also a sequential numbering for cards making up each
set (see Fig. 7) . Each set is easily distinguished within the file
because all cards have upper left corner cuts except the first one
which has a right corner cut.

Two permanently wired boards are used, one for punching and
one for reading. The reading operation, of course, fascinates visitors
as the punched cards are fed into the Document Writing System, and
the coded information is interpreted to produce the typed cards.
During the reading process, an operator reads from the Master
Catalog Card as the copy is being typed, (it has been found practical
to read on Thursdays.) When the day's reading has been completed
and the ink has dried, the stock is fed into the burster, a Standard
Register Selec-tronic Tab Card Burster. After the stock has been
burst (from each typewriter separately), the cards are assembled
into sets for each title and forwarded with the Master Catalog Card
and the School Catalog Record Card. A library clerk compares the
catalog cards and the book, notes on the School Catalog Record Card
that cards have been sent to the school, and then the book is shipped
to the school. The School Catalog Record Card is returned to the
master file where the filing revisor matches the set number of the
Master Catalog Card and School Catalog Record Card to assure
accurate filing.

The District is fortunate to have a cataloger who had had key
punch training before joining the staff. This cataloger recognized
what the IBM programmers confirmed— that efficiency lies in giving
one of the library clerks key punch training rather than in attempting
to provide library background to a key punch operator. Civil service
personnel advance from the beginning positions all too rapidly, and
the Personnel Commission is understandably reluctant to establish
"one-of-a-kind" positions. The operators must understand the con-
stitution of a set of cards so that the school librarian— and there is
only one in a school— need give only a minimal amount of time to
helping the school clerk who is not always full time in the library
and who is vulnerable for advancement or assignment changes.

Service calls for machine malfunction have been minor irri-
tations rather than problems. It was anticipated that the machines
would be inactive 10 per cent of the time— some months that has
been true, other months not. The present over-all period of non-
productivity is less than 10 per cent, but the Cataloging Department

is conveniently located only a few blocks from the service department
of IBM.

The greatest disadvantage of the IBM 870 Document Writing Sys-
tem is the insufficient number of impulses to direct the typewriters to
produce all the characters that the standard office typewriter does.
The coding that is necessary to produce the appropriate results on the
print-out of catalog cards would be difficult for the average key punch
operator, but key punch operators are not hired for this installation.
Functional limitations of the equipment are frustrating. Because of the
demands upon the equipment, increased capacity is needed.

The insufficient number of impulses available in the control
unit is an educational handicap. Unless they appear on dead keys,
diacritical marks, which are a part of the spelling of many words in
foreign languages, cannot be used. Since the cards are produced for
teaching- learning situations as well as for finding purposes, this is
a serious drawback, especially since there is an increase in the study
of foreign languages in the schools. It should be mechanically pos-
sible to use manual wiring with a distinctive color directive card,
and it can be presumed that the same procedure could be followed here
that is used for square brackets if all available impulses were not
already in use. Square brackets are used somewhat sparingly because
the only means of producing them is to manually change the wiring
on the reading board. A signal is given to the operator by a punched
card of distinctive color when this change is necessary.

A major disadvantage is that the machine is an independent
unit which can take advantage of the information punched into order
cards for only a few central office records. Order records are
abbreviated and coded so that they are not directly transferable for
punching catalog cards. The by-products of the punched order cards
which have been applicable to the IBM 870 include the punching of
the School Catalog Record Card for titles new to the Master Order
file. When the entry needed for ordering is not the same as that
needed for cataloging purposes, incompatibility results, and there
is no useful by-product. Additional by-products of order records
provide some expendable working records for Catalog Department
use.

The ability of the System both to input and output has delaying
tactics in that it cannot do both at the same time. Currently 50 per
cent of its time is now used for output. As output increases so does
input. The obvious answer is longer hours or another "inputter,"
but floor space is limited, and even such a small machine taxes it
too heavily. Until, or unless, space for additional equipment is
available, it appears advisable to use machine typing in conjunction
with manually typed cards for analytics. It would be preferable
to have the document writer do the complete typing for the sets since
that would save revision as well as typing time, even though all

cards, including those machine typed, get a least a once over lightly
check.

Some related problems involve the continuous stock as it feeds
through the typewriters. Any irregularity in stock causes wastage.
If stock is not carefully joined at the junction point, there will be
wastage— not only of the defective stock but of all cards read for that
title. If the stock is too stiff, it may not hold against the platen and
may be printed in incorrect alignment. The chip between cards, if
not precisely cut, may break and catch so that vertical movement is
hampered.

The single line information on the School Catalog Record Card
has necessitated new filing directives. Because of line length we are
limited to a single initial of an author's given name. This results in
interfiling the titles of several authors in the Master Catalog file.
In order to keep corporate entries within bounds, abbreviations have
been used, but even so the title is very limited. The Cataloging De-
partment depends upon the set number for accuracy in filing.

The greatest advantage of the IBM 870 Document Writing Sys-
tem is that it is now possible for the Department to provide better
service to libraries on a more current basis. There is no longer a
backlog of work for high schools. No longer is there an accumula-
tion of records for cards to be sent later after the books and shelf
lists have been received in the schools. The book can now be sent
with complete sets of cards, including the extensive analytics needed
for collected biographies and some literary works. Cards can be
sent for all editions instead of for only a single edition of a title so
that the catalogs represent current holdings. The reduction in re-
vision time is another major advantage, relieving the staff of con-
siderable tedious work. The advantage of individualized collections
in the schools results in some sporadic orders for specific titles.
These repetitive titles presented problems of retyping and revision
in the past which have been reduced by the System, but the "reading"
of these "re-runs" requires more and more time. The System has
also negated the need for space consuming storage of "extra" cards
in the hope of avoiding the retyping of offset masters or individual
sets of cards.

It is no longer necessary to remove the Master Catalog record
for a previously cataloged title from the file to produce a set of
catalog cards. The School Catalog Record Card is removed without
substituting a temporary record. (Since every title has both a
Master Catalog Card and a School Catalog Record Card, if the School
Catalog Record Card is not in file it is in process and not immediately
available.) A green card l/4n higher than the data processing card
is prepared, giving set number, first word of entry, and number of
sets needed to be used as a temporary record in the numerical file of
punched cards.

It is difficult to discuss costs, for cost studies so often do not
agree in the factors involved. Los Angeles costs may have little
relation to those in another school system. The figures that have been
used are not definitive. Some are carefully evaluated estimates, and
they are not comparable to figures for other installations. Overhead
costs have not been included. These figures do, however, show com-
parative costs for 1960-61 and 1963-64 for items included for the
Library Catalog Department of the Los Angeles City Schools.

Major cost factors for the years under discussion were as
follows:

1960-61 1963-64

Professional salaries $19,692.00 $45,852.00 (incl. $9,864

for IBM 870)

Clerical salaries 21,846.00 39,468.00 (incl. 10,530

for IBM 870)

Card stock 1,400.00 5,400.00

Offset stock 300.00 300.00
Catalog Master

record stock 25.00 11.00

Machine rental 3,660.00

Typewriters 350.00 350.00
Burster

(prorated 5 years) 485.00
Offset Duplicator

(5 years) 280.00 150.00 [repairs]

Filing cases 300.00 900.00

Total $44,193.00 $96,576.00

Book budget
(incl. "dup") $412,069.00 $604,125.00

Sets of cards prepared 45,553 98,157

On the basis of these figures, the over-all cost of producing a
set of catalog cards, including cataloging and classification costs for
new titles, was $0.97 in 1960-61 and $0.983 in 1963-64. These figures
are, of course, in no way indicative of actual cataloging costs which
are several times that per title. No allowance has been made for
staff time spent on the ever present "related duties as assigned"
that raise costs beyond those of direct production. General salary
increases for the continuing 1960-61 staff positions alone account
for $0.0913 of the $0.983 cost of a set of cards in 1963-64. A saving
should be evident in 1964-65 because the parallel files, essential

during the change-over period, are no longer necessary, meaning
that the work routines have been shortened. But the important factor,
so long as costs remain within reasonable limits, is that more work
can be accomplished, backlogs cleared, and more service to the
schools given rather than that a unit cost saving can be effected with
the System.

There is a need to advance thoughtfully in schools; for school
librarians have the responsibility of providing catalog services in a
format which students will meet in public, university, and special
libraries as well as in their school libraries. Neither must school
libraries lag behind the developments being made in other libraries.
It is quite possible that as these developments occur school libraries
may change mightily. Computerized methods may better serve the
library needs of the future. At present books are being moved with
appropriate catalog records more rapidly than in the past. This is
a service of the greatest importance to students. The Catalog De-
partment has been more concerned about accomplishing a job than
it has been about costs, although the costs are very important. With
the IBM 870 System the Department has accomplished more without
increasing the total staff although there have been changes in job
responsibilities and internal department assignment which are not
shown in the positions chart above.

The IBM 870 Document Writing System is serving the Library
Section of the Los Angeles City Schools well at this time.

Appendix
PUNCH ROUTINE CHART

When the cataloging has been completed, the general procedure
is as follows:

1. Master Catalog Card (MCC) which becomes the "original
document" for punching is prepared by the Cataloger (Library
Coordinator) .

Information is typed in appropriate format within the
dotted lines which defines the area from which our offset
duplicator will reproduce.

2. Need for format cards is determined.

By- pass Step 3 if format cards are not needed.

3. Additional format cards, when needed, are prepared by the
Cataloger:

Shortened shelf list cards
Extension cards
Tracing cards
Heading cards

Long headings of more than one line

Unusual arrangements, e.g., author-title analytical

cards.

4. Master Catalog Card (MCC) and/or format cards are as-
sembled with School Catalog Record Card (SCRC) and for-
warded for IBM 870 processing.

SCRC is a holdings record on which schools having the
title are noted.

5. Vertical spacing is figured and coded into the appropriate
spaces on the MCC and format cards.

6. Next consecutive "set number" is punched into delineated
area at upper right of MCC and SCRC.

7. Cards are punched with appropriate coding to assure the
correct catalog form on card stock in the reading process.

Same set number is punched into each Document Writing
System (DWS) card punched for that title.

Each DWS card is additionally numbered from first to
last for the set. Card 1, which has the upper right corner
cut, records the total number of cards in the set. All
other DWS cards in the set have the upper left corner cut.

8. Punched cards are sight revised.

Format cards are destroyed.

9. Number of sets per title required is entered on the DWS
card, and cards are ready for reading routine.

READ ROUTINE

CONTINUOUS OFFSET
MASTER STOCK

1 . Document Writing System cards
sorted, grouped and stacked in
hopper

2. Paper stock selected - card or
offset master

3. DWS cards read by System

4. Stock burst

Cord stock by-passes Step 5

5. Cards reproduced from offset
masters and masters destroyed

6. Cards gathered into sets and
assembled with books

CONTINUOUS CATALjOG
CARD STOCK

63
READ ROUTINE CHART

When an appropriate number of cards (newly punched or with-
drawn from prepunched file) is ready for reading:

1. Punched Document Writing System cards are sorted,
grouped, and stacked in the hopper according to the number of
sets needed for each title.

2. Decision is made of stock to be used— card stock for five or
fewer copies of a title; offset master stock for six or more
copies.

Continuous stock is fed into typewriters.

3. Document Writing System cards are read by the System.

DWS cards are returned to file.

4. Stock which has been read is burst.

Card stock by-passes step 5.

5. Cards are reproduced from offset masters.

Offset masters are destroyed.

6. Cards are gathered into sets and assembled with books.

PUNCH ROUTINE

1 . Master Catalog Card (MCC)

prepared as "original document"

2. Need for format cards determined

By-pass Step 3 if format cards
not needed

3. Format cards prepared

4. MCC, format cards and School
Catalog Cards (SCRC) assembled
for punching

5. Vertical spacing coded and
recorded

6. "Set no." punched into MCC
and SCRC

7. Document Writing System cards
punched

"Set no." and sequence no.
within set punched

Punched cards sight revised
Format cards destroyed

9. Number of sets to be read
entered on DWS card

BRO-DART INDUSTRIES' EXPERIENCE WITH
ELECTRONIC DATA PROCESSING

Arthur Brody

The primary objective of Bro-Dart Industries is to serve
libraries. As a result of committing the management and creative
and financial resources to this objective, Bro-Dart has become the
one company directly involved in practically every area of service
to all types of libraries— public libraries, school libraries (from
elementary through university), and special libraries.

Therefore, because of the similarity of functions and problems
and because Bro-Dart is continuously trying to improve its opera-
tions as well as to supply and anticipate the needs of its customers,
the experience of the company with electronic data processing (EDP) —
the successes and the failures, the progress of the system to date,
the equipment being used, and future plans— will be described in this
report.

A service organization such as Bro-Dart must be able to make
intelligent decisions as to how time and resources should be directed,
and, therefore, statistics must be developed by products and type of
service showing the total activity, requirements, and trends in the
library field generally. At the same time, the needs of an individual
library must be known and understood. Because Bro-Dart must
handle numbers and documents accurately and quickly, it cannot be
forgotten that each library reflects the personality of its administra-
tion and the special needs and nature of its patrons.

Ten years ago the company had fewer products and was not
then engaged in the extensive program of book services it offers
today. Reports and statistical analysis were taken from hand posted
information summarized manually, but as new products and services
were introduced, the manual maintenance of information became
more difficult and inadequate. Comparative slowness and an increase
in human error in analyzing a much higher volume of documents

Arthur Brody is President of Bro-Dart Industries, Inc., of Newark,
New Jersey.

caused the company to investigate the use of electronic data process-
ing. After a careful study, it was concluded that electronic data
processing was too costly for the operations of the company at that
time, but it was believed that it would ultimately be used. Because
the study caused a close examination of existing procedures, many
ways were uncovered to cut paper work costs and to get more useful
information without installing new equipment. Anticipating the future
use of data processing, it was decided to endeavor to gear all sys-
tems and procedures as if the installation were already made, a
decision that proved rewarding. For instance, although it may have
seemed elaborate at the time, a system for numbering all library
accounts was devised. This was a first step in coding information.
Each number assigned not only indicated the library's own serial
number but showed the state in which the library is located, the sales
representative serving that library, the type of library, and its parcel
post zone. All of this information is important in the day-to-day
servicing of library customers. A numbering system of the various
products of the company was devised, keeping in mind the kind of
reports which Bro-Dart would ultimately want to receive, as well as
the limitations and possibilities of data card sorting equipment. Both
of these numbering systems have stood the test of time remarkably
well.

During 1956 it was finally decided that the company's volume
had reached a point where electronic data processing was econom-
ically feasible. A local service bureau was first used for punching
and tabulating sales analysis information from source documents the
company submitted. The key punching by the service bureau was not
as accurate as desired, and, therefore, after a couple of months, the
company installed its own key punch followed later by the rental of
a sorter, a reproducer, and a tabulator after a close and continuous
analysis of the service bureau's charges. The IBM Department re-
mained small for its first few years. Payroll was added to its work.
Then came accounts receivable. At first, the work done by the data
processing equipment was paralleled manually. It soon became
clear that the machine could do a better job in this area, and so the
conversion was completed. General ledgers, accounts payable, and
a cost system followed in 1961 and 1962, some upgrading in type and
speed of equipment being required. The installation of an EDP sys-
tem into the Supply Division was carried out as it should be. A
careful feasibility study was made, a definite plan of installation was
evolved, and additional procedures were installed on an evolutionary
basis. No existing system was abandoned until the system was
proved on the EDP equipment. Although it took seven years, it was
accomplished with minimal costs and heartaches.

If at all possible an EDP installation should be made in care-
fully planned steps, and no existing systems and procedures should

be abandoned until such time as the new ones can be proven. This
author has never observed an installation where everything immedi-
ately started to work, no matter how carefully it was planned. One
would think that Bro-Dart having been so successful in its first in-
stallation, it would have taken the same approach in other operations,
but, unfortunately, this was not the case.

In 1960, Alanar Book Processing Center, the first contract
cataloging and processing center and a Bro-Dart subsidiary, was
just beginning to stretch in anticipation of its very rapid growth of
the past few years. All manufacturers of data processing equipment
presented ideas for the use of such equipment in handling all or part
of Alanar's operations. But upon close examination, it was quickly
learned that library variants multiplied by thousands of libraries
were far too extensive for existing equipment at any cost which could
be considered by a responsible library service organization. Most
people concerned with data processing have had to learn the hard
way that the range and variety of materials used by the library is
extremely great. There are thousands of sources for books alone.
The publishing status of the books and variations in editions and
bindings add to the number of units. This is only the beginning, for
the number of possible combinations in processing books for a large
number of libraries is enormous. Any small grouping of libraries
wishing to adopt more or less complete uniformity of cataloging and
processing procedures, especially through limiting the vintage of
books to be processed, can use data processing equipment to a con-
siderable extent.

Alanar's involvement is much broader in that Alanar has acted
to provide professional and clerical manpower for all types of
libraries— performing work in accordance with the requirements of
those libraries. Recent developments both in low cost custom
cataloging through highly sophisticated use of equipment and the kit
approach to some areas of cataloging and processing will be de-
scribed later.

Although serious thought of using data processing equipment
for Alanar was set aside in 1960, by 1961 a possible new use in the
books area had arisen. In that year the company entered the book
distribution field. The deciding factor was interest by the Library
of Congress in establishing a program for the supply of LC cards
with books. New facilities at Williamsport, Pennsylvania, were
established in the summer of 1961, and by the winter of 1961-62 the
books with cards program had become a reality. In the beginning,
the new company, Bro-Dart Books, had only one customer for its
wholesale business— that customer being Alanar. But even at that
time Alanar was doing acquisitions, cataloging, and processing for
several hundred libraries.

General experience with data processing in the book distribut-
ing industry had not been marked with success. On the contrary,
there had been many instances where its use had proved disastrous.
But in setting up a completely new operation, possible use of data
processing equipment offered many attractions. Looking to the future,
complete involvement appeared inevitable, and authorities in the field
seem agreed that a thorough grounding in tab card operations is a
great help in getting ready for a computer. Having decided upon the
desirability of installing a tab card system, a program was under-
taken to gear an entire organization for a substantial degree of auto-
mation. Unlike the company's tab card system in Newark, New
Jersey, used primarily for accounting and administrative purposes,
the new installation was to serve primarily as a production tool.

Many and long sessions were held with representatives of IBM
to develop a tab card system which would provide both the opera-
tional and informational controls desired. In broad outline, the sys-
tem adopted was as follows:

1. A tab card was key punched for the total number of copies
of each title to be purchased for stock and/or to meet specific
customer orders for titles not stocked or for those ordered in
excess of the existing stock balance. Information punched into
these cards included author, title, publisher, price, edition,
and binding.

2. If perchance the title was not in stock, the original key
punched cards were duplicated to create a customer's back
order file and were further duplicated to tabulate purchase
orders placed with the various publishers and were then held
as an open record of books on order.

3. As books were received from publishers, invoices were
compared with purchase orders and necessary invoice number
and date, together with discount received, were key punched
into the tab card previously used to create the purchase order.
Such updated cards became an open file of books received.

4. A copy of the receipted card, modified to indicate a quantity
of one only, was then prepared and inserted into each copy of a
title. Such cards contained all the information required to
invoice books shipped to a library customer.

5. Through the merging and tabulation of cards showing quan-
tities of a title on order, received, and shipped— to which was
added the cumulative totals of books ordered and shipped— a
valuable progress report could be prepared as a guide to ac-
quisitions and as an inventory record.

6. The original order card was used to pick the back order
book and then used to collate out the back order cards.

In order to bring together all information relating to a title
and to combine it into a single entry, the use of a distinctive number
was recommended. Because it was intended to stock the cards with
books, using the LC number for this purpose seemed to have merit.
But it was soon apparent that: ( l) cards for many titles were un-
available and, therefore, numbers were not available; (2) the same
LC card set might be supplied for the same title against a number
of imprints because the publisher might not have been careful to send
their own edition to the Library of Congress; (3) it was discovered
that occasionally the same LC number had been inadvertently as-
signed to two different titles. In short, the LC number would not
positively identify the exact title, author, edition, and binding. Al-
though an effort was made to assign temporary numbers in the ab-
sence of a proper LC number, the problems were acute; and the
number program was abandoned. The result was full reliance on
titles, editions, and bindings— the slightest variations in which
created unbelievable difficulties in carrying out the original aims
of the system. It does not take much to knock a machine system
into a cocked hat, and the frustrations faced were at once instructive,
tragic, and costly.

The most severe blow of all was the complete miscalculation
by the machine experts concerning the ability of the recommended
equipment to do the job. The basic equipment in use centered about
an IBM 407 printer and a 604 calculator, plus, of course, key punches,
sorters, etc. The inability of this equipment to handle the required
volume of records led to the abandonment of the acquisition and
inventory report and the customer back order file. Great reliance
had been placed on this aspect of the system, and new ways and
means had to be devised under great pressure to bridge the gap.
Many of the solutions to individual problems gave rise to other com-
plications, and the number of different systems being tried and
modified led to very complex problems. The unhappiest situation
was a sacrifice in service to company customers. Although Bro-Dart
had, and has, great faith in the future of electronic data processing,
it went swiftly from optimism to despair.

The biggest weakness in the early months of involvement with
the data processing system in books operations was that it was un-
dertaken without the protection of an existing workable system which
would have run in parallel until the "bugs" were worked out. As
mentioned before, such a parallel system had been available in the
company's earlier conversion of its supply operation records and,
unquestionably, was responsible for the ease with which the con-
version was made. Bro-Dart will never again start up any operation
or convert any system and procedure to EDP without having first a
system to fall back on, particularly during the early stages of the
installation.

Growth of book distribution caused a considerable increase in
the tab card equipment required to handle the input of more than
forty key punching and verifying units, working on both day and night
shifts. As the cost of the card operation approached the cost of a
computer, the next step was obvious. An order was placed for an
IBM 1440 computer, which would do everything being done by the
tabulating card system and also provide for that area of inventory
and book order control which had been lost in the early stages of the
tab card program. The IBM 1440 series had been announced at a
cost which was lower than that of the company's existing card han-
dling equipment.

Experience already gained in data processing had begun to in-
dicate directions to be followed in computer operation. The less had
been learned well. There are headaches in launching any large scale
data processing program. It was decided to provide for parallel
operations of the existing tab card system and any new computer sys-
tem for as long as it might take to ensure success of the changeover.

A separate department was established and staffed to prepare
for the programming job ahead. But first it was given the task of
reviewing all of the company's book operations and presenting in-
dependent recommendations as to the make and computer configura-
tion best suited to the needs of Bro -Dart, bear ing in mind the
company's rapid rate of growth. The result of this review was a
determination that a computer such as the IBM 1401 or the Honeywell
200 was required and not the IBM 1440. These preliminary investi-
gations took approximately four months, and, in February 1964, the
previous order for the IBM 1440 was cancelled and replaced with an
IBM 1401 magnetic tape oriented data processing system with a
high speed printer. The superior servicing facilities then provided
by IBM in the area of the company's operations were an influencing
factor in the final decision.

The next logical step in planning for computer installation was
the organizing of a Programming Systems Group whose major task
was analyzing the various aspects to be brought under computer
control, devising the total systems concept, and defining the specific
programming jobs. It cannot be overemphasized that a good, com-
prehensive, and useful systems concept must be delineated, dis-
cussed, and approved before any major programming effort is
undertaken.

In addition to four full time programmers, more than 1,000
hours of top management time and 10,000 hours of other executive
and supervisory time— outside the IBM department— were devoted to
preparing for computer installation and operation. Partial opera-
tions were planned under three main headings:

1. Inventory control, invoicing, acquisitions, back ordering,
and general book handling, etc.

2. General accounting, which had long been handled by data
processing equipment at the company executive offices at
Newark, New Jersey.

3. Other library services (e.g., preparation of indexes to book
catalogs, preparation and up-dating of lists, and simplified
ordering and interchange of data where library customers are
using data processing equipment) .

The IBM 1401 computer was installed in December, 1964. Ap-
proximately sixty basic programs are being used for daily operational
control, representing some 4,000 hours of programming time. The
programs cover order input, book picking, back order control and
acquisitions reports, purchase orders, receiving, invoicing, customer
reports, and shipping. Some of the programs are highly sophisticated
to provide for the great variety of ordering patterns at the disposal
of Bro-Dart and Alanar customers. A few of the hundreds of possible
variations are cited to suggest the degree of sophistication required
in programming.

A series of orders may require separate billings by individual
purchase order and /or line item number— a frequent requirement.
Cataloging and processing may be separate on the same invoice, on
a separate invoice, or combined with the book price. Another al-
ternative has been developed by Bro-Dart to effect a substantial
reduction in paper work, namely, the "Intend to Buy" system. Under
this system, the company assembles books against a tentative order
or listing, submitting what is an invoice in everything but name, on
the basis of which the purchasing office issues a confirming order-
eliminating partial shipment, open items, etc. The computer must
not only recognize and conform to such specifications, but it must
keep a record of purchases to make sure that customers' budgets are
not exceeded where these have been advised.

Special services offered in the way of LC card supply, book
jacket covers, hard binding or paperbacks, prebinding, etc., must
be identified and many variations in edition and binding preference
must also be accommodated. A detailed specification must be set
up for each library to cover its usual requirements, yet with pro-
vision for the library to override general specifications on an in-
dividual title basis.

The computer must also maintain a complete inventory record
of stock books on hand, as well as those that are on order, whether
or not more LC cards are available for that given title. It also must
maintain a complete back order file by customer so that as soon as
a book that was out of stock is received, it can be shipped to the

proper customer. It, of course, must also carry out the usual ac-
counting functions relating to accounts receivable, accounts payable,
payroll, costs, etc.

The biggest task which faced Bro-Dart apart from program-
ming, was to input all of the original records. What has been done,
in effect, is to assemble in machineable form essential information
on titles now in print. At this date the record covers about 140,000
titles with about 25 per cent to go. The problems of input are many,
and the decisions are important. There is virtually no limit to the
amount of information which it is physically possible to enter on a
computer record. But should you decide, as Bro-Dart did, that you
are going to use a fixed length record of each entry, a single record
twice as long as any other record will double the length of tape re-
quired for the whole and correspondingly increase the cost every
time the tape is passed through the equipment.

Bro-Dart determined the point at which the record length would
accommodate about 95 per cent of all titles in full and then edited the
balance within this limit. It so happens that the 5 per cent of ex-
tremely long titles includes many which have limited activity.

The title takes more space than any other single item, but the
same considerations apply to author, prices, discounts, and dates.
Those who have had experience in fitting the information into the 80
columns of a single tab card know how quickly those 80 places are
used up. Most entries require a number of cards; and one of the
advantages of the computer is that once you have entered the original
information, a longer single record can be maintained even though
only parts of it are used at any one time.

As already mentioned, the amount of input must be related to
the size and, hence, cost of maintaining records. It must also, of
course, be related to the many uses of output. If you shorten a little
for reasons of operating economy, can you accept the abbreviated
title when it is printed on the list or form you have called for ?

Errors occur through the slightest programming weakness and
although all programs are tested before use, only trial under full
operating conditions can demonstrate that they are completely sound.
The simplest weakness in a complex computer program can drive
you out of your mind, but can be solved with patience and fortitude.
The errors which remain are human errors. You may wonder how
something can be key punched and verified by machine and still be
incorrect. It cannot happen, but it does. Every effort must be made
to minimize input errors, for a pure system is a joy. However, we
live in a real world, and this means one in which mistakes occur.
Errors can and must be corrected. The trick is to watch for and
recognize the side effects. The computer may some day learn to
think, but there are times when its discretion is very poor.

In reference to input problems, Bro-Dart was not happy with
existing data collecting systems and has been involved in the develop-
ment of an interesting new unit which it calls the "korn-punch." It
is believed that this piece of equipment for data collecting may be of
great value for book-charging systems.

As mentioned before, Bro-Dart's attempt to use the LC number
as an address was not successful. Hundreds of hours were therefore
spent developing a code to serve the company's purpose properly
because although many long sessions have been held throughout the
library field to consider the practicability of a universal computer
number, little progress has been made. Many of the problems will
be apparent to anyone giving the matter thought. In the absence of
such a universal number, it was decided to use a computer assigned
number fitting the following pattern. The number, alpha numeric,
consists of a maximum of ten positions. The first three or four are
letters taken from an established publisher code which will be de-
scribed later. Here is a typical example: CRN CO 22L. CRN in-
dicates that the title is published by Crown. The next two letters are
the first two letters of the title and are used for purposes of rough
alphabetical sorting by title ( corresponding to the way in which the
stocks and files are maintained) . The next three positions are for a
number ( three digits are rarely required) . This number is always
distinctive for a title of that publisher having the first two letters
of the title shown. The last position is a letter which indicates the
type of binding (L for library, P for paper, T for trade, etc.) . The
composition of this number holds considerable importance and, as
mentioned before, was most carefully considered. It was chosen in
preference to a straight numeric code on the basis of its use in many
areas of operations and because after careful testing, it was de-
termined that Bro-Dart personnel made fewer errors when using an
alpha numeric code than when using a straight numeric code.

The publisher code developed by Bro-Dart has done an excel-
lent job for the company and for many others. The code has a maxi-
mum of four digits— all letters. A publisher such as McGraw Hill
is represented by the three letters MCG. The code is designed to use
meaningful and easily recognized letters whenever possible, particu-
larly for those publishers most frequently used. The letters MCG
indicate that books bearing the McGraw Hill imprint are also obtained
from McGraw Hill. Another code, MCGW (this time four letters) is
for the Webster Publishing Company. The fact that the first three
letters are the code for McGraw Hill indicates that Webster books
are obtained through McGraw Hill, and the computer prepares its
purchase orders accordingly. RAN is Random House, RANG Bernard
Geis, RANK Knopf, RANP Pantheon. All of these are ordered from
Random House. Both the company's book stocks and acquisitions

procedures make use of this coding means of dealing with publishing
families. The Bro-Dart publisher code now covers over 2,500 pub-
lishers, and the list has been made available to many customers at
their request.

It is possible to continue almost indefinitely describing the
various additional systems, procedures, and unique programs
Bro-Dart has been forced to devise just for the book distribution
operation. Some of these are unquestionably of interest for a library
installation, but others, of course, are not. If any librarian wishes
to visit Bro-Dart and explore the computer operations in detail, he
will, of course, be welcome. Although the computer has now been
in operation for five months, some systems continue to be run on the
data card equipment in parallel with the computer; and although this
is according to plan and pieces are falling in place nicely, the com-
pany is running about sixty days behind schedule as far as the com-
plete computer takeover is concerned.

However, certain uses of the equipment are now being made
and obtaining certain results which were not anticipated for many
months to come, for example: book catalogs. When interest was
first shown by Bro-Dart customers in book catalogs, a survey was
made to develop that product which appeared to offer both maximum
utility and economy. Most book catalogs produced previously fell
under a few broad classifications:

A straight photographic reproduction (usually reduced)
of actual catalog cards. An advantage of this system is that
it preserves the full depth of the original cataloging. Dis-
advantages include the need to disturb the catalog period-
ically for rephotographing, with the added expense involved,
the space involved in reproducing the total number of cards
in the catalog for each title, and the very considerable cost
of paper and printing for multiple entries of the same title
which must be printed over and over again.

Another approach has been actually to set in type some
or all of the information that appears on the catalog cards.
The purpose of this is twofold: first, to obtain the look of a
printed book, and secondly to utilize high speed listing equip-
ment such as the List-o-matic camera. This system has all
of the disadvantages mentioned above plus higher costs.

Other approaches have used computers or data card
systems— some using abbreviated entries, and although this
produces a very economical catalog, it lacks depth of cata-
loging. Others have key punched from the catalog card
everything on it and frequently additional information. Al-
though such a catalog does have tremendous depth, its cost
is enormous, and its physical size becomes unwieldy.

Bro-Dart has the facilities to produce book catalogs by both
reproduction and computer methods. A new type of book catalog has
been developed by the company and is a blend of the two systems. It
offers flexibility and economy, which will be apparent from a descrip-
tion of the end product. Bro-Dart calls it a Register - Index Catalog.
This type of catalog consists of a basic register in which there is a
single photographic reproduction of the full entry for each title in
the collection. The location of the entry on a specific number or
lettered spot on a number page provides a permanent and distinctive
index reference to such entry, the combined number of pages, and
location. Once cross index cards are set up to carry this distinctive
number for an entry in the register, it can be found without ever
being reproduced again.

Once the register has been established (including provision of
additional copies to cover anticipated increase in use) , it is only
necessary to prepare new sheets to include additional titles added to
the library's holdings. Such sheets serve the double purpose of
keeping all those interested in touch with new additions to the collec-
tion and updating the register on a continuing basis. The computer
index is the ever changing key to the library's holdings. Each entry
in the index (whether by subject, title, or author) gives the name of
author and title, the call number by which the book can be located on
the shelf, and the distinctive number which locates the fully cataloged
entry in the register. Since the year of publication is significant in
determining whether a particular book is likely to be helpful, this
information is also taken from the full catalog information when
available and included for each index entry. Inclusion of the year of
publication further reduces the need for referring to the full catalog
information. Each index entry can also be coded as designated by
the library to indicate in which campus library or libraries a title
is to be found. If this were to be done for every title, it follows that
it would be possible to separate all or part of the catalog index by
individual library if there ever arose a reason to do so.

To demonstrate the feasibility and efficiency of such a catalog,
Bro-Dart has, to date, produced three of them, one for a junior
college district, another for a co-operative public library system,
and still another for a government research library. All of the users
have agreed that the catalog has been most satisfactory. Additional
contracts have been accepted, but the number has been limited during
the period of computer takeover. However, the success of the book
catalog and other rapidly expanding activities caused Bro-Dart to
place an order for an IBM 360 computer for installation in 1966.

The computer undoubtedly can do a number of jobs in the
technical processing field. In light of the extensive catalog card
stocks which the company maintains— both headed and unheaded— and
with extensive facilities for reproduction, it is probable that

Bro-Dart's interest in computer preparation of catalog cards will
be of a minor nature. But as mentioned earlier, libraries prepared
to accept certain limitations in scope and some variations in the
physical form of their processing (e.g., the use of labels) can now
make use of data processing. In the role of a manpower service
organization, however, Bro-Dart follows a wide range of specifica-
tions to meet the requirements of the individual library which uses
company services when, as, and if it chooses. Apart from giving
opinions where invited to do so, the only way in which Bro-Dart
may influence the course taken by its library customers is by the
higher charges which go hand- in- hand with exceptional specifications.

Bro-Dart has recently developed techniques which will make it
possible in time to offer much lower processing costs, where the
simple label kit technique is acceptable, and yet preserve for the
library a high degree of flexibility over the complete area of available
in- print publications.

Nothing is less expensive than the printing press when large
numbers of items are to be imprinted with the same information.
Bro-Dart began two years ago to build a program to lower the cost
of a limited range of titles at the elementary level to libraries serving
schools and children. This program has matured as the cataloging
and processing kits are now being made available. A national library
mailing has just been completed of a book kit catalog listing 10,500
titles for which kits will be available for cataloging and processing
books going into libraries for the new school year beginning in
September. The cost to the library for books ordered from Bro-Dart
under this program, with kits applied either by the company or by the
library, will be the lowest ever, and yet a high standard is maintained.
The book and kit catalog referred to is a product of the computer.
The author and title listings include Bro-Dart's computer number for
each title. Where clerical help is hard pressed, a library can order
the books desired by simply listing the computer numbers or marking
them in a copy of the catalog itself.

As many library customers are in or entering the data process-
ing field, there is an increased need to find ways in which further
efficiencies can be gained by having the machines talk to one another.
There are undoubtedly areas in which this is possible and, while the
machines converse, Bro-Dart staff members may have more time to
talk to librarians, who are the source of many of the company's ideas.

A problem which has been widely discussed in using computer
prepared indexes is the difficulty of following established library
filing rules. This problem causes less trouble in book catalogs for
relatively small libraries where filing similar to that used in a tele-
phone directory can be tolerated. But in large libraries, the problem
assumes greater dimension. In order to determine how best to deal
with this important problem, a professional team is now making a

study under sponsorship of the Bro-Dart Foundation, a non-profit
organization established by Bro-Dart Industries to support selected
projects of wide interest to the library field.

Incidentally, the first project sponsored by the Bro-Dart
Foundation was the selection of a school oriented book collection of
approximately 5,000 titles plus audio-visual materials. This work
was undertaken by an independent professional committee under the
chairmanship of Professor Mary Gaver of Rutgers University Grad-
uate School of Library Science. The first phase of this list was
published in March, and the total collection will be available shortly.
At the request of many specialists in the elementary library field,
the book catalog format previously mentioned has been used for this
new library tool.

It is said that when one of the first data processing units was
made available for public inspection, a request was made by a news-
man that the equipment be instructed to add two and two. Many
minutes and many chuckles later the equipment responded with the
right answer. Adding two and two on a computer is about as silly as
driving a carpet tack with a sledge hammer. Yet the foolishness of
matching sledge hammer and tack is neither as great nor as frequent
as the wasteful use of data processing equipment. During research
into Bro-Dart' s own equipment needs and procedures, IBM and others
gave the names of many firms handling large numbers of items, such
as wholesale hardware supply houses, supermarket chains, etc. It
was astonishing to learn the number of large and costly installations
which were being used but a small fraction of the time. There were
two apparent reasons: ( l) Management felt it important to get into
the act in this day of computers but were either not prepared or
shied away from giving the management time and support essential
to a successful program or ( 2) The operation did not require as
sophisticated a piece of equipment.

Many library visitors, who come from all over the country to
see the company's book operations at Williamsport, have discussed
their data processing plans. Many have access to equipment avail-
able on university campuses or with associated agencies of local
and state governments. Windfalls are always welcome, and, if a
move to data processing is in your future, availability of adequate
equipment at low cost or no cost is a big help. However, just as you
judge the fitness of a book for your library's shelves without first
thinking about who is going to pay for it, it is well to make sure that
any uses of data processing equipment be efficient in relation to the
normal cost of such equipment. In time, even a prorated share of an
excessively expensive piece of equipment could be a drain on your
budget.

A number of tabulating and computer installations have drawn
wide attention among those in the library field interested in the use

of such equipment. Much of the work done at these installations, and
a high percentage of the cost involved, must be considered as neces-
sary research and development. And, as is often the case of pioneer-
ing work in any field, first results must be most carefully examined
to be sure that, at their stage of development, they represent the
degree of effectiveness and efficiency which would commend their use
to libraries wishing to incorporate such systems as part of their
day-to-day operations.

There is little which cannot be achieved if ample resources are
available, but, for the long pull, results must justify cost. There is
an apparent initial economy in following the pattern of work done by
others; but in any comparatively new development, the risk of buying
someone else's mistakes (however understandable they maybe) must
be considered.

Most librarians have devoted their life's work to service by the
library. The author's field has been service to the library. Bro-Dart
is presently in the midst of a substantial speed-up toward new and
greater goals in library service. To the extent that librarians can
find help in knowing more about what the company is doing, it is
theirs for the asking.

HOW TO DESIGN DATA PROCESSING INPUT RECORDS
FOR OPTIMUM RESULTS

John P. Kennedy

Card Design

Inefficiencies in machine processing resulting from poor card
design can be measured in milliseconds or microseconds per record.
Even when dealing with large files, this will usually add up to no more
than a few minutes per run. If the run is repeated frequently, how-
ever, a few minutes or a few dollars difference per run may be sig-
nificant. For a large library processing its circulation file daily,
inefficiency resulting from poor card design and requiring a few
extra milliseconds for processing each record could cost the library
hundreds of dollars over the course of a year.

Usually a more serious consequence of mistakes in card design
than increased processing time is increased time in coding and punch-
ing the data. Small differences in card layout can result in differences
of several seconds per record in coding and punching. While clerical
time is less expensive than computer time, the cost of a few extra
seconds in coding the source document or punching the record will
usually be more expensive than a few milliseconds of computer time.
The most serious consequence of poor card design may be an increase
in the number of errors made in preparing the input. Any change in
source document or card layout which will result in fewer errors in
the input data will probably prove to be economical even though it may
increase processing time. The acceptable tolerance level for errors
varies from one application to another, but in many library opera-
tions, errors eventually result in problems that require a considerable
amount of professional time to solve. In cases in which the most con-
venient layout for coding and key punching is not the best layout for
processing efficiency, priority should almost be given to the concen-
ience of the persons producing the records rather than to the machine.
The limitations and capabilities of the clerks recording and punching
the data are more important considerations in card design than the

John P. Kennedy is Research Associate jointly with the Statistical
Service Unit and the University Library, University of Illinois.

limitations and capabilities of the computer which will process data.

The most important consideration for the keypuncher is that
she not have to skip from place to place on the source document in
order to pick out items in the order in which they are to be punched.
Source document and card layout should be planned so that items to
be punched stand out for easy location and occur in sequence from
top to bottom and left to right. It should be realized that any coding
which must be done during keypunching will slow down the punching
and will probably decrease accuracy. Right justification of fields
should be avoided in punching. Right justification of numeric fields
often makes processing more efficient, but it decreases punching
speed and is one of the most common sources of errors in punching.
In using unit record equipment right justification is often essential,
but in computer input it is more efficient to let the computer take
care of justification than to require the punch operator to do it. An-
other technique for facilitating the punching of long numeric fields
and reducing the frequency of punching errors is to break the fields
into shorter elements. It is difficult to keep long unbroken numbers
in mind, and transposition errors are common in recording them.
The practice of dividing long numbers into shorter elements is
familiar in the telephone number and the Social Security number.
It is especially advantageous if the elements can be meaningful. For
example, in an accession number the first digits may represent the
year of accession, or in an order number, part of the number may
represent the fund on which an item is ordered.

Works on forms design are available which detail other factors
which can improve the efficiency and accuracy of recording informa-
tion on forms of various types. 1 These may be especially helpful in
designing dual purpose cards which are used for the original re-
cording of data which will be keypunched into it. The additional
factors which must be considered in designing cards for computer
input (because of the nature of the machines) are not difficult to
comprehend and require little technical knowledge of the computer.
The one essential requirement is that the computer be able to dis-
tinguish different types of information. This is accomplished most
often by the positions or fields into which different items are punched
in the card and by codes which may be used to identify items and
card types. Good card design requires the determination of the size
and sequence of the fields for essential items of information so that
keypunching and processing can be accomplished most efficiently.

There are four questions that must be answered in planning
the card layout. These are:

1. What items of information must be punched ?

2. How should each item be punched ?

3. How large a field will be required for each item ?

4. What sequence of fields will be most convenient ?

The factors that will be relevant in answering each of these four
questions follow.

What items of information must be punched ? It is easiest to
answer this question by beginning with the final products or outputs
of the system. These will almost always be printed documents. In
order to design the input records, it must first be determined what
the desired output is; then proceed to list the items of information
required for these outputs. In addition to items that actually appear
on the final printed reports, these items must be analyzed to deter-
mine what additional data that does not appear may be needed for
production. At this point it is advisable to include any information
which may be useful even though it is not being recorded under ex-
isting routines. If it is likely that information will be needed in
future operations, it is usually more economical to include it with
the original input. Examples of information items which have not
traditionally been recorded by libraries but are being considered
for computer input at the Library of Congress are the language and
color of books. The University of Toronto Library is including the
thickness of books in its records for the Ontario New Universities
even though thickness does not appear in the printed catalogs which
are the main current output of the system.

INTERNATIONAL BUSINESS MACHINES CORPORATION
T-QIL/ Form X24-6214.I

IBM CARD DESIGN AID Printed in U.S.A.
TYPE OF CARD: CARD NAME: SOURCE DOCUMENT:

Intonation Available Ind

Card!

Sequence

Method
of

R-Refeience
C-C Unification

Q-Ouantitative

Card Field Sia

Final Design

Interpretation

Trial

Final

Field

Sequence

Field

Site

Sequence

TOTALS — *•

Figure 1
IBM Card Design Aid

Equipment and forms suppliers can provide various aids for
card design. Figure 1 is a Card Design Aid supplied by IBM. It is
convenient to use in recording the required data for card design. The
first column of the form may be used to record the information items
which are thought to be potentially useful.

Now one should eliminate as many of the listed items as pos-
sible. Often items which are ordinarily recorded are found to be
redundant or useless. For example, some libraries have found it
unnecessary to record both call number and author and title for cir-
culation records. In a manual system the author and title provide a
useful check. If the call number on a charge is illegible or incorrect ,
the author and title can be used to identify the book. With a mechan-
ized system, this may not be needed.

After having eliminated any items that really are not needed,
the remaining items should be examined if it is necessary for them
to be punched. If processing is done regularly each day it is probably
unnecessary to punch the date; the computer can add the date auto-
matically to each record processed. The decimal point in Dewey
classification numbers is another element that can be supplied auto-
matically. There is no need to punch the point and carry it as an
extra position in every record when the computer can easily supply
it in printed output. Some information may be available through
table look-ups. If particular departments or locations control specific
funds, then it is unnecessary to punch both. The fund can be punched
and a table used to determine which department or location the item
goes to. Some information may be available in other files. If there
is a vendor name and address file, then it will probably be necessary
to identify a vendor only by a code in punching order and financial
records. If there is a borrower name and address file; then only
the borrower's identification number need be punched in circulation
records. Finally, some information items may be calculated by the
computer. If a library's loan periods are determined by borrower
status and type of material, then it may be unnecessary to punch a
due date for charged items since this can be calculated by the
computer.

In the initial proposal for the conversion of one library's shelf
list to magnetic tape, a three card set for each copy was called for.
The first card would have been an author card, the second a title
card, and the third would have given imprint, location, and order
information. After review, the use of master cards for author and
title with detail cards for each copy of the title was decided upon.
The cards layouts are shown in Figure 2. Since this library has
large numbers of copies of many of its titles, this simple change
from three card sets to the use of master and detail cards resulted
in important savings in punching and processing time. Examination
of the data items included on the detail cards suggests that some

enS

cn8

'OHIO «"«

cn»

en 8

en8

e»C

3iva "*

"Sod enc

enc

enS
cnc

enS

en S
en S

9999999999

67 61 69 70 71 72 73 74 79 76

PUBLISHER
9999999999

67 61 69 70 71 72 73 74 79 71

9999999999

67 61 69 70 71 72 71 74 79 76

9999999999

67 61 69 70 71 72 73 74 79 71

9999999999

67 61 69 70 71 72 73 74 79 71

en s
en S
0.5
ens

enS
ens

en s
en S

3D 1 yd «" s

en S

ens
enS

enS

enS
ens

ena
ml

enS

"ON m ?

en z

enS

eoS

en S

ena

en Z
ena
ens
ena

Q." v"*5
Z> < pen a

ven a

enR
en S

en a
en s

enS
ena
enR

enS

ena

tns

T FORM

:HINE CARDS

S
i%

o»3

e^S
enS
g en*
n: en »

en3
en z

ens

LJ "j*

en Z

en z

£ . enS
CC oc CO en S
< O O ata
Q- Z "*

en*

en 3
nt

enS
en*

en S

ens

enS
en £
en*

= 1

O o

"* en*
ens

,_ en S

en V

AdOO °> *

en 9

en?

en S

en 9

en S

MULTIPLE LAV

FOR ELECTRIC ACCOUNTS

rfe
-II

99999999999999

29 30 31 32 33 34 35 36 17 31 39 « 41 42

99999999999999

29 30 31 32 33 34 35 36 17 31 39 40 41 42

SERIAL
NO. VOLUME

99999999999999

29 30 31 32 33 34 35 36 17 31 39 40 41 42

99999999999999

29 30 31 32 33 M 35 36 37 31 39 « 41 42

99999999999999

19 30 31 32 33 34 35 36 37 » 39 40 41 42

99999999999999

21 30 31 32 33 34 35 X 17 31 39 <0 41 42

enR

•ON "»*

NOI 1 101 «"~

en ~

enR
o>S

enR
ens

31VQ «"*
NOI 1IQ3 "a

enR
en S

enR

-*!

CUTTER
999999999999

12 13 14 15 IE 17 IB 19 20 21 22 23

CUTTER
999999999999

12 13 14 15 16 17 11 192021 2223

CUTTER
999999999999

12 13 14 15 16 17 II 19 20 21 22 21

999999999999

12 11 14 15 IE 17 II 19 20 21 22 23,

999999999999

12 13 14 15 IE 17 11 19 20 21 22 23

999999999999

12 13 14 IS 16 17 19 19 20 21 22 23

2
C

I
C
2

en S

, Q *" "

en ™

DEWEY
999999999

en z

en <"

en *•
en <M

enS
en »

en »

en 9

en •>
en M
en -
en •*
en m

in i lyppi °>~

\JQI lypm °*~

e»~

en ~

ON ayvo"'

'ON (JHV'J •"-

ON (JUV'J "" ~

H3SWr>N OM10313

b38Wr.NOHJ.33n3

M3awn\ oui03n3

HSBwnNOblSanS

H3awnN OHiosns

bsawriN oaiosna

1 H31SVIAI

1 1 U31SVIAI
00

11 Vi30
ro

•

Figure 2
Multiple Card Layouts for Shelf List Data

unnecessary punching may still have been done. Several of the items
on the detail card will be constant for all copies purchased at one
time. It would be possible to punch the constant items, the inclusive
copy numbers, and the inclusive serial or accession numbers for all
copies purchased at one time. The necessity of including both copy
number and serial number and both edition data and publication date
might also be questioned.

The items which remain after such examination provide the
answer to the first question, what items of information must be
punched ? The next question to be considered is how these items
should be punched. Every alternative to keypunching should be
considered for each item.

One possibility is that some items may be prepunched. Cards
may be purchased with transaction numbers, order numbers, or
accession numbers prepunched. In an acquisitions system, sets of
cards may be punched by the computer when a book is initially
ordered. Then by merely adding the appropriate status code, the
cards can be used to update the processing file to reflect the current
status of the book as reports are received or actions taken in pro-
cessing it. In several serials systems now in operation, input cards
for reporting the arrival of expected serial pieces are completely
prepunched by the computer.

A second possibility is the automatic punching of items by gang
punching, reproducing, or duplicating on a keypunch. If none of these
methods for automatically punching the data are appropriate, then it
may be wise to consider possible alternatives to the use of punched
cards for input. Several libraries have decided on the use of optical
scanning equipment for conversion of data from shelf list records.
Both the University of Maryland and Southern Illinois University used
optical mark readers in preparing book cards from their shelf lists.
Johns Hopkins University, in converting more data from their shelf
list, found it economical to have a service agency retype the records
in a font which is readable by an optical scanner. Another alternative
to the use of punched cards which should be considered is the use of
punched paper tape. Whether or not this is a practical alternative
will often be determined by the availability of equipment.

If it is decided that punched cards will be the best form for
input, a final alternative to keypunching for some items may be mark
sensing. Mark sensing may be advantageous in circumstances where
a few short items of information are to be recorded at various sta-
tions in the library. The University of Missouri Library has found
mark sensing useful in recording data for catalog statistics and for
transfer and withdrawal statistics. In this type of use, the errors
which are likely to occur in using mark sensing are not critical. At
the Suffolk Cooperative Library System, mark sensing is used for
adding price, discount, and classification number to previously

punched order cards. Errors would be critical in this application,
but the director reports that in their circumstances mark sensing
has been extremely accurate. 2

The main alternatives to keypunching have been considered,
and a decision should be reached as to the best method for punching
each required item. The punching method selected for each item can
be recorded on the Card Design Aid. The third question to be an-
swered is how large a field is required for each item ?

At this point it is advisable to determine whether other cards
already in use include some of the same items of information. If so,
the size of the field required is already determined. The card
columns used for an item in other cards and the size of the field can
be recorded on the Card Design Aid. It is also advisable to consider
whether there are standardized layouts in use in the library or in
other libraries or organizations which might be appropriate. A
layout suggested by H. P. Luhn which is useful in many library ap-
plications is shown in Figure 3.3 This is the format required for
use in IBM's Keyword -In- Context (KWIC) and Selective Dissemina-
tion of Information (SDl) systems. It has been adopted with modifi-
cations for use in some procedures at the Pennsylvania State
University Library, the University of California at Santa Cruz
Library, the University of Illinois Library, and in the Urban Docu-
mentation and Retrieval Project. If it is likely that a library will
wish to use either the KWIC or SDI system, then it should probably
adopt a modification of this layout for those records which will be
used in the KWIC or SDI system.

In addition to the advantages of standardization, the 60 column
field for variable information has another merit. It makes it con-
venient to use two 60 character columns for printed output. This
output format is efficient in that it utilizes most of the 132 print
positions available on most printers. It also makes it possible to
print the two columns, with a 4 print position separation on 8-1/2"
X 11" paper, by reducing to about 58 or 60 per cent of original size.
A number of libraries are photographically reducing printed output
to this size for economy in reproduction. Reduction to only 58 or
60 per cent seems to be significantly easier to use and more pleasing
to users than reduction to 50 per cent.

If the size of fields is not determined by the use of a standard-
ized format or by the size allowed on other cards, then it will be
necessary to determine the appropriate size. The field should be
large enough to record the maximum number of characters that may
be required for recording the item, unless the item can be shortened
without loss of essential information. For some items such as date,
order number, or borrower number, it is easy to establish the exact
number of characters required. For other items, such as number
of copies or price, practical maximums can easily be set. For such

Q'*

lip

a s

C E

Q-1-

flu

n
u"

3 C

75 76 77 71
75 76 77 7«

I |f\

* »

U I

!iv

e =

o 111 Ul

Hit

s s

0*1

Jill

III5 1

s s

£ S

pi jjj yj

r ^r '.•

;S ;

5 * **

S X

p \fj Z

. -

i * Tg

S IS

tj _j 0

J —

- —

• »

» 2 5 "

S X

Bo. p

» „

i '

" »

• * TJ

s s

z u

e «

z< •

• 2

*"" * T^

s s

O LJ

S°= PS

= 15*0: =
SSfsS?:!

i! fl

it 59 60 61
it 59 60 61

:»• 11

K E

2 :

I - S 5

S S

r -:;?

- :-

" -.:

:| TI

K S

DID

*I D

X X
K S

bos

||1?

S S

I J

1 =

:5 ^f

s s

0 2

s£

'= -

oi'

-•

i i

S ;

; ;

B>-

= nq

s s
s s

: a

3 3

s s

Q i

;

;;=S

S ¥

J ^ 2

liiili

-5Ss

S S

s s

J!ii

s s

Q „

i]|i|

s s
s s

0 2

* ° x 1

s s

If;

3 3

?M y

S S

BUI

E ^

1 1 1~3

S S

» 2 " MN

R R

S " - "!

S S

8W»

^ 2

•^

- x « T;

K K

:!!

;liH

S5 £
S S

" * * 5S

S S

|» s

- * « ""

a n

:- s

l!i||

S R

p 1

•| §

8s;||

s s

fc>

IIIlT

2 Z

•5 =

3 r

|3S|i

2 2

sMit

2 £

5 o s ••

r i

llsit

™ 2

• 2 * ••

— s

; = 2 TS

— —

I S " T"

2 2

BTI

l:,lg

0. 0>

P:!

" "

D i

•» -«-

0r_

III

r« p.

s
E

__^

•:

' 0

1 1

-i.

Z z

Ill 3*

Figure 3

General Purpose Card Layout for Libraries
Developed by Luhn

items as entry and title, there are no logical maximums, and it is
difficult to determine practical maximums. The best approach is to
analyze a sample of the items. After such an analysis, one can de-
cide at what point it will be acceptable to truncate the longer items
or what size will accommodate a large enough percentage of the
records that the remainder may be handled as exceptions.

Needs and practices vary so much from library to library it
is impossible to suggest appropriate sizes for field such as call
number, author and title. Among the card layouts of which the author
has copies, there is a range in the number of columns allowed for
call numbers from 16 to 51, with a fairly even distribution between
the extremes.

The use of overly long fields in order to be certain to accom-
modate any item which may be encountered will increase the com-
puter time for each pass. In effect, most of the computer time may
be spent in processing blanks. In addition, available core capacity
may be used inefficiently and tape handling time increased. In pre-
paring a list of current periodical holdings at the University of
Illinois, it was found that 90 per cent of the entries were less than
80 characters and 98 per cent were less than 120 characters. A
title field of 256 characters would have been required, however, to
accommodate the longest entry. The use of fixed length fields large
enough for the maximum size of every item will often increase the
size of records by several times and will increase computer time
correspondingly.

If it has been decided that truncation of items is unacceptable
and that it will not be efficient to deal with oversize items as ex-
ceptions, it may be advantageous to use variable length records.
Input of variable length items is usually handled by using fixed fields
on the cards but allowing for a variable number of cards. Cards
then usually contain only one item or part of an item apart from
reference data to identify the card. The type of item on each card
is indicated by card numbers and codes for card type. It is also
possible to identify variable length items by the use of flags or codes
and by their sequence.

The use of variable length records decreases the size of the
records and the file and therefore decreases the input and output
time in each pass of that file. On the other hand, it means increased
steps in processing and therefore makes programming more difficult
and usually increases processing time. H. N. Laden and T. R. Gil-
dersleeve, in a recent book on system design, give a good discussion
of the factors that should determine the choice of fixed length or
variable length records. ^ Maximum efficiency is achieved by a good
balance of input-output time and processing time. If the run is input-
output limited, the use of variable length records will likely improve
overall time since it will decrease input-output time and increase

processing time. If the run is computer limited, the use of fixed
length records will be preferable. The exact way of balancing de-
pends on machine characteristics such as buffering capabilities and
timing, but the following conditions suggest the use of variable length
records:

1. The amount of variability in record size is extensive.

2. Among the variable items, the frequency of short items is
relatively great.

3. The file contains a large volume of records and will be
processed frequently.

4. The activity of the file is low.

You will notice that these characteristics are frequently found
with the files used in library procedures in which full bibliographic
descriptions are required. In such procedures, the use of variable
length records will often be advantageous. If the advantages do not
seem clear, however, it is probably best to use fixed length records.
The difficulties in programming and problems in sorting for files of
variable length records may be more serious than anticipated. Even
if variable length records are utilized, it will probably be necessary
to establish maximum sizes for the items.

After the size of each item has been determined and recorded,
the total number of columns required can be summed. In some ap-
plications it is highly desirable that only one card be used for each
unit record. For example, in a serials checking system there would
be many problems and opportunities for error if an issue could be
represented by more than one arrival card. In such applications, if
the sum of the field sizes exceeds 80, it will probably be necessary
to decrease the size of some fields. The simplest and most common
method is further to truncate fields such as the field for an abbrevi-
ated title. In some cases it may be possible to shorten several fields
by providing a common overflow field.

Occasionally it is possible to shorten fields without loss of in-
formation by more coding of data. An item such as a date may be
represented in as few as four positions, as opposed to the thirteen
or more which we usually use in writing it, yet still appear in con-
ventional form in printed outputs. In most cases where an item is
limited to a small number of possibilities, coding may be utilized.
If the number of possibilities is less than 10, single numeric charac-
ters may be used. If it exceeds 10 but not 12, as for months of the
year, numeric and zone punches may be utilized. Use of alphabetic
and special characters permits a greater number of possibilities.
Through the use of multi- punching, hundreds or even thousands of
possibilities can be coded in two columns. In some cases, it is pos-
sible to overpunch control codes in columns used for numeric data
or to combine in a single column the coding for two variables with

only a few possible conditions. Such heavy coding should be avoided
in most circumstances, however, since coding of the data is slower,
the probability of error in coding and punching becomes greater, and
computer editing for accuracy becomes difficult or impossible. Some
additional techniques that may be used for reducing the number of
columns required in special circumstances are suggested in the IBM
manual, Form and Card Design. ^

After determining the size of the field required for each item,
the final question to be answered is what order of items will be most
convenient ? If the format of the source document is fixed, the se-
quence of items on the source document is the most important factor
in determining the sequence for the card. If the source document
has not been designed or is to be redesigned, several factors may be
considered in determining the best sequence for items on the source
document and card. If other cards used by the library include some
of the same items of information, the same card columns should
usually be used for these items. This may facilitate programming
by making it possible to copy parts of file descriptions or subroutines
from existing programs. Keypunching may be facilitated since the
operator becomes familiar with a single location in which an item is
punched, and fewer different program cards may be required. An-
other factor that may be considered is the desired output format. It
is better to arrange items so that a minimum of reformatting is re-
quired in order to produce the desired output. In many library ap-
plications, one card is used for the input of one complete line of
information in the same format that it will be listed in computer
printed catalogs or processing lists. This usually insures a minimum
of programming effort and computer time.

The nature of the information may also influence the order of
items. The Card Design Aid shown in Figure 1 provides a column
for the classification of items as Reference, Classification, or
Quantitative. Reference items, such as order number or transaction
number, identify the other data on the card. Classification items
such as call number, borrower number, account number, and status
codes, are used to group records for reports. Quantitative data in-
cludes items such as quantities and prices. Conventionally, cards are
arranged with reference data to the left and quantitative data to the
right. This classification is not very useful in most library applica-
tions, but the practice of placing fixed length reference type items
to the left is helpful. To most of us, it seems natural to place the
fields for items such as call number, author and title to the left, and
reference items such as identification numbers, card types, and
card numbers to the right. If, however, reference items which are
usually of fixed length and must always be punched are placed at the
left of the card and items which may vary in length or may be left
blank are placed to the right, punching is expedited. After punching

the required data, the operator can touch the reject button and then
find her place on the next document while the card is being ejected
and the next card registered. Location of fields to be automatically
duplicated at the left or right end of the card also adds to the time
available to the operator for finding her place on the next document.

Finally, the sequence of items should be planned with con-
sideration for machine limitations and capabilities. In the use of
equipment which reads input serially such as the IBM 357 units which
are used in several library circulation systems, any blank columns
should be at the right of the card so that machine time is not wasted
in reading blank columns. If several items are to be used in one
sorting operation, it is desirable that these items be in adjacent
fields with the major element to the left and the minor element to
the right. In some cases, the order of items may determine whether
it will be possible to chain instructions and thereby save a little
processing time.

The location of items on dual use cards presents several ad-
ditional factors for consideration. In planning the layout for dual
cards, it is important to consider the visibility of items to be punched.
Information on the card may be partially concealed by the punch
housing unit or by the pressure arm of the keypunch. It may also
be necessary to position essential written information so that it will
not be obliterated by the punches. The use of dual cards has ad-
vantages in many applications. The card may include information
such as signatures which cannot be punched or which it is unnecessary
to punch. For example, in a circulation procedure, a tabulating card
form might be completed by the borrower. It would be necessary to
punch only a few items into the card for machine processing, but
the complete record including signature and address would be avail-
able for overdue procedures. The use of dual cards often eliminates
the need for typing the card to a source document through reference
items and the necessity of maintaining two files.

With the determination of the sequences of items on the card,
all of the information needed for the final layout is at hand. A num-
ber of card layout forms are available for use in recording the lay-
out and for ordering custom printed cards. IBM supplies a multiple
card layout form for sets of cards, a general purpose card layout
form, a dual card layout form, and a number of other layout forms
for more unusual types of cards. Figure 2 shows a multiple card
layout form, and Figure 3 shows a general purpose card layout form.
These forms include a number of guides and scales useful in drafting
the layout. Instructions for the use of these forms are included in
the manual, Form and Card Design.

91
REFERENCES

1. Marien, Ray. Marien on Forms Control. Englewood Cliffs,
N.J., Prentice- Hall, 1962. Sadauskas, Wallace B. Manual of Business
Forms. New York, Office Publications, 1961.

2. Curley, Walter W. "The Data Processing Program in Op-
eration at the Suffolk Cooperative Library System, Patchogue, New
York," see this volume.

3. Luhn, Hans Peter. General Rules for Creating Machinable
Records for Libraries and Special Reference Files. (Form No.
225-1487). Yorktown Heights, N.Y., IBM Corp., Advanced Systems
Development Division, 1960.

4. Laden, H. N., and Gildersleeve, T. R. System Design for
Computer Applications. New York, Wiley, 1963, pp. 92-97.

5. International Business Machines Corporation. Data Pro-
cessing Techniques: Form and Card Design. (Form C20-8078).
White Plains, N.Y., IBM Corp. Technical Publications Department,
1961, p. 8.

FLORIDA ATLANTIC UNIVERSITY LIBRARY

Edward Heiliger

Florida Atlantic University is a new upper -division and grad-
uate state university on the southeast coast of Florida, at Boca Raton.
It occupies a former government air field of some 1200 acres. The
first phase buildings were occupied last summer, the second phase
buildings are under construction, and the third phase are in the final
planning stage. The first phase library occupies three floors of a
five story building, the second phase will occupy the whole five floors,
and the third phase (first addition) will double the five floor space.
The Computer Center, with its IBM 1460 computer, is in the library
building during Phase I only. The IBM 360 system has a delivery
date of July 1966, the occupancy date for Phase II. The University
now has 1187 students. This number is to grow to 10,000 by 1970.
Colleges of Social Science, Science, Business Administration,
Humanities, Education, and Engineering form the university
organization.

The first year of work on the implementation of the computer
based system began in July of 1964 in an old Air Force firehouse on
the campus, with a staff of six professionals, eight clerks, and no
collection. As of today, approximately $500,000 has been spent for
books and journals, the professional staff has been increased to
fourteen, 23,000 titles have been cataloged and Library of Congress
(LC) cataloging information for these put on computer tape. Authority
files have been put on tape. A similar number of titles, bought in
block purchases, has been IBM listed (with author, title, and fixed
location number print-out) and is awaiting cataloging.

The first problem that had to be solved in the implementation
of the computer based system was to teach the computer to print out
the catalog in an order approximating that of the ALA Filing Rules.
This was resolved by developing a coding system for the use of the
catalogers, instituting new keypunching techniques, and doing some
special computer programming.

Edward Heiliger is Director of Library and Information Retrieval
Services of Florida Atlantic University in Boca Raton, Florida.

The second problem was to enable the computer to print out in
both upper and lower case, with all of the diacritical marks for the
western languages. A special computer chain was developed by us
and made to order by IBM. The Yale-Harvard-Columbia medical
libraries project and a group of colleges in Toronto took the same
chain, with modifications to suit their special needs. It was agreed to
agree on the first 88 characters so that any library wishing to use such
a chain and paper tape input from a standard keyboard, could do so.

A third problem was to create authority files on tape, enabling
the computer to print out authority lists and to provide the catalog
print-out with the necessary cross references. This problem has
been solved except for subject cross references. These have been
coded and punched, however, and are only waiting for programming.
The second edition of the catalog, coming out this summer, will have
them. "See also" references are often blind. The first edition of the
author catalog, which came out in September 1964, had 20,000 entries
for 14,000 titles. Not all authors require an author authority entry.
Only those where variant forms are a problem are included. Many
of these are corporate authors. The title authority file is much
smaller and is mostly for series titles.

The coding system for cataloging input has been adequately
described in two articles by Jean Perreault.l Decisions have recently
been made to eliminate all of the collation coding area except that
part concerned with personal and corporate authors. This does not
affect cataloging content but does limit the search capability for
information retrieval. Arguments favoring this abandonment went
like this: "Why should we go to the work of preparing input which
would enable us to list all of the books in the collection published in
France in 1909, on the subject of mo Husks, when a quick examination
of the subject catalog under mollusks would enable us to spot the
items published in France in 1909, very easily?" If the use of the
collation code provided more information than the cataloging copy
itself, then, of course, this reasoning would not be valid.

Important changes in computer programming and key punching
have just been made (March 25, 1965) which provide proper syllabifi-
cation for word breaks at the ends of lines and starts series entries
at the margins at all times. Proofreading is simplified. Formerly, we
were unable to read proof on the tracings beyond the 40-character
point. We are now able to read proof on the whole tracing. It is now
easier to make changes in cataloging. There is less rigidity, inas-
much as each area can be expanded further. There is less need for
frequent consultation between catalogers and key punchers, since the
problem of limited areas never has to be solved by asking the cata-
loger for deletions. LC forms of entry will be usable without ques-
tion, since none will be too long to be accommodated in Area 10.
Area 31 will accommodate considerably more information, e.g.,

"dashed" supplements, or as may be desirable in some future cases,
statements of holdings. Uniformity of format between main and
added entries will be achieved. The computer will not be required
to shift information from one line to the next. The above has been
accomplished by:

1. Pre -formatting for printing at 40 characters per line.

2. An additional 9 columns (41-49) may be used to account
for spillage due to punching multi-column or non-printing
characters. No text must appear beyond column 49.

3. Each new trace begins on a new card.

4. Columns 67-68 contain the two digit area number.

5. Column 69 contains the sequence number of each specific
trace (0-9), hence a maximum of ten traces are allowed for
each area.

6. Column 70 contains the card number (0-9), hence a maxi-
mum of 10 cards per trace. Example: For a particular item,
three subject tracings are needed. The first contains 60 char-
acters, the second contains 100 characters, and the third con-
tains 35 characters.

FLORIDA ATLANTIC UNIVERSITY LIBRARY CATALOG INPUT RECORD.

UZJL3

10 -BLLEti

&EOR&.&

E.Sttfc.!i

4 ' t ii

(\ F

>/ 0

22 It

LEI^ Isl

23- — * <^A ip

B l OJ6R

fc.Vtt \

Cf.

u It

/t^ r

ft L

, s

Re L K T i

w, _. <, c

e ijj i ST

S OF

f<e

i T

31-x ULU!&PC

a+£f Row i

M X) 1 t4

I w

C 1

1 1.

1 1C 1 9O£

. To u

H 1 C.A

^> '

\ T P

e M

f J

70 i«r~E;

> S A 1 of - c

t V J K to

0 K C

ft £

, e

y e

- 1

K l u-T i

QF B R 1 T L •

> rt £t i c

ij ce .

CMT

> 0 V

\T -

76 V LcDte

1910. t '

17 PI

R E M C € JO <

> x. s Jo

& Si _i_ O

1 Kj

78 TRouBi

ED uJO Ru D

MO. 1

up P l_e * 6 «JT

1 9OO

TO 1

^ 5

0 ,

c r

,GJf

80 £< €UE

•jfr, Neu, ^

L^,_E

U £ M K

IA !

\_\

"57

1 2

. A3

/* £ c i e u

L i s iv ^ g^i

-r i SB. ~ "

e, oft

31 / Sc i evj

.£- -&T . B

0 T. -- hi

l ^T .

/«•

//rBBSai!

: . W 1 L u l »i»

\ PL ft* E

we. 6

be.

i BL

cj^

W[ Bf

o c K e ( o ^

E >. R. L

Tiii.2

- ^

1 ^u

-'t-

' ?£"!*!

>Pi C.H i e \/t*

1C .j-j-^ Q

F D i?

1 T

l *;> t

*. ^

C 1

- ^ A

76-> [ 1 6 L C

^& ^ 3 EQ R c

*E+ (2p

l -

78 Sc i e "J"

r i a ii C,F -

r « C AMI

T<-DL

KJT

^J ^ !

•- c ft

^T M

: ^Kc. H i e \t e •

<e iJr S c

F Si R

V T

l SV

C. I

fj<

1 E .

22.

80 — »• £1^1.

r z^

Figure 1

IN AUTHOR CATALOG:

ALLEN, George Herman Michael Trevor
Allen, 1st Baron. 1876-1952.

An index of biographical materials
relating to scientists of the United
Kingdom, from the founding of the Royal
Society to 1900. To which is appended
the essay of Lord Brooke on the achieve-
ments of British science. London, Rout-
ledge, 1910. 647 p.

Reference books to guide us in a
troubled world, no. 12.

— Supplement, 1900 to 1950, by George
Fielding. New York, Putnam, 1957. 312 p.
QLU.M

BROOKE, William Clarence Scribblerus
Brooke, 3d earl. 1852- 1927.

The achievements of British science.
ALLEN, George Herman Michael Trevor
Allen, 1st Baron. 1876-1952.

Reference books to guide us in a
troubled world, no. 12.

— Supplement, 1900 to 1950, by George
Fielding. New York, Putnam, 1957. 312 p.
QLU.M

FIELDING, George, 1901-
ALLEN, George Herman Michael Trevor
Allen, 1st Baron r 1876-1952.

Reference books to guide us In a
troubled world, no. 12.

— Supplement, 1900 to 1950, by George
Fielding. New York, Putnam, 1957. 312 p.

These entries appear in the author catalog as a result of the coded com-
puter input appearing in the form of the preceding example.

Figure 2

IN TITLE CATALOG*

An index of biographical materials
relating to scientists of the United
Kingdom,

ALLEN, George, Herman Michael Trevor
Allen, 1st Baron. 1876-1952.

Reference books to guide us in a
troubled world, no. 12.

— Supplement, 1900 to 1950, by George
Fielding. New York, Putnam, 1957. 312 p.
QUl.M

The achievements of British science.
ALLEN, George Herman Michael Trevor
Allen, 1st Baron. 1876-1952.

Reference books to guide us in a
troubled world, no. 12.

— Supplement, 1900 to 1950, by George
Fielding. New York, Putnam, 1957. 312 p.
Q141.M

Identical entry under:

Biographical materials relating to
scientists of the United Kingdom.

These entries appear in the title catalog as a result of the coded com-
puter input appearing in the form of the preceding example.

Figure 3

IN SUBJECT CATALOG:

Scientists, British— Biog.

ALLEN, George Herman Michael Trevor
Allen, 1st Baronr 1876-1952.

Reference books to guide us in a
troubled world, no. 12.

— Supplement, 1900 to 1950, by George
Fielding. New York, Putnam, 1947. 312 p.
QU1.M

Identical entry under:
Science — Gt. Brit. — Histc

These entries appear in the subject catalog as a result of the coded com-
puter input appearing in the form of the preceding example.

Figure 4

Area 70 is formatted as follows:

Card Column: 1 through 40 67-70 71-80

Text of first card, First Trace 7000 Access No.

Text of second card, " " 7001 "

Text of first card, second trace 7010 " "

Text of second card, second trace 7011 " "

Text of third card, second trace 7012 " "

Text of first card, third trace 7021 " "

All former formatting rules remain valid. Maximum number
of cards per entry is thirty-two.

It was decided to bring out author, title, and subject catalogs
separately. There was some discussion of having a title index to
the author catalog, and this is still being discussed. The first author
catalog, with 14,000 titles and 20,000 entries, was slightly smaller
than either the title catalog or the subject catalog. Its 463 three-
column pages average about 30 titles per page. The computer took
four hours to print it out and another ten hours to do the edit and
sort (filing) work. The latter (edit and sort time) is a one time

1«6»- 1943 ne CANIIIY, PHI lip Jon», me- JglBl »e COBNCLL,

loo.-Ub I9t>0

AY*. Crn.t, 190«-

BM71.M

fi-
nd [1»6]

169 p.

(TH« »€f.r.
DT779.7.R26

I, Herlene -.-r.j, ^S1-J Bo, ion, little, Bro.n [1963] 427 p.

JK4B2S 1963. H2
HAkkA, Cenev. II

le, «A4 reading He CLOSKCY. Allen Lyle, 1922- fd. XU2J7.C35-J 1V&9

d] B.rC.n. «. He- PROGRESS In boron en. .[.try. v. 1.

Ken York, PI. e. linn, 196«- . For He CR«C«C». J L

COLLUUCM, Ce

(1963) He COLLUH, John I., 53.

Bo. ton, Hougnton Hifflli, [1961] 166

236 p. , Jo

16- Jsi»l P«2»92.rl2» Met

Figure 5

Page from the February, 1965, Cumulative Supplement
to the Author Catalog of the Florida
Atlantic University Library

expense. About 6600 titles were prepared for computer input before
the arrival of the computer on July 1, 1964. This material was never
proofread, but work is beginning on this for the second edition. Since
July 1, there has been daily proofreading. The first supplement to
the catalog, which included 3,518 titles, came off the computer on
November 3, 1964. The second (cumulative) supplement, with 7,829
titles, left the Computer Center on February 6, 1965. There will be
a third in May, 1965. After the production of the second edition this
summer, monthly supplements are planned. The second edition
promises to be twice the size of the first edition.

The computer copy for the catalog is sent to the Duplicating
Service of the University where it is photo-reduced to 58 per cent of
the original size by a photo-direct process using the Addressograph-
Multigraph 705. This by-passes the negative and produces a film-
based plate from which the offset machine produces 150 copies.
Three plates are produced every two minutes, enough to feed three
presses. A thermobind unit called Perfectbind binds and applies the
paper cover.

Copies of the catalog are to be found at twenty catalog stations
throughout the Library. There is no card catalog nor any catalog
center. Each cataloger has a copy of the catalog, as does each ref-
erence desk. There are copies in each faculty department. Some
professors have their own copies. Catalogs are also sent to the
libraries of the other state university campuses, to the nearby junior
college libraries, to the local public library, and to the Library of
Congress. No copies have been sold, but some have been loaned or
given. The Library has been approached with a proposal for mer-
chandising both its catalogs and its tapes. Suggestions are also being
considered that Florida Atlantic University Library might contract
with other libraries to provide their catalogs.

Faculty members find the nearby catalogs useful in book selec-
tion, checking the catalog before sending in an order, and sometimes
choosing an alternate title that the Library already has. They also
use it for counselling students, for preparing reserve lists, and for
planning research papers. All seem to like the book form and the
convenience. Students are equally pleased. They can consult a cata-
log wherever they may be in the Library, they can xerox a page or
pages of the catalog for research purposes or for bibliographies or
they can borrow a copy. They like the book form approach.

Until such time as the Library can go on line to the IBM 360
system, a weekly "official catalog" print-out will be needed. This
will have a full tracing. Author added entries will be included. New
author and title authority entries will appear too. This will keep the
author and title authority files up to date. The official catalog print-
outs will be cumulative weekly for a month, monthly for a quarter,
and quarterly for a year. One computer print-out with its carbon
copies will be adequate for cataloging and reference uses.

100

Supplement 2/1/65

Florida Atlantic University Library
Author Catalog

MAYNAHD, Sir John, 1B65- 1943
KUNOVALUV, Serge, $d.

Russo-Polish relations; an historical
survey.

London, Cresset Press, 194b. 90 p.
DK418.3.R9K6 194Sa

NAYO, Henry Bctram, 1911-

Introduction to Marxist theory*
New York, Oxford Univ. Pr. , 1960.

334 p.

First issued as Democracy and Marxism.

HXB6.H36 1960

MAYK, Ernst, 1904-

Animal species and evolution.

Cambridge, Belknap Press of Harvard
University Press, 1963. 797 p.
(IH371.H3J

HAZIA, Daniel, 1912- ed.

General physiology of cell specializa-
tion [by] ** [and] Albert Tyler.

New York, McGraw-Hill [c!963] 434 p.
QH573.M3

MAZLISH, Bruce, 1923-
BHUNOWSKI, Jacob, 1908-

The Western intellectual tradition,
from Leonardo to Hegel [by] ** [and]
Bruce Mazllsh.

New York, Harper [c!960] 52«J p.
M7VI.B7&

OUK, Anatole Gregory, 1900-
inland between East and West.
rinceton, N.J., Van Nostrand
.

MAZOUK,

P
2*0 p

[19561

He CAKTHY, Philip John, 1918- joint

•altar

STEPHAN, Frederick Franklin, 1903-

Sampllng opinions; an analysis of sur-

vey procedure, by ** and Philip P. Mc-

Carthy.

New York, Wiley [1*58] 4J>1 p.

HM263.S84

Me CARTY, Henry R.

The cooperative approach to audio-
visual programs. Prepared by •• and
Horace C. Hartsell.

[Washington, 1959 ] 80 p.
LB1043.M23

Me CAHY, James L., ed.

Psychology of personality; six modern
approaches.

New York, Logos Press, 1956. 383 p.
BF69B.N2214

Me CLELLAN, Aubrey L.

ira author

PIMENTEL, George C.

The hydrogen bond [by] ** [and]
Aubrey L. McClellan.

San Francisco, Freeman [c!960] 475 p.
Q0471.P3

Me CLELLAN, Aubrey Lester, 1923-

Tables of experimental dipole moments.
San Francisco, Freeman [1963] 713 p.

QDS71.M27

Me CLELLAN, Grant S. ed.
South Africa.
New York, H.W. Wilson Co., 1962.

ICO x

Figure 6
Sample Print -Out, Two of Three Columns

The dependence of the Library upon the Computer Center and
the Duplicating Center should be emphasized. At present, the Com-
puter Center is administratively under the Registrar, which has
resulted in a tendency to give programming priority to the Regis-
trar's work. The first draft of the University's constitution gives
independent status to the Computer Center, and this should create a
more equitable situation. Two of the programmers on the Computer
Center staff spend considerable time on library programming. One
devotes himself to cataloging programming and the other to serials,
acquisitions, and circulation programming. Although many on the
library staff can do simple programming, the programming has been
left completely to the Computer Center staff. The Director of the
Library is a member of the faculty's Computer Committee which
has made important decisions, such as the one to order the IBM 360
system. Hopefully this will become an advisory committee to the
Computer Center; it has been so indicated in the draft constitution.

101

The Computer Center, in catalog production work, takes about
one week to turn out the copy needed by the Duplicating Center. The
latter then takes about two weeks to provide the first 100 copies,
which serves the Library and faculty needs. The remaining fifty
copies take an additional two weeks. The welfare of the Duplicating
Center and of the Computer Center are of direct concern to the
Library and the library staff works closely with both.

The IBM 357 Data Collection System is used for preparing com-
puter input for the circulation records that are maintained on com-
puter tape. This equipment takes the book's ID card and the reader's
ID card and produces two identical transaction cards. One of these
enters the charge on the computer record, and one erases the charge
when the book is returned. The computer record is updated each
day, and each day the computer prints out from the circulation tapes
the following: a listing of all books in circulation arranged by call
number, giving the ID number of the borrower, the date due, and
indication of faculty, staff, graduate, or undergraduate status; a list-
ing of all books in circulation by borrower's ID number (his Social
Security Number), including the same data as above; a listing of
overdues in both orders; and a listing of hold requests. There is a
weekly print-out of books on reserve and a reference shelf list. A
meeting was held with the Computer Center staff on the matter of
statistics. The Library indicated a variety of statistics that would
be needed, and the programmers said that these needs could be
satisfied and that the Library would be supplied with statistics needed
on demand. To date, the Library has been so supplied.

A number of problems has arisen in connection with the circu-
lation system. The book cards are punched with the call number and
certain controls. The control punches are gang punched in the Com-
puter Center. The call number punches are made in the Circulation
Department of the Library. In one batch of gang punching, an error
was made and not caught. This resulted in some 800 faculty cards
being placed in new books. The computer refuses to accept transaction
cards made from these cards. Fortunately, a visual check of the trans-
action card shows this up. Shelf reading also helps. Another problem
has been the need to punch cards for material that does not have a
book card. This includes uncataloged government documents, maga-
zines, and vertical file materials. This was solved by releasing the
key punch on one of the IBM 357 stations to punch cards at the time
of charging. At present, one station handles the 1100 student load,
and the Library reserves the other for special charges. When the
load becomes heavier, the second station can be converted to key
punching at will. For uncataloged government documents, the Docu-
ments Office classification number is used. For other materials, a
straight alphabetical approach is used.

When Florida Atlantic University Library lends to another
library, it creates a borrower's badge for this library and checks the

102

materials through the IBM 357 system in the same way as for any
borrower. The borrowing library is given an identification number
established by the University Library. When the University Library
borrows from another library, a manual file system is used.

*£* Swis.

?*'-*£

-**-

^aa-i

t?r

Figure 7

Daily Circulation List, Call Number Arrangement
(ID Sequence is also Printed Out)

Material that is overdue is recognized by the computer, and
the item and the borrower are placed on the overdue print-out list.
The computer prints gummed labels, one name and address label
and as many others as are needed to print all the call numbers for
overdue materials for the person whose name appears on the name
label. These labels are placed on pre-message, 4- cent post cards
which become the overdue notice. When overdue material is returned,
the return card will clear the circulation list and the overdue list.
Faculty charges do not become overdue. All books are due two weeks
from Tuesday, so that overdue work can be batched.

Borrowers may place holds on material that is in circulation.
The borrower's ID number and the call number of the needed ma-
terial are entered on a transmittal sheet from which IBM cards are
key punched as input to the computer. The computer holds the data
and prints out a list of hold requests. When the particular materials

103

are returned to the circulation desk, and the return cards for them
are sent to the Data Processing Center, the computer recognizes,
upon making the match, that the material has a hold placed on it. A
list of materials to be held is then printed. At the present time,
because circulation is not too great, the processing shelves in cir-
culation can be checked against the hold print-out, and books listed
on the print- out can be retrieved before they are shelved. When
circulation becomes larger, the plan is to create a list of holds on
magnetic tape. Just before shelving, probably two to three times a
day, the return cards will be matched against the hold tape. The
result of the match is a "materials to be held" list which includes
just those materials, out of the many waiting to be shelved, that are
to be held. The staff member can then go to the processing shelves
and retrieve just those items that are to be held and shelve the rest-
probably less than 1 per cent of the total awaiting shelving.

Serials are now being checked in by pulling computer-produced
cards from a tub file. The computer record from which these cards
are produced is created by using the old serials coding form for pro-
viding computer input for each journal. This coding form has been
changed slightly and is now undergoing a major revision. The cur-
rent form is described in detail by Ted Srygley.2

The first part of the new form is shown in this article. Pro-
gramming has been completed for the use of the new form, and an
early delivery date has been promised for the new Current Serials
List. Both full title and short title have been coded in for all journals.
It was first thought that the short title would serve for the Current
Serials List, but users have experienced such difficulties with the
abbreviations that it was decided to use the full title for all serials
lists being used by the public. The first Current Serials Lists will
use the short title, but subsequent lists will use the full title. Serials
title entries will be the same as those set for the catalog. Program-
ming is being done for claiming letters. Human recognition will be
used for deciding for which items claiming will be done. A manual
run through the cards left in the tub file will determine which cards
to send to the computer for the claiming routine. There are too
many imponderables to rely on the computer for determining which
items should be claimed. The Library's holding list is unsatisfactory
and is being reprogrammed. The end result will approximate the
form of the UCSD list, with perhaps some effort at space saving.

It should be emphasized that serials records must be corrected
before they can be successfully computerized. After the information
about the journals being received had been put on computer tape, the
computer was asked for a check- in card for each issue of each
journal to be received during the next six months. In answer, it
gave only about half the information requested, because it had been

104

given either incomplete or incorrect information. This was cor-
rected, and now there are adequate check- in files and correct serials
records on tape.

Subject print-out of serials can be easily arranged and provides
a useful approach to the serials collection. One such was printed out
last year and the Library is asking for another shortly. Faculty
members (new ones particularly) frequently ask what journals the
Library has in a certain field. In adding serials titles, such a listing
is useful.

Acquisitions procedures are in a *go" state, but waiting for the
completion of Business Office programming. The transmittal sheet
and processing information list elements include: author, title,
edition, date of publication, series, place of publication, volume num-
bers, price, notifications code, number of copies, fund number, call
number, date of order, status, process number

L I B R A RV SERIALS
3- IS-t>S

(coil. •»-»<>)

CARD FORMATS 00 - 46 MEW »t>o
SO - »» CMAMOC

99 BEUtTE

tAHO k
COLS' r

1 ••», •

'111! ....^

se«.i*u «»

8
8

150

FUIL TlTtS

^-.v-f

. r

a
<

•

2.0.O

- E

iff

ijt J

f L

190

J6O

>j'j » r t ^

vtKioon. CITY IT.Tt n C.UUTK.Y NO.

'"I:

Figure 8

Partial Coding Form for Entering Serials Information

on Computer Tape. Additional Form Includes

Financial and Binding Information

105

TLV ACTIVE TITLES.

1CATION PUBLISHED LATtST ISSUE RtCEIVED

USUALLY REC*O. 2 MOS.CAALV

Figure 9
Weekly Serials List. Must Type Required List for Serials

(accession type number instead of present LUHN number), purchase
order number, pagination, and vendor. The book request form
(transmittal sheet) goes to acquisitions for approval of purchase
where the bibliographical data is verified and the request form cor-
rected. Upon approval, the transmitted data is key punched and sent
to the computer.

The computer will then print out on order the following: pur-
chase order, the Processing Information List, a Drop List for items
already in the Library, and cards for the Acquisitions Department and
the Cataloging Department for their use in providing data for the up-
dating of the Processing Information List. This list will include all
items that have been ordered but are not yet in the catalog, including
unprocessed gifts. It will indicate the whereabouts of each item in
the work process and will be printed out on a daily basis.

106

o*oo«o**ooo

sss*»s»se oo ooo« oooot t/

SSH9MMM OO *OS3 OOOO2 2/

i «3 00 09*0

» «0 00 0-»fcS

Figure 10
Master Tape Record List for Serials

The Library's organization is arranged to suit the needs of the
new system. In this area, as in others, the Library is changing as
the system develops. Data Processing, Liaison Services, and In-
formation Retrieval Services are the three main divisions. Data
Processing includes technical services plus circulation; Information
Retrieval Services include reference work and planning for the uses
of the IBM 360 system; Liaison Services concern daily contact with
the faculty on faculty needs in courses and research. The latter
division now has only one person, who spends his time consulting
with faculty members and reporting back to the Library staff. This
has proved to be highly effective. Faculty needs are determined
ahead, fruitful suggestions are made to faculty, faculty gripes get a
hearing, and the professor gets a feedback on student reactions. The
Head of Liaison Services also serves on the faculty Curriculum
Committee, and provides that committee with library information
needed in planning for new courses.

There is the possiblity of making all three division head posi-
tions research positions, with these and the department heads being

107

Chief of Chief of
Cataloging Acquisitions

Chief of
Circulation

1 Assistant
Acquisitions Librarian

1 Social Sc
Reference

Assistant
Cataloger

Head of
Informal
Retrieval

on
Service

Chief of General
Reference Service

Science
Reference

Humanities
Reference

Assistant
Cataloger

Figure 11
Florida Atlantic University Organization Chart, March 1965

responsible to the Director. The Data Processing Head would then
conduct the user study of the system and maintain liaison with the
Computer Center, the Information Retrieval Head would plan the uses
of the IBM 360 system, and the Liaison Services Head would accept
responsibility for development of the SDI (Selective Dissemination
of Information) system and gathering of faculty reading interest
profiles for the computer, as well as carrying on his liaison work
with the faculty.

As planned, the SDI system will treat only of book materials
and will match the reading interest profiles with LC subject headings
assigned to the books. The system will be designed for the University
situation, but full advantage will be taken of the IBM experience with
its successful internal SDI system.

The use study is badly needed to help give direction to the
development of the system. User reaction to the book catalog, for
instance, can help determine those features of the catalog which
should be preserved and those which should be changed. Readers
must be trained in the use of the new machinery and the products of
the new machinery. This library is concerned with the best ways
to do this. It is interested in the relative use of the author, title,
and subject catalogs. Which is most important? Do supplements to
each need to be produced with the same frequency? Do subject
headings in the computer-produced catalog do the same job that they

108

do in a card catalog ? Is the filing order better or worse because of
concessions to the computer ? In the matter of the Processing In-
formation List, the Library is interested in who the users of this list
will be and what information they will seek from it. Will two forms
of the PIL be necessary, one for public use and one for staff use ?

The completion of Business Office programming on March 31,
1965, will begin the completion of the system, and user studies can
begin this summer. The catalog will then have passed through one
edition and its supplements, and one generation of users will have
had an opportunity to test its use.

We decided at Navy Pier that information retrieval was not our
dish of tea. We also decided that bibliographic searching was strictly
a manual job. With the advent of the IBM 360 system, we are re-
considering information retrieval. We suspect that the costs per
search will be much lower. Also, the on-line approach feasibility
makes this possibility more attractive. The potential use of the IBM
360 system is concerned with non-reference library applications, too.
The present research for reference uses revolves around ways and
means of instructing the computer in how to conduct bibliographic
searches in the computer catalog record. It is hoped that by creating
a device which will enable the computer to establish a hierarchical
system for the LC classification and subject heading list, a better
kind of literature search will be made possible. On-line applications
for non-reference uses present interesting possibilities. For in-
stance, when we are on-line to the computer, a circulation transaction
can be halted by the computer if the borrower has any books out
overdue. This could eliminate the fine system. The hold requests
could also be handled more efficiently. In serials, updating of the
computer record could be done immediately as the journal is checked
in. In acquisitions, bibliographic checking could be done immediately.
An experiment is now in progress on bibliographic checking, using
the present equipment. The titles indexed in the 1963 Essay and
General Literature Index were checked against this library's cata-
log by a clerk. The staff is now programming to conduct a search
of the same titles on the catalog tapes by the computer. In both
cases, both author and title approaches are being used to see if one
is better than the other. The cataloging use of the on-line approach
and read-out facility of the IBM 360 system may do such things as
eliminate the print- out of the "official catalog" of the authority lists
and of the shelf list. Catalog corrections may be made on this basis
also.

The Computer Center is staffed with capable programming and
systems people. The IBM 1460 computer and the IBM 1403 printer
equipped with the special print chain is servicing the Library, the
Registrar, the Business Office, and some courses in the College of
Business Administration. Because we are waiting for the IBM 360

109

system for research-type computer use, some use of the equipment
by the science faculty has also been made that should have been done
on a larger computer. The result was much more programming and
computer time than the type of problems justified. Recently, however,
an arrangement with the University of Miami was made to use their
IBM 7090 at the same rates as are charged to their own students and
faculty. It should be repeated that the Computer Center should be
administratively free of any one area so that it can be just in allotting
time. With all library records on computer tape, the Computer Center
becomes extremely important to the Library. Its staff must under-
stand library problems, particularly the reasons for needing certain
print-outs on a regularly scheduled basis. In turn, the Library must
have staff that understands computer problems. Working together,
great things can be achieved.

The following is a list of the computer produced control docu-
ments either in use or to be produced by the end of this school year:
(Those with an asterisk are not yet being delivered. Those without
an asterisk are now being received. The only exception is the SDI
notices, which will be delayed until the next school year.)

Catalog (irregular)

Catalog Supplement (every three months this year, monthly

thereafter)

Cataloging proof sheets (daily)

Authority Lists (author, title, and subject) irregular

Circulation Lists (daily)

1. Listing by call number

2. Listing by borrowers' ID numbers

3. Overdue listing by call number

4. Overdue listing by ID number

5. Listing of hold requests

Reserves (weekly listing of all books on reserve, arranged by

call number)

Reference shelf list (weekly listing, arranged by call number)

Circulation statistical listing (every three months)

Overdues label listing (weekly), for overdues mailing. (All

books are due two weeks from Tuesday.)

Current serials listing (twice weekly)

Serials listing by subject ( irregular)

Serials title listing ( monthly) , until Current Serials Listing

comes out

Serials listing of changes and deletions (monthly)

Serials check-in listings

1. Cards for check- in, computer produced, (one for each
issue of each journal)

110

2. Serials master record (status of data for check-in),
monthly

Serials holdings list (irregular)
*Serials claiming letters
Serials orders lists (annual)
*Serials and bookbinding lists (irregular)
Serials and book orders (irregular, frequent)
* Processing Information List

1. Complete data form (daily) for internal use

2. Brief data form (daily) for public service use

*Budget control listing (daily)

*Invoice, voucher, receiving, and inspection print-outs for
Business Office use.
*SDI notices to faculty (irregular)
*Shelf list ( irregular)
*Official catalog (weekly)

The systems approach in library computerization extends be-
yond the Library. It soon became apparent that both the Registrar
and the Business Office should be in an even larger system of which
the Library would be only a part. The Registrar is now completely
computerized; and the Library made use of this fact to encourage
use of the Social Security Number as an ID number, to justify an ID
card suited to library needs, to provide a broader base for statistical
analysis, and to print out student addresses for overdue notices.
The Business Office programming is just being finished and is being
done with full knowledge of library needs. Overdue fines will be
handled by the Business Office, including sending notices and col-
lecting overdues. Budget listings, invoices, vouchers receiving, and
inspection print-outs will be of mutual interest to the Library and
the Business Office.

Beyond the system within the University, we are thinking more
and more of the system extending to the other state university
libraries, the other state university business offices, and to the state
financial center. The other university libraries now want to take
advantage of what Florida Atlantic University Library has learned
and have approved appointment of an Inter -university Library Com-
mittee to discuss ways and means of doing this. The committee will
be composed of technical processes librarians, who will be most
concerned. There has been discussion of using the telephone com-
pany's "Tel-Pak" system for transmission of computer data, for
facsimile transmission (using Xerox's Scanner- Printer) , for voice
transmission, and for typewriter hook-ups. This may be meaningful
if other campuses decide to go in the same direction. It will be very

Ill

helpful if the new state university campuses in Pensacola and Orlando
adopt Florida Atlantic University Library's system. Members of
its staff are the official consultants for the library planning for those
campuses, and they are intent on extending our system. It is obvious
that such an extension can save money for all concerned. This
library is training librarians from other libraries in the use of the
system and is making information about all aspects of the system
available to everyone.

REFERENCES

1. Perreault, Jean M. "The Computerized Book Catalog at
Florida Atlantic University," College and Research Libraries,
25:185-197. May, 1964; and "Computerized Cataloging: The Com-
puterized Catalog at Florida Atlantic University," Library Resources
& Technical Services, 9:20-34. Winter, 1965.

2. Srygley, Ted F. "Serials Record Instructions for a Com-
puterized Serial System," Library Resources & Technical Services,
8:248-256, Summer 1964.

3. University of California, San Diego, Library. Final Report
Serials Computer Project. ... La Jolla, California, 1964, Appendix
5B.

THE ONULP BIBLIOGRAPHIC CONTROL SYSTEM
An Evaluation

Ritvars Bregzis

The Ontario New Universities Library Project

The Ontario New Universities Library Project (ONULP) was
established in October 1963 to compile basic college library collec-
tions for five new academic institutions in the Province of Ontario.
The Project, which terminates in June 1967, is scheduled to select,
acquire, and process approximately 17,500 titles consisting of 30,000
volumes of in print materials for each of the five institutions: Scar-
borough College, Erindale College (both are new liberal arts colleges
of the University of Toronto) , Brock University, Guelph University,
and Trent University. The progress of the selection and acquisition
program to date indicates that the total number of titles acquired by
the target date will be approximately 25,000, and each institution
will receive approximately 40,000 volumes.

The Project, which is administered by the University of
Toronto Library, is also responsible for providing the participating
institutions with a functioning bibliographic record covering the
materials processed by the Project. The University of Toronto
Library's experience with the maintenance of subsidiary catalogs
located outside of the Central Library was sufficiently persuasive
to indicate that the maintenance of card form catalogs in five dif-
ferent locations would not be successful. Further complications
were to be expected because of additional materials to be acquired
by the individual institutions themselves: out of print materials,
serials, and materials in subject areas which do not fall within the
immediately planned acquisition by the Project. Records of these

Ritvars Bregzis is Assistant Librarian (Technical Services) of the
University of Toronto Library.

112

113

materials had to be compatible with those of the materials received
from the Project. At this point automated compilation of catalogs in
book form appeared to be a suitable method.

Moreover, automation appeared to offer a potential for other
library functions in the two new colleges at Scarborough and Erindale.
This coincided with a vital interest in automated library operations
in the Central Library system of the University of Toronto for which
a large building program, involving the division of the central collec-
tion and a corresponding rearrangement of its bibliographic records,
is being planned. In this complex picture, the ONULP presented an
opportunity to explore the suitability of automated bibliographic con-
trol for some vital library operations.

The design of the ONULP bibliographic control system there-
fore evolved into a project of more general scope than just the com-
pilation of catalogs for five almost identical collections of 25,000
titles each. Three basic requirements emerged clearly: compatibility
with similar work in other institutions, flexibility in accommodating
the entire spectrum of bibliographic data, and exactitude in the
identification of all significant data elements. Within the limitations
of reasonable economy as well as the principal objectives of the
Project, its participants, and the University of Toronto Library, the
ONULP system attempts to meet these three basic requirements.

II
The Structure of the Master Record

The ONULP bibliographic control system is based on a Master
Record which contains bibliographic data in a systematically or-
ganized format. Since all important categories and elements of
bibliographic data in the Master Record are independently and directly
addressable, the entire information in the Master Record constitutes
a store of building blocks which can be arranged in any desired
combination and pattern thus facilitating a number of bibliographic
services: shelf list records, catalogs of various formats and scope,
specialized reading lists, records required for circulation control,
bibliographic information service to the teaching staff, a variety of
statistical data, and so on.

The Master Record consists of a Title Record tape file which
contains complete bibliographic information about the individual
titles recorded, the Name Authority tape file which contains all
names used as entries in the Title Record having alternative forms,
and the Subject Authority tape file which contains all subject terms
used in the Title Record along with the required references to these
terms.

114

The Title Record is organized into bibliographic data cata-
gories, and these are further sub-organized into data elements where
applicable (see Appendix I) . Most of the bibliographic data categories
and elements are of variable length. The total length of the entire
title record has a maximum of 1500 characters. There are no re-
strictions on the length of the individual data categories, as long as
the total length of all used categories combined does not exceed 1500
characters. For example, the category "title" can be of any length,
provided that its length together with the sum of all remaining used
categories does not exceed 1500 characters. In practice this 1500
character limitation can affect only records of titles which are longer
than four full catalog cards. The statistical incidence of such record
length is practically negligible.

The Name Authority and the Subject Authority files are similarly
structured. The authorized form of the name or the subject con-
stitutes a category and so does the group of alternative forms of the
name or subject. The linkage ("see" or "see also" reference) is in-
dicated in the coding of the alternative term. The maximum length
of each name or subject record is 1700 characters. Within this max-
imum length of the record each of the individual categories (autho-
rized term or reference term) can be up to 600 characters long.

For the compilation of the Master Record a specially designed
data sheet is used which reflects in its design the structure of the
Master Record as well as a close likeness to the accustomed bibli-
ographic forms. For each line of the data sheet, one Hollerith card
is keypunched (see Appendix II) . All data are entered in the Master
Record in the form of character codes designed to provide output in
upper and lower case characters modified by diacritical marks where
required (see Appendix III) . For this purpose an IBM 026 keypunch
with a modified A- 2 keyboard is used. All keypunched information
is periodically converted to magnetic tape on an 8 K IBM 1401 which
has been modified to accept and handle the array of 100 characters.

The organization and maintenance of the Master Record tape
file is performed on an IBM 7094 model II. The bibliographical data
are stored on the Master Record tape in call number sequence. All
uses of this bibliographic information derive from the Master Record
which stores data in a format that is largely compatible with any
other structurally systematic format of bibliographic data identifying
all significant data components with a sufficient degree of flexibility.
Fundamental to the compatibility of the data format of the ONULP
Master Record system is the concept of individually addressable units
of bibliographic information which are systematically controlled and
are independent of conventional bibliographic (i.e., catalog) require-
ments. The latter can be derived from a standard system; a stan-
dard system cannot be derived from a specifically oriented fusion of
bibliographic information units. The principal function of the Master

115

Record is not to store catalog data; its function has a far wider scope.
The Master Record stores bibliographic data for any bibliographic
requirement, catalog production being just one of them. For this
reason the Master Record contains more data than necessary for the
compilation of catalogs.

Ill
The Catalog Compilation System

The catalog compilation system produces one of several im-
portant services which can be derived from the Master Record of
bibliographic data. This system compiles three forms of catalogs:
a classified list of all bibliographic units (titles) which is produced
in card form to serve as the official shelf list; and author-title
catalog which includes titles of bibliographic units and names re-
sponsible for the contents of these bibliographic units; and a subject
catalog which includes all topical and name subjects that constitute
the subject matter of the bibliographic units.

In its first major operation the system selects from the Master
Record data necessary for the compilation of the required catalogs
and creates three data files: the shelf list record file, author-title
catalog file, and subject catalog file. Data in each of these three files
are formatted to meet the requirements of shelf list records, author-
title entry records, or subject entry records (see Appendix IV) .
Author-title and Subject files then are sorted in alphabetical sequence
on the first fifty characters of the entry within each entry record.*
This operation concludes the compilation of the catalogs containing
all primary approaches (entries) and those secondary approaches
which have been generated from the data contained in each title rec-
ord. However, alternative approaches (references) still remain to
be included in the catalog files. For this purpose, during the second
major operation of the system the title record tape is matched against
the required authority tape, and references to all entries or terms
found on both tapes are selected and added to the title record tape
(see Appendix V) . Another sorting operation, this time conducted on
a computer compiled sort field, interfiles all references in their
alphabetical positions and completes the detailed alphabetical sub-
arrangement of all entry records.

*"Entry" refers only to the name (personal or corporate) , title,
or subject term under which a work is entered. "Entry record" refers
to the whole record of a work displayed in a catalog.

116

In the design of the ONULP system filing arrangement, em-
phasis was placed on the development of a formula that would accom-
modate filing without manual aid which otherwise would have to be
provided in the form of specific instructions at the input stage. It is
believed that in the file arrangement the principally determining
factors are those which derive from the relative relationships be-
tween entry records rather than from the characteristics inherent
in the entry record. It is also accepted that it is unnecessary to
simulate all exceptional sequence patterns employed in manually
maintained files since many of these exceptional sequence patterns
have been established to overcome difficulties which are not present
in a mechanical file arrangement.

Since the objective of the ONULP filing arrangement is to seek
a filing pattern that would facilitate finding bibliographic data in a
book form catalog, it is expected that subsequent adjustments of the
initially designed pattern will have to be made based on practical
observations of the inadequacies of the initial filing pattern. In de-
signing the ONULP filing system, economy of operation was one of
the factors considered important. The time required for the alpha-
betic sort procedures of the ONULP program now constitutes approx-
imately 15 per cent of the total IBM 7094 computer time which is
required for the compilation of the entire catalog from a relatively
short master file. It was felt that the accommodation of filing pro-
cedures to serve conventional convenience rather than the essential
requirements would not represent an economic use of computer time.
Moreover, it was difficult to accommodate all important requirements
with the same degree of convenience. Thus, for example, it was ac-
cepted that initially it is more important to organize the units of
corporate entry systematically than to arrange names and titles
beginning with these names in the customary accepted order. It is
evident that the initially established filing arrangement will have to
be refined, and that experience will have to indicate the direction
and the priority of the required improvements.

The fourth and final processing operation is the construction
of the catalog page format (see Appendix VI) . During this operation,
the information that constitutes the entry records is set up in line
format as required for the specified column width. In addition, the
total page format is calculated, and the listing tape is arranged
sequentially across the entire width of the page so that the n-th line
in the first column on the page is followed by the n-th line of the
second column which in turn is followed by the n-th line of the third
column. These three lines from three different entry records across
the page constitute a print line which then can be reproduced on the
IBM 1403 model II printer.

The listing of the output tape in page format concludes the
catalog page compilation process. The individual pages are printed

117

on 15" X 22" plain white tabulating paper and are photographically
reduced to 60 per cent of their original size. Offset master is used
for the reproduction of the required number of copies.

IV
Cost Analysis

One of the principles underlying the development of the ONULP
bibliographic control system is that immediate economic considera-
tions are subordinated to the objectives of the system. The ONULP
system is a practical research project, and its research objective is
to develop a bibliographic control system that can meet the entire
range of bibliographic requirements, from catalog production to co-
ordinated bibliographic data exchange. It is believed that only after
these objectives have been attained is a search for the most economic
method of attaining these objectives really possible.

Functionally and economically the ONULP processing work con-
sists of three major operations: cataloging, data processing, and
physical preparation of the processed materials. Since the process
of cataloging— evaluation of bibliographic information and arriving
at cataloging decisions— is beyond the present range of automated
information processing methods, this work is performed by conven-
tional methods. Verification of the required bibliographic data or of
the physical units on hand (searching) , and the interpretation of this
bibliographic information (cataloging) constitutes 32 per cent of the
total staff salary cost of the Project. (Here it must be added that
only a small part of all materials requires original cataloging; most
of the materials acquired for the Project are found cataloged and
recorded in the University of Toronto Library's Official Catalogue)
Physical preparation of the processed materials takes 16 per cent
while data processing operations account for 46.5 per cent of the total
cost of the Project.

This cost analysis is concerned only with the data processing
operations. Under the Project conditions bibliographic searching,
cataloging, and physical preparation are practically unaffected by the
automated bibliographic control system as compared with conventional
catalog compilation methods. This distinction permits comparison of
the economy of the automated catalog compilation system with that of
the manual compilation of customary card catalogs. Both methods
cover the production of the catalog from the point where the evaluated
bibliographic information leaves the cataloger to the display of this
information for public use (see Appendix VII) .

118

The cost of the computer compiled catalog in 150 copies is
more than double that of the manually compiled card catalog in one
copy. It is interesting to note that in the present ONULP work ar-
rangement the largest single factor under both methods is the staff
salary account, 58 per cent and 65.6 per cent respectively. Under
normal operating conditions, the data sheet being filled in by the
cataloger and the data processing operation being carried out with
substantially reduced professional supervision, the staff salary ex-
penditure would be reduced by approximately 50 per cent. The actual
staff time value, nevertheless, would not fall below the staff time
value required for customary catalog compilation methods. It is
interesting to note that reduction of staff participation is not indicated
in a conversion from manual to automated methods.

The other major expenditure account covers rental of data
processing machinery. This is a cost factor which is unique with the
automated method. It is difficult to measure the relation of the value
of this account to the staff salary expenditure since computer rental
costs vary very greatly depending on the institution and the status of
the work. However, the cost of this account is substantially lower
than the cost of staff time.

The third major expenditure account is for reproductive print-
ing. This cost is four times higher for the book form catalog than
for unit cards. It should be noted that this comparison is made on a
first-year basis. The reproductive printing cost of the book form
catalog would progressively increase from year to year.

The principal economy of the automated book form catalog ap-
pears to be in the available multiplicity of copies of the catalog.
Since the compilation of the required information for the catalog rep-
resents a fixed cost not dependent on the number of copies produced,
and since the reproduction of the first copy takes approximately
65-90 per cent of the total cost of the reproductive printing depending
on the number of copies printed, it appears that the book form cata-
log becomes an economically attractive method if distribution in
large numbers is required.

On a one- copy basis the automated ONULP catalog costs twice
the amount of an equivalent customary card catalog. However, since
at least six copies of the catalog are required for the five participants
and the Project, a minimum 250 per cent economy has been obtained.
In fact, the present distribution of the catalog is twenty copies for
each institution, and there is indication that more copies will be re-
quired. Along with this wide distribution of the bibliographic in-
formation, effective control and convenient maintenance have also
been attained.

119
V

Observations

The most significant conclusion which can be drawn from the
experience of the design, programming, and initial operation of the
ONULP bibliographic data control system is that it is the structure of
the data format in the Master Record which has permitted^ a variety
of gradually emerging output requirements to be implemented without
any modification of the originally established data format. The unit
concept of categories and elements of the individual bibliographic
record permits an unlimited variety of combinations of these cate-
gories and elements to be exploited for programmed construction
of data patterns.

This flexibility appears particularly important in view of the
future objectives of bibliographic data processing. The present ex-
perience with the ONULP system indicates possibilities in developing
automated bibliographic data transmission for various forms of
bibliographic co-operation. Centralized or co-operative catalog in-
formation exchange and accessibility of union records on a regional
or wider basis are the most obvious immediate benefits. Preliminary
calculations indicate that such services would be relatively economi-
cal in terms of their attainment.

Bibliographic data transmission and co-ordinated utilization,
however, would require a compatible data format which would be
sufficiently standardized to ensure a certain minimum of scope, def-
inition, and level of detail of bibliographic data to every participant.
Although the ONULP data format is designed with these objectives
in mind, it is not set up to meet such requirements with its present
degree of specificity. Since the ONULP system is a research project
as well as a practical application of limited objectives, the ONULP
data format incorporates only the amount of detail which was found
necessary to test the principle of compatible data format and to meet
the limited requirements of the Project. Experience with the ONULP
system is already being used in the development of a more specific
system for the University of Toronto Library with a view to regional
bibliographic control requirements.

The ONULP system has been rich in lessons pertaining to in-
formation systems development. This work has involved members of
the University of Toronto Library staff as well as members of the
University's Institute of Computer Science, which has been the prin-
cipal advisor, and staff members of the IBM Toronto division which
carried out the practical machine system design.

Regular design meetings were held in which four regular mem-
bers participated: the information retrieval specialist of the Institute

120

of Computer Science, the IBM systems engineer, the head of the
ONULP processing division, and the Assistant Librarian (Technical
Services) of the University of Toronto Library. Many aspects of the
design bearing relationship to the uses of the products of the planned
system, or which were recognized to be of potential interest to the
University of Toronto Library, were referred to the Chief Librarian
and to the Senior Staff of the University Library representing the
interests of the participants of ONULP. In addition, individual staff
members of the University Library were consulted as required and
several surveys conducted by members of the Catalogue Department
staff. In retrospect, it can be observed that an approximate total of
225 man-hours was invested by two librarians and two IBM staff
members in the design meetings alone; and that it would have been
desirable, if time had permitted, to enlarge the participation of the
library staff.

The principal observation to be made pertaining to the design
of the operational system is that the implementation of the system
was several times more complex than originally estimated. There
were two reasons for this severe underestimation of the task: lack
of any significant experience with complex conceptual data process-
ing on the part of the machine system designers, combined with lack
of accurate and detailed knowledge of the theory and the precise
structure and functions of the elements of bibliographic data on the
part of the library personnel. The various theoretical and practical
problems which arose during the design stage of the ONULP system
point out most dramatically how little the bibliographic information
profession (i.e., librarians) knows about the information in its
custody! Extensive, most important, and urgent research and edu-
cational work is required if our profession is expected to take sig-
nificant advantage of the presently available tremendous power of
information processing devices.

The programming phase of the system emphasized this problem
in detail and most convincingly. Again and again it appeared that the
specificity and logic of the external (not to mention the internal) or-
ganization of bibliographic data carefully described by the librarian
stops very much short of the level of specificity of the structural
detail of bibliographic elements required in the programming of the
system. Some of this lack of communication facility between the
librarian and the programmer is due to a tendency to take the obvious
for granted. The explicit absence or the manifest generality of in-
formation causes the programmer to ask the librarian for more
information or for more detailed information. The "obviously" im-
plied information, however, creates results which frequently require
considerable time and ingenuity to diagnose and correct.

A rather simple but significant example can be noted in pro-
gramming the disregarding of the initial article in filing title entries

121

which are automatically generated by the computer program. The
programmer was given an explanation of the structure of the entry
system and a review of the various forms the title entry may take.
As it turned out, this review was not detailed enough. In terms of
the system, apart from the title main entry, the title added entry, and
series title added entry, there may also be an alternative or con-
structed title added entry, analytical title main entry and analytical
title added entry. It took some time to discover that in the printed
catalog those title added entries, for which the initial article had not
been disregarded in filing, were actually alternative title added en-
tries. For the librarian, it had been "understood and obvious." The
programmer, however, naturally took the instructions literally.

This raises the question of whether for certain kinds of pro-
gramming the programmer does not have to be a librarian. I believe
not, because fundamentally the problem is not one of communication
between the librarian and the programmer. It is the lack of knowledge
of the systematic and specific organization of information and the
lack of a logical and standardized form for recording of this in-
formation available to the librarian. In the cited case, a librarian
programmer would not have fared much better because he himself
likely would not have been able to analyze and express in logical
and sufficiently detailed form the "obvious" differentiation between
the various types of title added entries. Numerous similar incidents
bear out the need for much more exact, accurate, and specific def-
inition of the bibliographic problem rather than increased involvement
by the librarian in the actual preparation of the detailed working
drawings of the information processing work at which the draftsman
will be always more efficient than the architect.

The ONULP program is written in Cobol for the IBM 7094
computer, the programming work having been done on contract by
the IBM Toronto Data Centre. The choice of the programming lan-
guage represents a compromise between the resulting processing
time and cost on one hand and the ease of writing, change, and adap-
tation of the program on the other. For processing economy con-
siderations, the choice should have been a lower order language
requiring a minimum of compiler instructions for translating it into
machine code language. Cobol was selected for three principal
considerations.

First, the consideration of convertibility of the program for
use on other computers was an important one. The Project involved
compiling bibliographic data and catalogs for five individual insti-
tutions some of which have already expressed interest in continuing
and developing further the automated method of bibliographic control.
It could not be expected that all of these institutions would have
access to the computer configuration employed for the ONULP sys-
tem. With a non- compatible language, practically the whole pro-
gramming effort would have to be repeated by each of the interested

122

institutions. It is estimated that conversion of the present ONULP
Cobol program would require only a fraction of the total programming
effort in order to convert it for use on similar installations. Thus it
is estimated that the conversion to IBM 7040 would require only ap-
proximately 15 per cent of the programming effort that was required
to write the program of the ONULP system.

Convenience and ease of programming was another considera-
tion. It was expected that numerous subsequent modifications in detail
of the ONULP system would be required, and that these would have to
be implemented by programmers on the library staff, perhaps with
limited technical background and experience in complex machine-
oriented methods. A high order language with maximum similarity
to customary forms of expression was therefore considered a valid
choice for the benefit of both the ONULP and the participating insti-
tions individually.

The third consideration behind the choice of Cobol was an early
observation that the nature of bibliographic information combined with
the requirements of the design of the ONULP system resulted in a
degree of complexity of information structure which would require
an extreme analytical consistency from the programmer if a lower
order language would be chosen. Thus it appeared that, for instance,
Fortran would be impractical.

The very specific and demanding nature of bibliographic data
soon revealed the unexpected complexity of the programming and
machine processing time requirements. The ONULP Cobol program
is lengthy in terms of processing time. Masses of machine instruc-
tions have to be generated for the already copious Cobol instructions
that constitute the program. This characterizes Cobol in general.
In addition, the application of Cobol to bibliographic data (i.e., con-
ceptual or word information as compared to conventional business
information) showed a decrease of the actual IBM 7094 processing
rate below the speed of tape moment. Since the initial calculations
had been based on tape speed which normally is the limiting speed
factor for IBM 7094 operation, the ONULP processing time was found
to be considerably higher than estimated. The IBM 7094 processing
power is really put to use by the ONULP system. The severity of the
test was also indicated by the fact that during the one year of pro-
gramming work, several Cobol compiler errors were detected.

A good part of the processing complexity is accountable to the
programming language. However, a considerable factor is the mas-
siveness of bibliographic data and the character-by-character mode of
operation upon this mass of data under very complex patterns of data
transformation. Present experience appears to indicate that process-
ing information in word form poses problems to the machinery and to
the methods of automated data processing which until recently were
developed primarily for quantitative data processing. The promise

123

of the latent potential of information processing methods requires
the testing of these methods under the full load of automated informa-
tion transformation and communication requirements. The ONULP
system has attempted to face this problem squarely without com-
promise. With the question and answer console as the next immediate
step in automated bibliographic control, it is obvious that the pro-
cessor and the processing methods will be required to handle the full
complexity of bibliographic data without any appreciable human as-
sistance. It is possible that computers employing associative mem-
ories controlled by list processing techniques can offer some solutions
to the time economy problem which is so significantly underscored
by the ONULP Cobol programme. The digital memory techniques
seem to indicate a dilemma between the processing time economy
and the human oriented form of communication.

The operation of the bibliographic data processing system in
place of the conventional card catalogue production system has been
successful. The preparation of information for keypunching the input
record is simple, and only a few days are required to acquaint the
catalogers and other staff with the method of recording information
on the data sheet. The data lay-out on the keypunched Hollerith card
is equally explicit, permitting relatively easy sight verification of the
keypunched information. Sight verification, however, is a tedious
process and is due to be replaced by machine verification. The ex-
ceptionally low rate of input errors has been the result of competent
staff and a simple input procedure requiring only a minimum of in-
structions to be remembered or consulted for the preparation of
input.

As most of the ONULP materials are copies of the University
of Toronto Library holdings, very little cataloging is involved, and
data sheets are prepared by library assistants from the Official
Catalogue copy. For processing of materials for a collection that is
currently being cataloged, the data sheet would be prepared by the
cataloger in lieu of his manuscript copy. This would eliminate the
data sheet preparation as a separate step (see Appendix VII) . When
in addition the professional administration of the data processing
operation is reduced to normal operating requirements, the total
salary cost of the preparation of the input record for automated
processing equals approximately the preparation of cards for a card
catalog. The cost of data processing equipment rental and repro-
ductive printing is approximately equivalent to that of data processing
staff time cost, and it is over and above the direct cost of the card
catalog. However, the total cost appears modest when viewed in
terms of multiple book form catalogs.

A few profitable observations can be made about the format
and serviceability of the computer-printed book form catalog. The
availability of upper and lower case characters and most of the

124

diacritical marks necessary for the display of bibliographic informa-
tion in most languages which can be expressed in Latin characters
contributes substantially to the accuracy, necessary differentiation,
and style of bibliographic information display. However, it is realized
that the 120 character print chain which produces this result is only
a temporary compromise between what is presently available and
what is necessary. Apart from at least three type fonts for the Latin
alphabet, the research library uses require also the Cyrillic, Arabic,
and Greek alphabets which can be used intermixed with Latin alphabet
text. Particularly essential is the Cyrillic alphabet since every 15th
to 20th new title acquired by large research institutions is in a lan-
guage using the Cyrillic alphabet.

The readability of the page in the book catalog also depends
largely on the added emphasis that is attached to selected informa-
tion of key importance. Various display techniques can help to en-
hance readability. However, the positioning and grouping techniques
would be helped considerably by additional variation in character size
and density. Additional type fonts are called for. The requirement
for a variety of type fonts and alphabets indicates perhaps that a
technique other than impact printing has to provide the solution.

The problem of the printing method is further complicated by
limitations of printing speed. Six hundred or eleven hundred lines
per minute printing speeds become insufficient for printing the large
quantities of bibliographic data which are required for a catalog of a
medium to large research library where constant cumulation and
reprinting of data becomes a very voluminous chore of the data
processing system. Until the techniques and the required philosophi-
cal and logical bases for bibliographic data processing and trans-
mission are developed to the point where on-line consoles can take
over the major part of the present catalog display functions, printing
of catalogs will remain one of the principal problems.

It should also be remembered that the variety of characters
available for information display has to be paid for by a sacrifice of
printing speed. The 120 character chain has only two sets of 100
characters instead of the customary five sets of 48 characters on the
standard chain. t

tThere are 120 character positions on the chain, but only 100
positions are used for the following groups of characters: 26 alpha-
betical characters in lower case, 26 alphabetical characters in upper
case, 10 numerical characters, and 38 special characters. The cost
of this print chain to the University of Toronto Library was $2,007.
The modifications of the IBM 1401 necessary to accommodate this
chain cost $2, 832 in additional rental annually. Other minor modifica-
tions were required for the IBM 026 in order to facilitate the prep-
aration of input suitable for the printing with the 120 character chain.

125

Since the rotation speed of the chain cannot be increased, the
printing speed is reduced to two-fifths of the maximum 600 lines per
minute. Diacritical marks are imprinted by overprint while line
space movement is suppressed. The presence of even one diacritical
mark in a print line cuts the printing time therefore for each such
line approximately in half. Depending on the quantity and distribution
of diacritical symbols, the effective printing speed on a 600 lines per
minute printer ranges approximately between 110 and 225 lines per
minute, or 200 to 400 lines per minute on a 110 lines per minute
printer. It is readily seen that it does not require a particularly
large catalog to be periodically cumulated in order to reach a satura-
tion point in printing economy.

The computer produced master page is only the beginning of the
printing process. The offset press takes over the reproduction of
copies. The cumulation of the ONULP author-title and subject cata-
logs during the second year of operation covering 10,000 titles will
amount to a total of 4,400 pages to be printed. For 150 copies of the
catalog, this represents approximately $10,000 offset printing cost.
It has, however, to be borne in mind that the offset printing of the
first copy costs approximately $7,500; the number of copies produced
therefore adds very little to the printing cost, and the catalog dis-
tribution can be liberal.

A thumbnail estimate, however, can readily show that periodic
cumulated reprinting of the entire catalog of a medium size research
library would keep a university press busy almost on an around-the-
year basis. For these cost reasons as well as reasons of cumulation
effect on consultation ease and reasons of up-to-dateness, it ap-
pears that the computer produced book form catalog in itself cannot
be the ideal solution to the bibliographic data display problem in large
research libraries, although on a short term basis and for limited
data quantities, the cumulated book form catalog has many features
to recommend it.

For the medium size and large research library, it appears
that direct access to digitally stored bibliographic data can be eco-
nomically competitive with computer printout catalogs, as well as
giving instantaneous and direct answer to the bibliographic request.
The concept of the library catalogs of the 1970's therefore may well
be a combination of printed catalogs for the consolidated portions of
the collection and keyboard-and-screen terminals capable of display-
ing the bibliographic information entered into the system before this
instant.

The serviceability of the information displayed in the catalog
can be further improved by an optimized page layout, variation in
the degree of density, cumulation pattern, and system of alphabetic
arrangement (filing) . Cumulation pattern and alphabetical arrange-
ment are particularly important. The cumulation pattern chosen for
the ONULP catalog production was determined from the correlation of

126

the printing cost and the convenience of consultation of the catalog.
Appendix VIII shows the effect of the various cumulation patterns.

The system of alphabetical arrangement involves the com-
bination of the requirements for convenient consultation of the file
and the economy of the sequencing operation. There appears to be
a choice of three basic systems which can implement the alphabetical
arrangement. In the first place the filing arrangement can be manual-
ly aided by inserting in each bibliographic record symbols which
place this record in a relative position with other records in an
anticipated context. Although the easiest to operate on a limited
scale and accurate in a correctly anticipated context, this method
cannot function effectively in a context which initially cannot be cor-
rectly assessed. Moreover, on a large scale this method tends to
constrict the versatility of the potentially required file arrangements;
it also tends to interfere with retrieval techniques designed to bring
out key words and key phrases from entries and titles.

Another method of file arrangement is to establish an ideal
total file of all potential entries which then can be coded sequentially.
The entries which are to be included in the file are similarly coded
and their relative position established by matching of the two files.
This concept assumes a large enough pattern which would always have
gaps between any two actual entries. Given the ideal pattern and
only records with alphabetic patterns defined at the input stage, such
a system could be a convenient one to operate. However, the practi-
cal limitations are set by the indefinite variety of potential informa-
tion, extremely dense unforseeable clustering of information, and,
most important, by the entry records automatically generated and
therefore lacking any other identification (i.e. a relation code) than
their inherent character value.

The third method is based on an entirely programmed and
manually unaided sequencing by the values inherent in the characters
constituting this information, in combination with the coded charac-
teristics of the fundamental bibliographic unit. For reasons dis-
cussed above the ONULP system chose the third method. The
inadequacies seen in the produced catalogs indicate that subsequent
adjustments will be needed before a general and largely automated
filing method is developed.

One of the most important and potentially far reaching obser-
vations derived from the design and initial operation of the ONULP
system is the inability of the customary bibliographic systems to
provide a sufficiently systematic basis for a significant utilization
of the power of the data processing machinery and techniques. Both
the customary system of entry and, in particularly, the system of
subject terminology cannot facilitate effective retrieval of biblio-
graphic information on the basis of authorship and subject value.
The corporate and other formal entries which the ONULP system

127

has accepted with the customary cataloging rules cannot facilitate
approaches to the significant elements embedded in these entries.
Similarly the Library of Congress subject headings which are used
for the organization of the ONULP subject catalog are not suited for an
an analytical or integrated approach to subject information. The ca-
pacity of the data processing system in this case is largely invalidated
by the ineffectiveness of the subject heading system to integrate and
analyze concept terminology automatically. Instead of a powerful
analytical tool, this subject catalog can be only an alphabetically or-
ganized array of fragmented information.

Along with the urgent need for an agreement on a compatible
data format which can facilitate effective communication of biblio-
graphic information, a quest for a theory of systematic organization
of concept terminology is essential if automation of bibliographic
information processing is to yield the tremendous potential of which
it is capable.

In conclusion, it can be noted that the ONULP bibliographic con-
trol system was designed to meet a limited practical requirement
and at the same time to explore problems of more general application.
Many of these problems cannot be answered without a considerable
practical exploration. Several others require fundamental research
in the theoretical bases of bibliographic information. These had to
be compensated by expedients of compromise nature.

The design and implementation of the system was a complex
team work requiring co-operation between librarians and data pro-
cessing experts. Numerous factors have contributed to the success
of this work. However, although the success largely depends on the
characteristics and capacity of the equipment, on tested methods of
information processing, and on the expert judgment and decisions of
systems engineers, the librarian cannot evade the final responsibility.
Like an architect who alone has to know intimately and who can con-
trol the whole range of expert engineering functions and drafting
techniques, the librarian alone is in the position to know what the
general design has to incorporate and what attainment constitutes a
fair price for the engineering of the desired effect. In order to carry
this responsibility, the librarian like the architect has to know in
detail what constitutes good design; he has to know what constitutes
the engineering work and what effect he can expect for a given in-
vestment in methods and materials. And like the architect he does
not have to be the engineer and even less the draftsman.

128

Appendix I

ONULP
Data Categories and Elements

The Title Record |

1. Call number of the title. This category is further sub-
divided in the following elements:

a. main class

b. sub-class

c. class number

d. first Cutter number

e. second Cutter number

f. date.

2. Form of publication.

3. Brief edition statement.

4. Edition code.

5. Number of volumes.

6. Size. This category consists of two elements:

a. height of the book

b. thickness of the book.

7. Origin of the publication. This category is further sub-
divided in four elements:

a. country of origin

b. city of origin

c. language of the publication

d. date of the publication.

8. Any special aspect other than those enumerated.

9. Paging and illustration statement. The two elements of this
category can be separated if desired.

10. Principal author (main entry).

11. Conventional title.

12. Title.

13. Full length edition statement when different from 3, above.

14. Imprint.

15. Series note.

16. Other notes (except Contents).

17. Secondary bibliographic approaches ( Tracings) . This
category is subdivided in five elements as follows:

129

a. subject term. This element is further subdivided in the
following forms:

i. name as subject term
ii. topical subject term
iii. descriptor term

b. added entry (other than title and series)

c. title added entry

d. alternative title added entry

e. series added entry

Up to ten approaches of each group can be assigned to one title
record.

18. Contents note. This category can be used in a systematic
organization mode indicating the numbers or volumes of the
contents. If so identified, the individual parts of contents
function as analytical approaches to the larger work.

19. Copy and location. This category consists of two elements:

a. copy number of the title

b. location of the copy.

Appendix II

ONULP
Input Card

444444444

5555555555555555555555555555555 5TT5T"5"5T"5 555555555555555555555 5T5V5 j 'j 5 5 5!5 55555555

130

Appendix III

ONULP
Table of Special Character Card Codes

LOWER case characters

UPPER case characters

Card Code

Character

Card Code combination

Character

0 (zero)

follow key -board <

058 - A

A (capital)

058 - Z

Z (captial)

12 - 3-8

. (period)

058 - 6

0-3-8

11 - 5-8

058 - 7

b (musical flat)

11 - 6-8

058 - 8

I ^cript «L")

12-0

11 - 7-8

' (apostrophe)

12 - 4-8

(

058-11-7-8

* (quotes)

12 - 5-8

)

058 - 11 - 0

& (ampersand)

12 - 6-8

Diacritical marks

12 - 7-8

058- 1 VVV*

case)

0-1

058 - 2

058 - 3

- (hyphen)

058 - 4

3-8

058 - 5

- (umlaut)

0-7-8

058 - 5-8

o (angstrom)

11 - 4-8

058 - 6-8

~ (tilde)

0-6-8

4£ (musical sharp)

058 - 12

•J (cedilla)

0-4-8

058 - 0-1

Diacritical marks A

0 - 2-8

(Lower

case) |

6-8

11 - 3-8

(underscore)

5-6

4-8

Note that diacritical marks are keypunched before the character they
modify; e.g., c 058-6-8 followed by c

131

Appendix IV

ONULP
Catalog Entry Record Formats

JA83. A6 1951
Allen, John Jilliam, 1665-

slxteenth century. London, Methuen

xxii, 527 p.

"The Bibliographical notes, pages
517-523, have been revised and brought
up to date by Mr. J.W. Cough."

1. Political science - History 2.

State , The I. Title

BR3C

ERIN

GLPH

SCAR

THEM

1. Author-Title Catalog listing
-* main entry record

JL195. C6
The alignment of political groups in

Canada, 1641-1867
Cornell, Paul G.

The alignment of political groups In
Canada, 1841-1867 1962 x, 119 p.

BROC

ERIN

GLPH

SCAR

TREN

All

: Quint!
jthor

PQSS01. L8 A19 1962
Joaquln, 1873-1944, jt.

Alvarez Qulntero, Serafln, 1871-1933
El ijenio alegre 1962 525 p.

BRDC ERIN GLPH SCAR TREN

PQ1356. A63 1962
Allen, Louis, ed.

Beaumarchals, Pierre Augustln Caron de,
17J2-1799

Le barbler de Seville 1962 111, 90
P-

E77. 075 1960

Philadelphia. Transactions, ne»
ser., v. 47, pt. 2
Driver, Harold Edson, 1907-

Indlans 1960 165-456 p.

BROC ERIN GLPH SCAR TREH

title added entry record

name added entry record
(joint author)

name added entry record
(editor)

series added entry record

132

Alberta - Politics and government

JL500. A9 17
Irving, Johr» Allan, 1903-

Alberta. 1360 «i, 369 p.

BRDC ERIN GLPH SCAR TREN

2. Subject Catalog listing
subject entry records

JL500. A3 »3 1362

KIV, 258 a.

BRDC ERIN GLPH SCAR TREY

JL500. A3 M36

"il lory, Janes Rutsel 1

in :anada. 1954 xii, 204 a.
BRDC ERIN GLPH SCAR

3. Shelf list record cards

Thomas, L.C.

The Libe
xii, 230 o.

JL500. A3 T5
•>erta 1353

BR3C ERIN GLPH SCAR TRE«(

PM6086. G4J card 1

Genest , Erai le

Dictionnaire des citations ;
dictionnaire des phrases, vers et mots
celebres employes dans le langage
courant avec precision de 1'origine.
Paris, F. Nathan r!962]

423 p.

Cover title: Dictionnaire des
citations francaises.

SEE NEXT CARD

PM6088. G43 card 2

1. Quotations , French 2. Quotations

I. Titls II. Title: Dictionnaire
des citations francaisss

GLPH 1

133

New Titles
Changes

Appendix V

ONULP
Processing Chart

The Master Record

f Names
References
History notes

sort by call number

Subjects
References
Scope notes

card to tape conversion

file maintenance

134

Catalog Compilation

entry record construction

135

Catalog Listing Tapes

.Shelfi-ltst.
_ 3*5 _
_oards_

Subject
Authority
list page

136

Appendix VI

ONULP
Page Format

PQ1B94. M66
Hoore, VIII Grsyburn

thought

Th t d r u.i •"""' "5 "" ""

PQ1842. H3S
Hall, H. Colon

HT609 C4 1961

gPS14. CS4

1961 .11, 244 p.

Uutllnei of bloch»l>try 1963 .111,
391 p.

Currl., P.tfr

QP601. N35 1964

""Jl'o!""' *'lf It , f

OD7. PS

Barber, Will lap Henry
VoM«lr«: Ondlde. ' " 1 62 p.

s,,, , ,h ,„ ,h "T1-"7

CK3515. H4 1962
Stylei of «ddr«99

Falrlle, Hilton
64 p.

BROC ERIN GLPH SCAR TREN

HC106. R35

P92355. 384
no. 7)

64 f.

PQ2246. «3 F3
r.lrl l«, nl lion

BJ41. "3

no. 7
4162. SBS 19S3

BROC ERIN GLPH SCAR TREK

2BO p.

192 p. (• M.ntor book)
Till.

H«ggl«, Donald R.
BROC ERIN GLPH SCAR TREK

rc», 1950.
v. In 1

PR3S62. 58
1962.

68 p.

""ir'^r.'po^.-ri-Mjc.,

D419. C4 1961

Piul Cl*»<l«l. 1957 111 p.

PNSll. 53

• d. ] 1961 372 f.

IfeMfkl

Murloch, Irlj

211 p.

D375. 1)7 SB 1962

«, 114, [2] p.

The study of «n

!B»-lo81 1. Title

o
o

137

•8

8 S"^

. ^ o

,IH o

i-T

/^^s**^

o m

CM O

CM CO

1 1 1 1 1 1

•a

^^^

^— ^

--— *. ^^

s^ s^

•r- (

r-H J2

CO lO

™ O

0
0

^ :£

GO O

C r3

<«^ <•**

0) g
bD §

s s

bJ9

1 1

°l °.

CM CO

-3

o g

-M O
CQ 53

^ , x^^i

»>x

•H

o o o m o

m in

rt
0

3 <u

o in t* t* co

co co oo T~i TJ<

t- CM

m IH

^ &

i-i

•gH

s °

« ^

CO ^

i — i

o
=1

O5 O5 0

«S Ji ^

«1

fl ctf
g«
O

o
o

ONULP
actual ope

m o m
c- m co

". °l °.

•* CM CO

^»N

o m o o o m o

m in

o t> o m i> t> co

C- CM

o
$J

oo m co co oo IH •*

in IH

CM rH T~) r~( ^^

•p

« el

o o (5

•+J

If iii]

Q) >• | «M «M CO

^ CQ

HOI CO

a JH be be ^ ^ ^

H fl

h §

® "S 2 3 ^ ^ S

5 s

CQ
O

j 1 1 1 1 1 -=

1 «

— S

_j ^j ^1. CX ^( p_|

1 S

•4^ jj >* >N flj CD " TTTTIT1

>v m

nj e5 CO fl) -4-> -4-> **l'**l

Q Q W W fl fl

XSOO AHV^VS d»IVJ,S

138

i — i
b? ^ 6^ 5S -r> co

(M O5 C<l OO &S •

r-H

r^ £3 ^. 2 « l<0'

o 13 °°>>
o+2 a

o m m feR

O <M rH 0
t» t~ O5 O

£0 §

r-T r-T Cl-

a 'H ^

S^S^*-^ s-~"~~-^ ^~~^*~*.

ooomoommin

Co O

oinmt-int— Tfmo

oo ^

o>oooocotcoT^'^f<T*'~J%

<~^BI

l-T r-l t-

[—71

1 1

Tt*

1 ^

1 — '

'S'

0 °

T3 *2

01 g

-a

-t-> O
CO "^

.f-, "co

in o

O «D

o o o

(NO O

3 &

in i i i i i i i cp^
t-

t- in o

co" i ^ i

^fl

i — i
oo

•g

ft 1

€

H-5 ^

P r-j

§ |

o o o

m co
o o

(MO O

t> m o

m i i i i i i i L-"

•* »»

co i ^^ i

s™-'

c a i

co -2 -*•-

st elements

"ti fH r? W bD 3 2

1 _f 1 1 I I 1 !

1 i i I 1 ill %

§H^rHr-(f-lHJ §^3

C8 CO 0) H ra CO

•S'S'M00'S'^2CO

"1 rt ™ r*>>>x) a S ^
iJ o o r-< rHr2<.5.;;c3

bC
0

rH "*•* -^

° o o 'S

o
0

•8 % g-a-S'airtgS

5 sja;nd ;uijcd™

Jrs! 8 § "S "S £3

•s^^mgcOcOOO

| -uioo ;asjjo

WSSCQWOOOH

XSOO AHVTVS .MVvLS

XSOO ONISS30OHd

r-l

I— I

o §

h CM

p, QC^-s

in*

£j O CO

O ^ OH

r^1 CQ «4H

o co o

£o §

11 *

Q) ^ OJ2
•Q, Q) .g

fl "S m

C i— i CO

s°

0
10

ill

r-T

53 ei_i

I— 1

* ° 0

•JJ CO •*-<
0 « TJ

1 1

^ 2 *

"co"

r— i

o b£ pj

S 'S ®

2 .2

g-g B1

QJ *3

1— 1

m CO

0) 'S

fac §

CQ «0 ]tQ

"^ r— 1

CO »^

•1-1 O *O

Q} C3

-s

4-> O

CQ £j

O
IO

to co -S

Bj CQ OJO

r^ W

_ CO O

oo"

CM"

oo"

CQ 0 I-H
fl> S «

53 O "S

£ ^ 0

C1! Q} ^^

jq-

si i

(Computer compiled)

ONULP
• the actual operation

CO
iH

co
i — i

o
ob"

O
O
rH
1 1

IO
CM

io"

for the processing o
ir of additional titles

:essing work of the O

m w v

t-< & o

™ S £

4) 3

B i g

** "8 t|

*•§ a

CO m O

s ^

•3

O • CO

"CQ

£3 CO •**

•4^

d O5 m

_£2 ^ S

13
1
EH

—

o 'S o
0-3 _

53 rQ rH

139

O EH

rt|^^ 0°^

•a-S o-^5§

-a <«£.8.S n

2 ™

3 I

•iH S

140

Appendix

Cumulation Effect & Printing Cost

0 Not cumulated

Third- yearly -
to date

Quarterly - to date N^

Third-yearly - current
Quarterly - current

("Bi-monthly & semi-annually - current
1 Quarterly & semi-annually - current

(•) Bi-monthly - to date

" ~" O ' Monthly - to date

10 20 30 40 50 60 70 80 90 100

Total number of cumulation units to be printed during the year

COMPUTER INDEXING OF THE PRESIDENTIAL PAPERS
IN THE LIBRARY OF CONGRESS

George R. Perreault

Preface

"The index to the Abraham Lincoln Papers is a direct result
of Public Law 84-147, dated August 16, 1957, the object of which is
to inspire informed patriotism, to provide greater security for the
original manuscripts, and to make the Lincoln and other Presidential
Papers more accessible and useful to scholars and other interested
persons. The law authorizes and directs the Librarian of Congress
to arrange, microfilm, and index the papers of the 23 Presidents whose
manuscripts are in the library. An appropriation to carry out the
provisions of the law was approved on July 31, 1958, and operations
began on August 25."

The above paragraph was taken from the index to the Abraham
Lincoln Papers and best describes the purpose and goal of the Pres-
idential Papers Program. This paper explains how the index was
prepared on unit record equipment, the problems encountered, and the
solution or partial solution of those problems with the use of a small
scale computer.

Solution with Unit Record Equipment

Merely microfilming the manuscripts of the various presidents
would not increase their usefulness unless there was some tool that
could be used to locate the manuscripts on film. The responsible
staff members at the Library of Congress decided in 1958 that this
tool would be an index by writer or recipient of the manuscript and
in addition would contain the following information:

George R. Perreault is in the Data Processing Office, Administrative
Department, Library of Congress.

141

142

1. date of document

2. series number

3. number of pages

4. additional pertinent information.

With the number of entries varying from 1,089 in the Taylor
index to about 530,000 in the Taft index, it was a natural decision to
use unit record equipment then available in the Library of Congress
to assist in the tremendous alphabetic sorting task. At that time the
Library had IBM model 407 tabulators, model 83 sorters, model 85
collators, and model 519 reproducers. Five years later it converted
to a small scale computer operation.

Librarians soon learned that the punched card has the inherent
disadvantage of containing only eighty columns of information with
very rigid fields of information within these 80 characters. These
restrictions caused no problems for items with fixed lengths such as
president number, date of document, or case file number, but prob-
lems were anticipated with only 36 of the 80 columns remaining for
the name of the writer and recipient of the manuscript. Rather than
attempting to use a rigid limitation of 19 columns for the writer and
17 columns for the recipient, it was felt that one variable length field
of 36 columns would offer greater flexibility for indexing. The name
of the writer or recipient would be terminated by the words "TO,"
*FR" (From) or "VS" (VERSUS). Figure 1 shows the card format.
This successfully allowed the indexer to use more columns for the
writer or recipient on the merits of each individual entry. As an
example, the following entry would be impossible to place in two rigid
fields of 18 and 17 columns: "Birmingham Peace Jubilee to WMK."
The name of the writer here requires 24 columns.

The use of one variable length field solved one problem but
created another. How could the mechanical sorter tell where one
field ended and the other one started ? It just could not. Sorting the
entire field would place all BFR" entries before all *TO" entries for
the same writer or recipient rather than chronologically within each
writer or recipient. It therefore became necessary to select an
arbitrary number of columns (one that would not extend up to the TO,
FR, or VS in the majority of entries) to sort mechanically, followed
by a hand sort for all conditions that were not properly alphabetized.

Filing Problems

In addition to the limitations of the mechanical sorter to handle
a variable length field, there were many filing rules that could not be
performed because they varied from character for character sorting

143

and the collating sequence of the sorter. One may think he sorts
character for character until he uses a mechanical sorter. In the
"Date of Document" field, for instance, the following entries were
used. The list indicates the desired sequence:

Description Entry

1. Cross Reference See Also

2. Month, year, and day 1912 MY 12

3. Month and year 1912 MY

4. Compound Month 1912 MY-JE

5. Year Only 1912

6. Compound year 1912-1913

7. Estimated Date ND*1912-1913

8. No date ND

Character for character sorting would place them in the follow-
ing sequence:

1. Year Only 1912

2. Year and month 1912 MY

3. Year, month, and day 1912 MY 12

4. Compound month 1912 MY-JE

5. Compound year 1912-1913

6. No date ND

7. Estimated Date ND*1912-1913

8. Cross reference SEE ALSO

All of these deviations from character for character sorting in
the "Date of Document* field had to be hand filed, but greater prob-
lems arose in the writer or recipient portion of the card. Typical
examples of further deviations would be names with a prefix such as
Mclntyre, McCooley, O'Bryan, O'Leary, and words that should be
ignored, such as "MRS" and «ETAL."

Sorting and Hand Filing with Unit Record Equipment

To hand file the cards properly, it was first necessary to list
the cards on the IBM 407 tabulator. The listing was then checked for
discrepancies. The card representing the discrepancy had to be
found in the card deck and then hand filed in its proper sequence.
The card decks were then listed again to verify that all hand filing
was done properly.

144

File Maintenance with Unit Record Equipment

The editing for spelling, consistency in indexing, etc., was also
done at this time. To change an entry, it was necessary for the
editors to locate the card representing the entry on the list, write
the change on the card, and submit the card for keypunching. When
the cards were returned, they had to be compared for accuracy of
punching, then hand filed in their proper sequence.

Certainly many bibliographic projects on unit record equipment
must follow a close parallel to the foregoing system. The main
problems are hand filing and file maintenance, the latter meaning the
ability to make additions, deletions, and changes.

Solution with a Computer

On January 17, 1964, a small scale IBM computer was put into
operation at the Library of Congress. It consisted of 8,000 positions
of memory, 4 magnetic tape units, a 1403 printer, and 1402 card
read punch. Although its main purpose was to serve the Library's
business and fiscal needs, we felt certain that it could be used to
great advantage in solving some of the problems in preparing the
indexes to the Presidential Papers. It would have been ideal if an
entirely new system could have been developed at this time, but un-
fortunately a great deal of work had been done on the indexes to vari-
ous presidents long before the computer arrived.

The computer system decided on has four main phases. Each
phase on the accompanying flow chart is on a separate page. The
following narrative description is written in this sequence and carries
the step number of the flow chart to allow a cross reference between
narrative and flow chart.

Phase 1— Create Index on Magnetic Tape (see Appendix)

Strangely enough, the problem of sorting a variable length
field led to the system technique that was used to solve some of the
hand filing rules. A sorting operation on a computer is much like
that of a mechanical sorter in that it will only sort character for
character, and the length of the field to be sorted must be clearly
defined.

( l) It was evident that to solve the problem of a variable
length field it would be necessary to establish a fixed length field on

145

tape. This was done by searching the "Writer or Recipient" portion
of the card as it was being transferred to tape for "TO," "FR," or
"VS." When this limitation was found, that portion up to but not in-
cluding "TO," "FR," or "VS," was placed in a fixed length field of
44 characters that we called the "Sort Key." All unused portions of
the sort key were filled with blanks. The "Date of Document" field
was also searched, and appropriate substitutions of values (see
below) were made that would give the desired arrangement according
to the collating sequence of the IBM 1401 computer which is special
character first, followed by letters, followed by numbers. This sub-
stitution of the date field was placed in the rightmost portion of the
sort key leaving 36 positions for the writer or recipient. Each entry
on tape would now contain the card image and the 44 character sort
key establishing a 124 character fixed length tape record. Figure 2
is an actual "tape print out* of a few tape records.

Listed below are the substitutions made in the "Date of Docu-
ment" field. These substitutions can also be seen in Figure 2 by
comparing the entry to the date portion of the sort key:

Entry in "Date of Document" Field Substitution in Sort Key

SEE 00000001

SEE ALSO 00000002

1916 MY 6 19160506

1916 MY 19160588

1916MY-JE 19160599

1916 19168888

1916-17 19169988

ND*1916-1917 19169999

ND 99999999

It was at this point in the process that it was discovered some
of the hand filing rules could be solved by manipulating the sort key
in much the same manner as the "Date of Document" field had been,
either by removing words and characters or by substituting values
that would produce the desired sequence. For instance, it was a
relatively simple matter to substitute "MAC" for all "MC" entries,
because "MC" should be sorted as "MAC." Names like O'Leary and
O' Bryan should sort as if there were no apostrophe. The apostrophe
was removed, and all characters to the right of the apostrophe were
shifted one position to the left.

An entry such as "Walen Mrs J A* was a little more complex
but, nevertheless, was handled without too much difficulty. These
entries should be filed as the last entries for Walen J A and chrono-
logically within themselves. Removing "MRS" from the sort key
would make the entry sort with "Walen J A" and by placing a "9" in

146

the rightmost position of the sort key would give the entry a greater
value than any other entry for "Walen J A" thereby forcing it to fall
after all "Walen J A* entries. Figure 2 line 9 is a good illustration
of this type of data manipulation.

The original computer program also successfully filed all
numbers from 1 to 100 as if they were spelled out. This was accom-
plished by finding a number in the entry and locating its spelled out
equivalent in a table located in memory. The word was then sub-
stituted for the number in the sort key which, of course, would force
the entry to sort as if it were spelled out. It was later decided that
this filing rule would not be followed. In its place we may substitute
spelled out words for key abbreviations. These key abbreviations
will vary from index to index so the computer program would be
developed with the flexibility to change the abbreviations used for
each index. The number of abbreviations that can be handled would
normally be determined by the core storage capacity of the com-
puter, but a technique could be developed that would provide for
thousands of abbreviations.

Although these are the only filing rules that were programmed,
there are many others that could have been easily incorporated but
were not required in the indexes to the Presidential Papers.

On the surface these few filing exceptions sound trifling, but
in the Benjamin Harrison index (about 77,000 entries) we were
amazed to find that 47-1/2 man days of hand filing and 21 days of
machine sorting were eliminated and that the sorting of the "Date of
Document" column was so flawless it is no longer edited. The con-
tribution of computer application to this program increases in direct
proportion to the size of the index and therefore will be of crucial
importance in preparing some 400,000 index entries for the Wilson
Papers and 530,000 for the Taft Papers. The Presidential Papers
Section has estimated that 8 months of mechanical sorting and 1-1/2
man years of hand filing will be saved in preparing the index to the
Taft Papers through this system.

( 2) With the index now on tape, the tape file is ready to be
sorted by the sort key. Because the manipulated date is part of the
sort key, this places the file in sequence chronologically within each
writer or recipient. The file is now ready for a preliminary listing
to be used for editing.

(3) As the index is being listed, the computer program assigns
accession number or sequence number to each entry thereby creating
a unique identification. (See Figure 3). The entry plus this acces-
sion number is printed and written on another tape at this time, but
the sort key used in the sort is dropped as it is no longer needed in
the process. The new tape index now contains only 88 characters— 80
for the card image and 8 for the accession number.

147
Phase 2— Edit and Change Report ( see Appendix)

(4) To make a change, the editor records on the change form
(see Figure 4) the accession number and only the changing informa-
tion, (in a unit record process it would have been necessary to re-
cord the entire entry.) To change the sequence of an entry, the editor
records the "from accession number" and the "to accession number"
that will place the entry in its proper sequence. After the changes
have been keypunched, the cards are ready for an edit and change
report. (See Figure 5) .

The cards are now read by the computer and matched against
the tape index on accession number. The entry which is printed
before the change and after the change is then reviewed by the
editors to verify that all changes were made correctly. If a dis-
crepancy does exist, further changes are submitted, and the above
procedure is repeated until there are no errors.

Phase 3— Update Index (see Appendix)

( 5) The change cards are now ready to update the tape index.
All entries on the tape index that are to be changed are deleted, and
the remainder of the file is written onto another reel of tape at the
same time the corrected entry is written on an alternate tape. The
net result is a reel of tape representing the index less all changes
and a reel of tape containing only changes.

Phase 4— Final Report (see Appendix)

( 6) The change tape is now sorted by accession number and
is ready to merge in the main file. If the changes had been included
with the main tape, the entire file would have to have been sorted to
place the changes in their proper sequence within the index.

(7) The main file and changes are merged at the same time
the listing for photo offset is made. (See Figure 6). This computer
program also has the ability to run only selected pages of the index
in the event a page or pages are damaged.

148

Conclusion

This paper describes the solution of a fairly extensive but rel-
atively simple editing and filing problem by computer. It was not
necessary to attempt to develop a program that would accommodate
all filing rules, nor is it certain that such a program is possible for
a highly complex dictionary catalog. However, after a limited ex-
amination of the 180 pages of the "Filing Rules for the Dictionary
Catalogs of the Library of Congress," which is presumably as com-
plex and detailed as any codes the Library might attempt to com-
puterize, the problem may be somewhat less difficult than had been
anticipated.

In reading through the Filing Rules to note all the filing excep-
tions, it soon became obvious that the task would be endless. How-
ever, a pattern in these exceptions quickly became apparent. While
a printed manual of this type must necessarily describe each ex-
ception separately in detail, many of the exceptions were actually
similar to one another (some identical) when expressed in computer
logic. Moreover, many filing exceptions in the printed code would
turn out not to be exceptions in character for character sorting on the
computer.

Below are just a few examples of the many areas in which there
are similar or identical exceptions:

1. commas after surnames

2. commas after forenames

3. Bible entries-New Testament before Old Testament

4. corporate entries

5. the rules for person, place, thing, title.

Actually, all the rules for exceptions in the above areas are
identical in nature, with only one exception. The computer cannot
differentiate between such things as name entries ( Lincoln, Abraham)
and place entries ( Lincoln, Nebraska) . These entries could however
be identified by the cataloger as the entry was being prepared for
computer input by coding "A" for person, "B" for place, "C" for
thing, and "D" for title. Placing this code in its proper location
in the sort key would force the entries to sort A, B, C, D or person,
place, thing, title.

For a successful operation, a joint effort by the library and
data processing personnel will be required. In this partnership the
cataloger will have to understand thoroughly computer sorting to
supply the codes properly.

A project before us now is the author and title index to the
Catalog of Copyright Entries which is somewhat more complicated
than the indexes to Presidential Papers. Efforts here will necessarily

149

be slanted to the filing rules used in this index rather than to a gen-
eralized program. However, greater depth in filing is planned, and
it is confidently believed that the majority of the filing rules used in
this index can be accommodated. Problems in foreign languages, ab-
breviations, and in filing numbers as if they were spelled out can be
anticipated. The easiest solution to the latter two (abbreviations and
numbers) would be to use a variable length record that could be large
enough to spell out all words and write out all numbers.

A limitation of computer sorting that should not affect the out-
come is the number of characters that can be sorted. Although there
are no statistics on truncating names, experience indicates that a 50
character sort key would properly sort most entries. This means that
abbreviations or numbers could be used beyond the first 50 characters
without affecting the result of the sort. The maximum number of
characters the IBM 1401 computer can sort is 99, but it is desirable
from a data processing point of view to keep the sort key to a
minimum.

The computer file maintenance procedure obviously is superior
to a unit record process. It permits the editors to work entirely
from a listing whereas the unit record procedure; required the editors
to locate physically the card on the listing, make the change, and
then re-file the card. The development of the change report allowed
the editors to restrict their reviews to changes only rather than re-
viewing another complete listing after the changes had been made.

There are no statistics developed to indicate how much time is
saved with this updating procedure; and although it is not as dramatic
as the sorting technique, it will eventually save countless hours to
editors dealing with hundreds to thousands of entries, particularly
after the editors become better acquainted with the process.

150

Appendix I

PHASE 1-CREATE INDEX ON MAGNETIC TAPE

NEW INDEX WITH
ACCESSION NUMBER.
THIS NUMBER SERVES AS
A UNIQUE IDENTIFICATION
FOR EACH ENTRY AMD ALSO
REPRESENTS THE SEQUENCE

THE SORT KEY OF 44
CHARACTERS IS NOT INCLUDED
IN THIS RECORD. EACH RECORD
IS NOW 88 CHARACTERS LONG.

SEARCH WRITER OR RECIPENT, DATE
OF DOCUMENT, AND GENERATE A SORT KEY
AS PART OF THE TAPE RECORD.

INDEX IN" SHELF LIST ARRANGEMENT.
EACH RECORD IS 124 CHARACTERS LONG.
SEE FIGURE 2.

SORT INDEX BY SORT KEY.

ABOVE SORT PLACES INDEX IN
SEQUENCE BY DATE OF DOCUMENT WITHIN
WRITER OR RECIPENT.

PRELIMINARY INDEX IS REVIEWED
BY EDITORS. CHANGES ARE RECORDED
ON THE FORM SHOWN IN FIGURE 4.

151

PHASE 2-EDIT AND CHANGE REPORT

TO PHASE 3

MATCH CHANGES AGAINST INDEX ON
ACCESSION NUMBER. PREPARE REPORT
SHOWING ENTRY BEFORE AND AFTER
CHANGE IS MADE. IF ERRORS ARE
FOUND IN THE CHANGE REPORT, CHANGES
FOR THESE ERRORS ARE MADE AND STEP
4 IS REPEATED. NOTE CHANGES ARE
NOT ACTUALLY MADE AT THIS TIME.
SEt PHASE 3-UPDATE.

PHASE 3-UPDATE INDEX

> 'MATCH CHANGE CARDS AGAINST INDEX

ON ACCESSION NUMBER. WRITE INDKX BUT
DO NOT INCLUDE CHANGES. WRITE A
CHANGE TAPE THAT INCLUDES ALL CHANGES.

TO PHASE

152

PHASE 4-FINAL REPORT

RYAN MRS L M TO MRS J ft UALOH

fill

191Z SE 16

en?
S'!

D.1I
£•2

-J)l
<

h44

Z55

(/)7|
111
CM

«•••

'^AlA BF \J»lM» ol RECIPlfN*

OATC Of DOCUMENT

«ooo oolooooo

CASE

rue
Illll

no. or

oTo

ADDITIONAL INFORMATION

00 0000000 000000000

4''1
5 0
|

>s
• i

•

in|iiiiniiiiiiiiin|i|ii|iiiiiiiiii
iissfttilttsifsstitlifitiiftiiittasti

IllllllltlllltllllllllllllllltlSltSSI

444444|44444|4444|4444444444444444444

iiiiliiiiiniiiiiiiiiiiiiiiilliiiiiii

l(llll(tlliilil|(l(llllll|lMllltlilt

J7777jnTJ7J7777JI?l7?7777MJl»7777»I
lllllllllllllMtlllllllllllllllllllll

l|lll9S|4lllllllll|llllMIIIIMIHMI

i till ••• n •at4aan*»anMBii ••••»•* *•*»•••>

lli ll 1 1 i 1 ll 1

2222|Z|22Z22
333333333333
444444444444
5S55555|5b55
IIIMIIIIllS
777777777777
III Illll III!

Illlllllllll

••••••••••an

1 1

31
44
55
(i
7 7
II

• •

1 i 1 1 1
22222
3)131
44444
55555
iSCC
77777
Illll

Illll

«••••

111

222
3)3
444
555
(((
777
III

III

• ••

111111111111111111

222222222222222222
1111111111))))))))

444444444444444444
555555555555555555
tllllliilililKKt
777777777777777777

iiiiiiiiiiiiiiine

Illlllllllll 19 II 9 9

• ••••ii*a»nanMn*niiid

Figure 1.— Card Format

153

< SORT KEY GENERATED BY COMPUTE

-J

"- CARD IMAGE

WRITER OR RECIPIENT

DATE

WRITER OR RECIPIENT

DATE

MABE D E

19120916

27 MABE D E FR WW

1912 SE 16 '

MACCOCLEY J

191209 6

27 MCCOOLEY J FR WW

1912 SE 6

MACDOUGALL R V

19120916

27 MACDOUGALL R V FR WW

1912 SE 16

MACINTYRE J

19120916

27 MCINTYRE J FR WW

1912 SE 16

MADISON C W

19120916

27 MADISON C W FR WW

1912 SE 16

'MADISON C W

19121099

27 MADISON C W

1912 OC-OE

OBEAR A C

19120916

27 OBEAR A C FR WW

1912 SE 16

OBRYAN G S

19120916

27 O'BRYAN G S FR WW

1912 SE 16

'OBRYAN G S *>

1912011',

27 O'BRYAN MRS G S TO WW

1912 JA 14

OLEARY J

19120916

27 O'LEARY J FR WW

1912 SE 16

OMEN W A

19120916

27 OWEN W A FR WW

1912 SE 16

PARSONS B C

00000001

27 PARSONS B C

SEE

SCOTT

OC000001

27 SCOTT C R

SEE

WALEN

18901231

27 WALEN J A TO WW

1890 DE 31

MALEN

18S01288

27 WALEN J A TO WW

1890 DE

WALEN

18908888

27 WALEN J A TO WW

1890

WALEN

18909988

27 WALEN J A TO WW

1890-91

WALEN

9999S99S

27 WALEN J A TO WW

WALEN

99S9S999

27 WALEN J A TO WW

•ND1890

WALEN 9

19120916

27 WALEN MRS J A FR MRS L M RYAN

1912 SE 16

Figure 2
Sample Magnetic Tape Records

03612960

22 MCVEIGH H R TO DSL

1888 AP 20

IftRA AP 1

03812720

22 MACVEIGH J FR DSL

1886 OC 11

03812880

22 MACVEIGH JAMES TO'GC

1886 SE 23

Lflia NO ?n

8WITH CIRCULAR
7VJ7

03813040

22 MCVEIGH N TU GC

1888 MY 28

03813200

22«MCVEY CHARLtS S TO GC

1885 OC 3

IflHC, NO 19

03813360

22 MCVICKER J H TO DSL

1887 SE 27

1BB& SI 12

1TELEGRAM

03813520

22 MCWHIRTER A J TO GC

1887 MR 1

03813680

03013760

22 MCWHOOD E JR FK CLP
??«MC-uiit LAttS c P rn QSJ

1885 SE 17
IDA? ND in

03813840
Oi8l3920

22 MCWILLIAMS J FR DSL
22 MCWILLIAMS J TO DSL

1885 AP 24

03814000
03814080

22 MACWILL1AMS WILLIAM TO GC
22 MACY E J TQ»GC

1885 St 10 '
18H7 JL 4

03814160
03814240

22 MACY MRS L S FK DSL
22 MACY S J TO GC

1887 JL 21
1888 MY 21

2
2

03814320
03814400

22 MACY MRS S J TO»GC
22 MCZANE A TO f MHAKTON

1904 JA 14
1887 Ft 19

03814480
O3814560

22 MADDEN D TO GC
22 MADDEN G H TO J BLACK

1895 NO 25

03814640
O3B14720

22 MADDEN H A TO GC
22 MADDEN H A FR DSL

1889 JA 25
18H9 JA 29

03814800
03814880

22 MADDEN J TO GC
22 MADDEN T E T AL TOoGC

1887 JE 7

03814960

03815040

22 MAOUIN L L FO»F F CLEVELAND
22 MADDGCKS H F FR DSL

1892 NO 15

03815120
03815200

22 MAODOX L TO GC
22 MADDUX H C FK DSL

1880 OC 21
18B6 JL 26

3WITH CLIPPING

03815280
03815360

22 MADDUX S W TO DSL
22 MADDUX -T W TO GC

1888 JE 11
1892 NO 9

2
1TELEGRAM

03815440
03815520

22 MADOUXS S W TO GC
22 MADIGAN A H TO GC

1868 JE 7
1886 DE 11

1
1TELFGRAM-V30

03815600
03815680

22 MADIGAN J TU GC
22 MAOIGAK W T TO DSL

1887 MR 1
1886 DE 4

1TELEGRAM-V30

03815760
03815840

22 MADIGAN W T TO DSL
22 MADISON A Z FR OLP

1888 AG 21
1885 AP 30

03815920

03816000

22 MAOISON A Z TO GC
22 MADISON COFFEE POT CO TO GC

1888 OC 29 ,
1886 DE 8 i

4
JWITH ADVERTISEMENT

Figure 3
Preliminary Listing with Accession Number

154

[

Accession

•o
MM ,

loossdon
•o
TO j

WRITER OR RECIPIENT

DATS

1 1

?.l

Accession
So

RON ,

Accession

•o

TO j

, Cent if Ca.
AJOmOHU, IWORKATTCII . Cross C

rd CHABCB OCBB
- 1 Add
2 Delete

ftj

Fig. 4— Change Sheet

HtfSIOfNTHL >«>t«S tOlT MMMII HOM CASE H

CHANGE N.UHSES, NO.
.. <DOS 01071970 M

DOCUMENT iiUiJ P«US

•EFEOEHCE

YES

. CHANGE

r*M

•KHUCC

0*2720
0(2720

„

LLKMSON J FH H IE«»L€ 1*7

NO « 2 « »I«I tf 1«7 M, 10

OM770

095600
095600

ST»« C J« TO'JHl 1*1

•» 1> 1 2

NO
NO

TES J B TO.J«1 1*1

J* 2* 1 1 DUN61ISON 1110 m 23

IES j e to«j»i 1*1

J» 24 1 1 DUN6LISON 1130 M> 2>

Figure 5
Edit and Change Report

155

i »MACAULAV GRAHAM CATHERINE PR GW
1 «MACAULAY GRAHAM CATHERINE TO«GW
1 MACAULAY GRAHAM CATHERINE FR GW
1 »MACAULAY GRAHAM CATHERINE TO«GW
•MACAULAY GRAHAM CATHERINE TO*GW
•MACAULAY GRAHAM CATHERINE FR GW
•MACAULAY GRAHAM CATHERINE FR GW
MACAULAY ZACHARY TO*W W GRENVILLE
MACAULEY AULEY TO«GW
MACAULEY AULEY FR'TOBIAS LEAR
MACAY SPRUCE
MCCABE ALEXANDER TO GW
MCCABE HENRY TO GW
MCCALL GEORGE TO*GW

•MCCALLISTER ARCHIBALD TO PA COUNCIL
MCCALLISTER ARCHIBALD FR GW
MCCALLMONT JAMES TO GW
MCCALLMONT JAMES TO GW
MCCALLUM KENNETH TO* JAMES GORDON
MCCALMAND JEAN FR«JOHN MCALISTER
•MCCALVEY WILLIAM ET AL TO»I PUTNAM
•MCCARMICK GEORGE TO GW
•MCCARMICK GEORGE FR GW

MCCARNY FRANCIS-BRITISH STRENGTH
1 MAC CAR THY EUGENE

1 MCCARTHY MARTAIGUE JEAN BAPTISTE DE
1 MCCARTY DANIEL TO GW
1 MCCARTY DANIEL TO»GW
MCCARTY DANIEL FR GW
MCCARTY DANIEL TO GW
MCCARTY DANIEL TO GW
MCCARTY DANIEL TO GW
•MCCARTY DENNIS FR GW
•MCCARTY DENNIS TO«GW
•MCCARTY DENNIS FR*GEORGE MERCER
•MCCARTY DENNIS TO«GW
MCCARTY DENNIS FR GW
MCCARTY DENNIS TO«GW
•MCCARTY DENNIS TO GW

MCCARTY DENIS
•MCCARTY PATRICK TO»GW

MCCASKEY BRITISH SHIPS 6 TROOPS

•MCCASKEY ALEXANDER TO*GW
•MCCASKEY ALEXANDER TO*GW
1 MCCASKEY ALEXANDER TO GW
1 »MCCASKEY ALEXANDER TO GW
1 MCCLANACHAN ALEXANDER FR GW
I MCCLANACHAN ALEXANDER FR GW
1 •MCCLANACHAN ALEXANDER FR GW

1787 NO 16

2 V14-P199 1

•1789 OC 30

« ]

1790 JA 9

4 V17-P289 ]

1790 JE

8 1

1791 MR 1

4 I

1791 FE 10

1 V17-P164 1

1791 JL 19

1 V17-P181 1

1794 NO 28

6 MEMORIAL ]

1791 SE 6

4 1

1791 NO 14

2 V18-P22 2

SEE

SALISBURY NC 1791 MY 30

1786 JE 26

4 1

1775 MR 31

2 1

1783 JL 21

4 1

1779 MY 4

2 EXTR-VERSO-GALBRETH MYS 3

1782 JA 1

2 CERTIFICATE OF SERVICE 1

1791 OC 8

3 V19-P8 ]

1792 FE 25

6 V19-P9 1

1782 MY 8

2 HAZEN MY 27 ]

1782 FE 22

1 V2-P36 2

1777 FE 7

2 1

1786 OC 31

4 TWO SAME DATE 1

1786 NO 27

3 V13-P247 1

1778 SE 29

2 DEPOSITION 1

SEE

JONES JOHN PAUL 1785 AP 13

SEE

MACARTY-MARTEIGUE J B DE

1769 DE 6

3 1

1784 FE 22

2 SPRAGUE TRANSCRIPT

1784 FE 22

1 V11-P130

1797 NO 2

1797 NO 6

1798 SE 19

1755 NO 22

2 V1-P288

1755 DE 30

1 V8-P18 RECEIPT

1756 JA 30

1 V3-P52

1756 MY 3

1 V8-P23 RECEIPT

1756 MY 10

1 V3-P171

1756 JE 2

1 V8-P29 RECEIPT

1756 NO 25

1 V8-P40 RECEIPT

SEE

MCCARTY DENNIS

1797 AP 26

•1780 JE 1

2 INTELLIGENCE REPORT

1789 JL 4

2 V19-P16

1789 JL 8

2 V19-P17

1793 AG 19

5 V19-P18 WITH RECOM

1793 AG 20

2 V19-P20

1777 AP 30

1 V3-P111

1778 MR 26

2 V5-P184

1778 MR 26

Figure 6
Used for Photo -off set

LIBRARY APPLICATIONS OF DATA PROCESSING EQUIPMENT
AT ABBOTT LABORATORIES

Walter A. Southern

I. Introduction

The Science Information Services (SIS) of Abbott Laboratories
is a medium-sized industrial research information service. Its
primary function is to serve Abbott's one-thousand man research
staff; but being a company-wide service, all North Chicago divisions
and, to a lesser extent, Abbott's international offices, avail them-
selves of SIS services.

The SIS has a staff of five librarians, nine information scien-
tests, and fourteen clerical assistants. It is a collection of 32,000
volumes with 2,000 volumes being added annually; it receives more
than 1,000 different journals, and it has an annual circulation of
135,000 items. Since Abbott is a member of a strongly research-
oriented industry and because the SIS promotes its services exten-
sively, the SIS is a very active and heavily-used department.

The SIS first made use of data processing equipment in its
operations in 1951 when its journal circulation slips and charge rec-
ords were produced on punched card equipment. Since then various
recurring lists such as our list of current journals and our internal
dictionaries have been added. In 1959 an automated information
retrieval system for the current literature was begun. An IBM 101
Electronic Statistical Machine located in the department is used for
our retrieval operations. In addition, the department has an IBM 870
Document Writing System.

Many of our data processing operations are handled for us by
our corporate Data Processing Department. Close rapport has al-
ways been maintained with this department for professional guidance
in developing our own automated procedures and to make use of their

Walter A. Southern is Head, Science Information Services, Abbott
Laboratories, North Chicago, Illinois.

156

157

equipment. Abbott's Data Processing Department has a card IBM
1401 computer, a Burroughs 220 computer, a Burroughs 280 com-
puter, and accessory equipment.

Conveniently accessible to the SIS, in Abbott's scientific divi-
sions, are IBM 402 and 407 accounting machines, an 007 collator, a
514 reproducing punch, a 548 interpreter and a 1620 computer with
discpack.

This report reviews SIS' past experiences in the use of data
processing equipment for library procedures. In these operations
there are three distinct areas where automation is, or will be, used:
information retrieval, serials procedures, and acquisition-cataloging
procedures.

At the present time, SIS is mid-way in a study of all procedures
which lend themselves to automation. This study is complicated by
the fact that the company is presently planning for a new large-scale
company-wide computer system. The computer needs of the SIS are
being considered in this corporate study. Since the final decision as
to which computer will be obtained has not been made, the final plans
for operations must be delayed until this decision has been made. In
the present study, however, the feasibility of developing computer
programs on an interim basis which will make use of present com-
puter equipment is also being considered. Such a plan may mean
that the procedures will have to be redesigned for the new computer,
but hopefully it is expected that it will mean only a minor reprogram-
ming job. Eventually input-output consoles should be in our library
on-line with our new computer for information collection and inquiry.

II. Information Retrieval

The SIS has had an automated information retrieval system in
operation since 1959. The system now contains 35,000 literature
references on Abbott products and other drugs pertinent to Abbott
operations. With a five-year coverage of the literature in the sys-
tem, information searches are now very productive. Approximately
200 searches per year are now being run.

The search system is currently based on the use of an IBM 101.
In designing the system of mechanized searching, a method making
use of random numbers in preference to direct coding was selected.
A random number system was selected for several reasons: (l)
Only one punched card per article is needed which keeps the size of
the search deck to a minimum. (2) An open-end dictionary is pos-
sible which allows us to add new index terms in any subject area as
needed. (3) Machine time required to search the punched card deck

158

is kept to a minimum, and (4) It is possible to correlate all index
terms with an individual article as desired. From the Fischer- Yates
tables, SIS has had generated and tested for randomness some 10,000
random numbers. The search field for the random number coding
makes use of a 10 X 40 field on the punched card (columns 1-40 and
rows 0-9) . Twelve-digit random numbers are used for the index
terms. Thus, random number 063-123-269-304 which represents
penicillin V is punched as follows: digit 3 in column 6, digit 3 in
column 12, digit 9 in column 26, and digit 4 in column 30 (see Figure
2) . Mathematical computations have shown that it is safe to super-
impose 50 index terms per article in our 10 X 40 search field. Be-
yond 50 superimposed index terms, false relationships show up
excessively in searching. The maximum number of terms ever
allowed is 70. In addition, rows 11-12 of columns 1-40 are used for
supplementary direct coding. Terms coded here are either role in-
dicators or frequently used terms, e.g. clinical report, LD5Q, intra-
venous drug administration. Other coding for an article consists of
the year of the article (columns 73-74) , the corresponding abstract
number (columns 75-79), and the machine control punches (column
80). (See Figure 3).

SIS information scientists abstract and code the current litera-
ture. Strictly controlled chemical and biological dictionaries are
used in assigning index terms. For each index term which is in the
dictionaries, two punched cards are prepared. The first is used in
preparing the printed dictionaries. (See Figure 4). The index term
is punched alphabetically in columns 16-80 and the corresponding
random number is punched in columns 1-12. Machine controls are
punched in columns 13-15. These cards are filed alphabetically and
used whenever a new edition of our dictionaries is prepared on the
IBM 1403. The second punched card is a tub index term card. The
index term's random number is punched as it appears in the search
field in columns 1-40. The term's subject serial number is punched
in columns 41-45, and the term itself is alphabetically punched in
columns 46-72. These tub cards are gang punched on an IBM 519
Document Originating Machine in quantities of either 25 or 50, de-
pending upon the frequency with which the term is used. The index
term is interpreted onto the top of the card by an IBM 557 Alphabetic
Interpreter. This gives a visual quality control check so the term
is properly filed. Green card stock is used for original tub index
term cards. These cards are then used as recorder cards when the
supply runs low.

The tub cards are filed horizontally in two tubs 10' 5" x 2' 5"
maintained in the SIS. Each tub accommodates 2,400 index terms
which are filed behind guide cards which are 1-1/2" longer than the
tub cards. The index terms are typed on the guide cards for easy
identification when pulling tub cards.

159

In processing an article for machine indexing, a Documentation
Assistant assembles from the tubs those index term cards which have
been assigned to the article. A card punched with the date of the
article is also pulled as are other punched cards which correspond
to the abstract number assigned to that article. Other machine con-
trol cards are assembled into the decklet. These decklets are then
sent to the Data Processing Department where all the random num-
bers corresponding to the indexing numbers are accumulated through
all the other cards, including the master search card (a top quality,
edge coated card) and the duplicate search card. An IBM 519 is used
to accumulate this information. The two search decks are then re-
turned to the SIS.

The remaining punched cards for each decklet are kept together
by abstract number and put into dead storage. The system was
planned so that it can be converted easily to searching on a computer.
In the accumulating procedure, the serial numbers corresponding to
the index terms are not accumulated. It is the serial numbers which
will be transferred to tape, along with abstract numbers. Searching
will then be done by serial numbers rather than random numbers.

Studies show that it is faster and less expensive to pull and
accumulate tub index term cards than to key punch all the random
numbers. One disadvantage is the amount of space needed for the
tubs. Should this operation ever be expanded beyond two tubs, an
additional disadvantage would appear: the distance a Documentation
Clerk must walk to pull cards from more than two tubs.

Requests for machine searches are programmed by information
scientists. The formulation of the search question must be carefully
done in order that search terms be correctly selected and arranged.
A maximum of four searches can be run at one time. Once the search
is programmed, the Documentation Clerk runs the search on the IBM
101. The appropriate random numbers, direct punches, and/or date
are wired on a panel board. Once the proper abstract numbers have
been selected, they are put through the IBM 870, and a printed list of
pertinent abstract numbers is produced. The corresponding abstracts
are then pulled from the abstract card files and a check made to see
that the abstracts are pertinent to the search request. The final
answer to the requester is provided by a packet of the pertinent ab-
stracts. All the procedures connected with punched card preparation
and the machine searches are well handled by Documentation
Assistants.

For machine searching, a permanently wired panel is used
which will accept a question containing up to four index terms: ABCD.
A typical search would be: The use of ristocetin (A) in combination
with penicillin V (B) in the treatment of bacterial endocarditis (C)
in children (D). The search panel has been wired to search all pos-
sible combinations of the ABCD terms. If this search, for example,

160

did not provide an answer containing all index terms ABCD, it is
quite possible the inquirer would be interested in the articles con-
taining index terms ABC, ACD, or AD. This search procedure is
very useful for generic-specific searches for a question such as:
Has hydrochlorothiazide been used in the treatment of diabetes mel-
litus ? Here again, the programming must be emphasized. The in-
formation scientists must fully understand these coding principles
for best results. This request would be programmed: hydrochloro-
thiazide (A), thiazide derivatives (B), clinical reports (C), and
diabetes mellitus (D). If no ABCD cards were produced, it is likely
that the requester would be interested in clinical reports on the use
of other thiazide derivatives in the treatment of diabetes (BCD), or
he might be interested in BD cards which would be experimental
reports on the uses of these drugs in experimental animals.

In systems using random numbers, false relationships of two
types are inherent. The first is that the components of several
punched random numbers may form the random number of an index
term not in the article. And the more index terms assigned to the
article, the more likely false relationships will develop. For this
reason, too, it is always more desirable to search on four terms for
desired information rather than one term since this cuts down on
false relationships. The second type of false relationship is the well-
known blind Venetian- Venetian blind example. The terms you are
searching for are in the article, but not in the relationship you de-
sire. Both types of false relationships must be sorted out before the
answer is supplied to the requester.

A search of the 35,000 articles in the system now requires
four minutes for panel wiring and 78 minutes for the IBM 101 run.
Additional time is required to pull the pertinent abstracts for the
abstract file.

As SIS' store of abstracts grows and as more and more machine
searches are being made, it is inevitable that the present system
must be transferred to a computer. These plans are presently being
made and, as noted earlier, the present system was designed for
easy transfer to the computer. A program is being prepared so that
the serial numbers (corresponding to index terms) and abstract
numbers on the decklets now in dead storage will be transferred to a
tape. It has not yet been decided whether an inverted file or serial
file will be used.

Computer searching will make it possible to eliminate all false
relationships caused by superimposing index terms; it will not, of
course, eliminate the Venetian blind type[ syntax] of false relation-
ship. Computer searching will also make it possible to run searches
faster and more conveniently. In addition, it will be possible to
multiplex fifty searches. As a result, SIS expects to run many more
searches and not be concerned with the time factor of running

161

searches as it now is with the IBM 101. A further goal is to prepare
a second tape with abstract numbers and complete bibliographic
citations for each. It will then be possible to produce printed bibliog-
raphies and also to make available a selective dissemination of in-
formation (SDl) program which would be a much-used service by
SIS clients.

The present use of the IBM 101 gives the convenience of run-
ning searches whenever needed. With computer searches, one hour
per week will probably be assigned for this purpose. The convenience
of searching will be lost, but the benefits to be gained will outweigh
the inconvenience.

III. Serials

The procedures for the control of current journals present the
greatest challenge in SIS operations because of the importance of
current journals in the research operations and because of the heavy
use of current journals by the scientific staff. It is quite natural
then that our first use of automated library procedures was for the
preparation of journal circulation lists. This automated procedure
for journal circulation lists which was begun in 1951 is still in use
today. Circulation slips and duplicate charge cards include the
following information: name of journal, copy number, serial number,
and employees' names and department numbers [ See Figure l] .
Punched cards are used for the input. When the system was started,
the IBM 407 accounting machine was used for producing the lists,
but today the IBM 1403 printer is used. In addition to the circulation
lists and charge cards, lists of journals routinely received by in-
dividuals are also produced on an annual basis. These lists are used
for annual reviews of current reading, both by the individuals and
by their department managers. Other by-products are printed scrolls
of all circulation lists which are used for many reference purposes,
printed guides for the storage of circulation lists, and statistical
data. On an annual basis, SIS receives a report on the number of
individuals receiving journals routinely, the number of journals re-
ceived by each department, and the total number of journals routinely
circulated (98,357 journals in 1964) . These data are of special
interest at this time as SIS is planning additional research facilities
away from the present location; they will be useful in planning
library facilities for this area.

The next step is for computer control of all journal procedures.
This will include: purchase orders for new journals, annual renewal
of journals, checking in of current journals, journal claims, overdue
notices for journals out of circulation, a union list of journal holdings,
plus all procedures covered by the present automated system.

162

A computer program for all SIS journal procedures has been
developed by Mr. Donald H. Kraft, Industry Marketing Representative
for IBM in Chicago. The program has been developed for the IBM
1620 computer, but the basic principles are applicable to any com-
puter system. The flow charts for this program follow. The flow
charts of a second program, using punched card equipment instead
of a computer were prepared for SIS by Mr. L. R. Chapman, Sales
Representative of IBM's Evanston Office.

A Generalized Library Serials Check- in & Routing Procedure Using

Using Data Processing Equipment

MONTHLY

The Call Card calls from disk
storage the computer program
which generates the monthly
Expected Serials File. This
file consists of a punched IBM
card for each copy of each
serial expected the following
month. The card is punched
with journal name, volume
number, month, copy number,
destination library, bindery
code, class number.

Renewal and bindery notices
are written when needed.

Trigger

Routing

Slips

Daily Check- in Procedure

As the serials arrive, a clerk
pulls from the Expected Serials
File the corresponding IBM
cards. These cards will trig-
ger the printing of routing
slips.

163

Daily Preparation of Routing Slips

Routing
Slips

— >

The Expected Serials
Cards are fed into the
computer. The disk file
contains the routing
lists, which are stored
magnetically on its disk
surfaces. The computer
then prints the paper
routing slips correspond-
ing to the Expected
Serials Cards. The rout-
ing slips contain serial
name, number, issue,
copy, and date, as well
as a numbered list of the
recipients.

Mail

The serials clerk staples
the routing slips to the
appropriate serials.

164

Mail to next
name on rout
ing slip

After a reader has finished the
serial, he returns it to the library.
The serials clerk key punches an
IBM return card with the journal
identification and the line number of
the last name scratched off the list.
The return card can contain informa-
tion for more than one transaction.

The return card is read into the
computer, which searches the disk
file for the appropriate routing rec-
ord. The computer then posts to-
day's date beside the next name on
the disk record, indicating the date
the serial was sent to him. Over-
due notices are printed for those
serials not returned within the loan
period. The return cards are
destroyed.

165

Weekly Printout of Current Routings

Call Card

Routing

Informa-
tion

Computer

Routing

Inform a
tion

Each week, a program is initiated
by a call card to print "CURRENT
ROUTING INFORMATION." The
computer prints the journal
name, identification number,
date, volume, number, copy as
well as the names of the next
two persons on the routing list.
This list is consulted by the
serials librarian to answer
queries relating to the where-
abouts of particular journals.

A sample is shown.

CURRENT ROUTING INFORMATION
WEEK OF 12/18/64

THE LIBRARY JOURNAL, 7825, 10/64. VOL. 34, NO. 4, COPY 1
JOHN JONES DEPT. 72

BILL SMITH DEPT. 90

166

Information relating to additions, de-
letions and address changes to the
routing lists are punched into IBM
cards. These cards are fed into the
computer which makes the necessary
changes to the lists stored magnetically
on the disk surfaces.

Monthly Claims Reporting

Expected
Serials File

Serials Librarian

Cards remaining in the Expected Serials
File at the end of the month are examined
by the serials librarian. Those cards rep-
resenting claims are read into a printer or
computer, and claims reports are written.

Claim

167

Serials Holdings List

Periodically, the computer can print a
Serials Holdings List from informa-
tion stored magnetically on the disk
file. This list can be prepared in
multiple carbon copies or directly on
reproduction mats by the computer.

A Keyword -in- Context (KWIC) Index
can also be prepared by the computer,
thereby indexing the holdings list by
each significant title word

IV. Acquisition-Cataloging

Up to the present time none of the acquisition- cataloging pro-
cedures have been automated. But in our study we believe that a
total systems approach will be established for them. Whether or not
the procedures will be developed on an interim basis for the Bur-
roughs 220-280 system has not been decided.

The following procedures will be included in the automation of
the acquisition- cataloging operations:

1. Purchase orders.

2. Outstanding order file.

3. Claims.

4. Book catalogs (and possible catalog cards until the complete
collection is automated) .

5. Book charge cards, book pocket labels, and book spine
markings.

6. Book overdue notices.

7. New book announcement lists.

8. Printed lists of books on specific subjects.

9. Various statistical reports.

The automated procedures being considered for the acquisition-
cataloging procedures follow closely those developed by Mr. R. E.
Durkin for the IBM libraries at Kingston and Poughkeepsie. Flow

168

charts, diagrams, and examples of their operations follow. This
progran, called the Program Library Tape, is in the process of being
documented by IBM, and the descriptive brochure should shortly be
available.

Program Library Tape

A. Basic Equipment needed:

1. 1401 computer with 4,000 positions of core storage

2. 4 tape drives

3. 1402 punched card reader

4. 1403 printer.

B. Hourly Costs:

The estimated hourly charges are in the area of $45 - $50.
C. Program Features:

1. AUDIT - Checks, and in some cases, corrects all key-

punched input; prints a proof-reading copy of
the input; and transfers information from
punched card records to magnetic tape.

2. PRINT - A printing program to prepare reports of

listings in any of variable page formats.

3. INDEX - A permuted title indexing program.

4. TWOUP - A second printing program which prints data

on two column page format.

5. THREE In combination, they print 3x5 cards.
(&) FIVEP -

6. FOURD - Allows further refinements in producing 3X5

cards.

7. SUPVS - The basic function of SUPVS is to control the

loading and unloading of the other programs.

V. Centralized Information Retrieval Systems

It has been evident for many years that no company can by it-
self adequately index, abstract, and disseminate all current and past
literature pertinent to its operations. This is especially true in the
pharmaceutical industry with its extensive interdisciplinary research.
As a result, many new centralized retrieval systems are now in
operation. And we must not overlook the many proposals for inter-
national cooperation in medical documentation services, some of

169

which will certainly materialize. On an industry basis, the American
Petroleum Institute and the American Society for Metals have already
established computerized information retrieval systems. Other
established automated information retrieval systems are now avail-
able in the chemical-pharmaceutical areas.

Because of these developments, the best laid plans of a pharma-
ceutical information service for developing its own retrieval stystem
are going to be influenced— perhaps even drastically reduced— as
these new centralized services develop. More likely, however, it
will be that the individual company will tailor-make these informa-
tion services to meet its own information needs.

In the Abbott SIS the following automated centralized informa-
tion retrieval systems will be influencing the development of its own
retrieval systems:

1. Steroid Index of Patents

The U.S. Patent Office for several years now has been issuing
a punched card index of all new steroid compounds reported in the
literature and in U.S. patents. An electronic sorter is needed to use
this service. No plans have been announced to transfer this operation
to computer usage. But it is not inconceivable that this will be done
since the file is now becoming unmanageable.

2. Information for Industry Uniterm Index to U.S. Chemical Patents

This service has been in operation for ten years. The basic
service is a manual, book- type, Uniterm index of all United States
chemical patents. A computer tape index became available two years
ago; thus both manual and computer retrieval methods are available.
An improved computer search system which would incorporate all
minor terms (terms used less than ten times in any one year) is
being planned. If this should become available, it would greatly en-
hance the value of this service for pharmaceutical firms.

3. RINGDOC

This indexing- abstracting service of the pharmaceutical litera-
ture was started in July 1964 by Derwent Publications in London. It
is perhaps the most advanced centralized information retrieval sys-
tem now available. It provides 40,000 abstracts each year along with
manual, punched card, and computer tape indexes. The present tape
index contains only alphabetical index terms and bibliographic cita-
tions. But in October 1965, it is planned to add a detailed chemical
coding system.

4. Chemical Titles

In January 1965, the American Chemical Society made available
computer tape indexes for all entries in their publication, CHEMICAL

170

TITLES. The tape format and search programs are designed for an
IBM 1401/1410 computer system. The tapes will enable users to
make their own searches and enable us to provide a selective dis-
semination service. (This new service also provides for the alterna-
tive of having searches made in the offices of Chemical Abstracts in
Columbus for those companies not having computer services.)

5. FARMDOC

For the past three years, an abstract-index service of pharma-
ceutical patents from the major countries of the world has been
available from Derwent Publications in London. Both manual and
punched card indexes are supplied. As this store of patents grows,
it is likely that the punched card index will also be supplied on com-
puter tape. This service has become so successful that the supplier
is planning to extend his patent coverage into the areas of food and
agriculture.

These five examples of centralized information retrieval ser-
vices have been noted to illustrate how national information suppliers
will be influencing the operations of company information services
in the future. More and more, the company information service
must be planning for reprogramming and "packaging" computer tape
indexes for their own use, making use of computer service centers,
and justifying the costs of national services in relation to company
needs.

171

ILLINOIS UNIV.ORG.AB.

I 6 4400 PO

Return within two days.

_ZAUGG_

•FREIFELDER ILLINOIS UNIV.ORG.AB. 1 6 4403 PO

JOHNSON R P.
SCHOCK

Return within two days.

COLE J W

: ZAUGG

432-

• COTA

^ «F'*EIFcLOEa

48?

PIECZ : "CLOSE 482

SMART

' JOHNSON R P

482

SHIPTON

SCHXK

480

COLE 'J W

432

*CGTA

453

PIECZ

453

SMART

453

SHIPTON

4!>3

ABBI

ABBOTT LIBRARY

*Pass Journal to next person.
Do not clip or mutilate Journal.

Figure 1

Computer Produced Journal Circulation Lists
and Charge Records

172

pmrMAYYMTWKVT. P1?NT(*TT.T.™ 00^5

,«sss»,

"„"£

1 II 1 1. 1 1 II 1

Tf** •«*•*•
1

-. ooooooooooooHooo'ftoooo'oiio'ODODOiiocooooooG

muN

00000||00|0|OOOOOOOOOOOOtlOO

OOOOCC

3 11111111111111111111111111111111111

* 22222222222222222222222222222222222

B33333I.3333333333333333333333333333
444444444444444444(4444 4| 4444444444

55555555555555555555555555555555555
6666bEG6SS666666666666B66G66C6 66666
77777777777777777777777777777777777

"iimiiiiiimttiiimtiiiiiiiiiiiiiiiii

9999999993999999999999999|999999S9999999

inn

22222
33333
44|44

5555B
6GCG6
77777
38808
99999

111111111111111111111111111
222222222222222222222222222
333333333|33|33333|3||33333
4444444|4444444444444444444

555J5555|555555|J555555|555
666 6|G6666666666686S6666666
|7777|77777777|777777777777
8J88S8|a38||S888S8888B888a8
99999999999999999I9IS9I9999

1 1
22
33
44
55
36
77
88
93

111111
222222
333333
44<44|

555555
G66G5 3
777777

388088

iiiiu

'•mtmM

Figure 2
Index Term Card for Phenoxymethyl Penicillin

1 1 °6881'J

I 11 I ,.,.,',"':!,««.,

."."«i

•7r.;r

„„

HuMMa

JOOOOOOOg 0000000000 OOOOOOOOUOGgOOOOgiBgl

00000

000000000000000000000000000

| 0 0 0 0|9

1 1 1 n 1 1 1ll 1 1 11 Iflll 111 1 1 1ll 11 1 1 1 1 1 1 1

2Q2222|22222222222222222222B|22|222

| | B 33333J33 333|33333333333J3333333333
| 444444| 4444444 4|4444444||444444|444

| 5555E5555555555555555|55|55|555555;
g6666GGS666G66gB[SGGG666GG6£G666G66
7777777|7|7777777777777777777|77777

iiittii|ai|iini|iiitiiiiiiiiiiii||i|i|<

Iliatll 91111191 11111111191 111 II III! Ill II

11111

22222
33333
44444

55SSS
68665
77777

88888
99999

111111111111111111111111111

222222222222222222222222222
333333333333333333333333333
444444444444444444444444444

555555555555555555555555555
666666666666(66666666666666
777777777777777777777777777

8888888888S 888 8888888888888
"99999999999999-999999 99 9999

1 1
22
33
44
55

88
99

1 1 1 1 1ll

222222
33333|
444444

555555

61.665 S
777777

aS||63
999999

Figure 3

Search Card. The Term "Phenoxymethyl Penicillin"
is Superimposed on Other Index Terms

173

-RANDOM NUMBER DICTIONARY TERM -.- SEO NO

157 194 271 330 C GIG. ACETYLCHOLI NE-LIKE, AUTONOKIC AGENT, DIRECT- A 11694
157 194 271 330 D ACTING /3/ II i A 11694

-062 135 214 230 «ACETYLCHOLINEST£RASE -INHIBITOR /!/ A-IZO2A.

062 135 214 230 B i ACE T VLC H3L I « STE *ft SE INHIB., ENZYME INHIB., NICOTINE- A 12024
062 135 214 230 C LIKE, CHOLINE*3IC, MUSCARI NE-L IKE , AUTONOMIC A 12024

062 135 214 230 D AGENT *$ ENZYME INHIB., ACETYLCHOLINESTERASE INHIB. S A 12024

062 135 214 230 E I SEE SPECIFIC AUTONOMIC ACTIONS II $$ GANGLIONIC A 12024
062 135 214 230 F BLOCKING, AUT3N3MIC AGENT t* SPASMOGENIC /MUSCLE SKEL./ A 12024

-062. .135 214 230 G II_t . 4-1.2024-

EXOPHAGUS OIS. /539/ Ibl

ACHLORHYDRIA SEE STOMACH FUNCTION DIS. /544.0/ Ibl A 12678

053-067-171-177 oACHROMOBACTE.I JUI

053 067 171 177 B BACTEftlA /9/

109 176 190 354 'ACIDIFIER /3/

012 042 058 165 'ACIDITY /4/ A-13640

012 042 058 165 B PH, PROPERTIES, CHEM. , PROPERTIES, PHYS., /4/ /SEE A 13640
012 042 OSS 165 C ALKALINITY, II, NEUTRALITY lit A 13640

ACIOOPH1LE SEE EOSINOPHILE /5/

292 311 339 349 •ACIDOSIS /788.'6/ Ibl

Figure 4

Biological Dictionary. A Computer Produced Dictionary Used
in Connection with Abbott Abstracts

GENERAL REFERENCES

Griffin, H. L. EDP Procedures in Technical Library Opera-
tions. IDO- 16881. Phillips Petroleum Co., Atomic Energy Div.,
Idaho Operations Office, AEC, 1963.

IBM DSD Technical Information Center. Selected Papers on an
Integrated System for Disseminating, Storing, and Retrieving In-
formation. TR 00. 1103. Poughkeepsie, New York, IBM Development
Laboratory, Data Systems Division, 1964.

IBM Data Processing Division. Index Preparation and Library
Processing at Monsanto Chemical Company's Research Center.
White Plains, New York, IBM Data Processing Division, 1964.

Nicolaus, John J. The Automated Approach to Technical
Information Retrieval; Library Applications. Washington, D.C.,

174

Department of the Navy, Bureau of Ships, (For sale by the Super-
intendent of Documents, U.S.GPO) 1964.

Southern, Walter A. "Mechanized Processing and Retrieval of
Bio-Medical Information," Methods of Information in Medicine,
1:16-22, January 1962.

DATA PROCESSING EQUIPMENT AND THE LIBRARY
A Review of Available Tools and Their Potential Application

William E. Greiner

As it is preferable to discuss data processing equipment in
general, using UNIVAC equipment as typical examples of the tools
that are presently available; this paper will be concerned with four
main topics:

I. THE TOOLS OF THE TRADE
II. POTENTIAL AREAS OF APPLICATION
HI. MATCHING THE TOOLS TO THE JOB
IV. RECENT TECHNOLOGICAL DEVELOPMENTS.

It is hoped that the following comparisons will prove helpful to
librarians.

I. THE TOOLS OF THE TRADE

Data processing equipment, on the basis of general character-
istics, falls into three general classes or types:

A. Punched Card Equipment

B. High Speed Card Processing Equipment

C. Electronic Data Processing Equipment (Computers).

A. Punched Card Equipment

For purposes of this discussion, punched card equipment is
defined as equipment whose input is limited to the punched card,
and whose output is limited to a printer and/or a punched card.

There are two types of punched card installations:

1. Complete, "free-standing," installations.

2. Peripheral installations, operated in conjunction with
electronic data processing equipment.

William E. Greiner is Manager of the Educational and Library Mar-
keting UNIVAC Division of Sperry Rand Corporation.

175

176

The complete, "free-standing," installation is just what the
name implies. It is a complete, independent, operating installa-
tion consisting of a number of basic punched card units, capable
of creating and processing data in punched card form, and pro-
ducing usable output.

In most instances the major mission of the peripheral in-
stallation is to create input data for an electronic data processing
(computer) installation, where the actual processing and output
is performed by more sophisticated equipment.

There are two manufacturers of punched card equipment in
the United States, IBM and UNIVAC, which market equipment
originated by their predecessor company, Remington Rand. The
physical specifications of the IBM and UNIVAC punched cards are
identical. However, the columnar format and punching code of the
two cards are entirely different. Reference to Figure 1 will illus-
trate that the IBM Card contains 80 vertical columns, each con-
taining 12 punching positions; whereas, the UNIVAC card contains
90 columns, each column containing 6 punching positions.

Generally speaking, the punched card equipment line supplied
by both IBM and UNIVAC is almost identical. It includes the
following units:

(1) Key Punches and Verifiers.— To create the punched card
and verify the accuracy of the punching.

(2) Reproducing Punches.— To transfer punched data from a
deck of punched cards to a deck of unpunched or partially
punched cards. Or to create a deck of identical punched cards
from a single punched card (gang punching) .

(3) Interpreters.— To decode and print on the face of the
punched card the English language equivalent of the punched
code.

(4) Sorters and Collators.— To sort and/or merge decks of
punched cards into desired sequences.

(5) Calculators.— To read, calculate, and punched desired
results in punched cards.

(6) Printers.— To produce usable printed reports from data
contained in punched cards.

(7) Summary Punches.— To simultaneously punch cards con-
taining data included in reports produced by the printer.

Certain pieces of UNIVAC Punched Card Equipment have
unique features and characteristics which might justify brief
comment at this time.

(1) UNIVAC Key Punches. All punching is withheld until all
key depressions for a given card are completed. With the

177

depression of a single "trip" key, all holes are punched
simultaneously. This is in contrast to the "instantaneous"
IBM punching principle where the hole or holes are punched
as keys are depressed. This punch die -punching technique
permits an operator to correct any detected errors prior to
the actual punching of the card.

( 2) UNIVAC Optical Scanning Punch. The UNIVAC Optical
Scan Card can record up to 40 columns of marked data as
compared with IBM's 27 columns. The UNIVAC card can be
marked with any soft pencil, ball point, or standard pen. Sen-
sitivity control is variable. The Optical Scanning Punch can be
programmed to store constants, generate characters, and make
logical decisions (see Figure 2) .

(3) UNIVAC Collating Reproducer. This machine combines
the functions of a collator and a reproducer. Each function is
separately operable and permits simultaneous reproducing and
collating when desired.

Notwithstanding the speed, capacity, and flexibility of punched
card equipment and the economy of its operation when properly
applied, this type of equipment has certain basic limitations when
compared with other more sophisticated types of data processing
equipment. These limitations include the following:

1 l) Limited Card Column Capacity. In certain instances 80
or 90 columns are not sufficient to record the data from a
single transaction.

( 2) Limited Memory and Programming Capacity. The memory
and programming capacity of the punched card printer and
calculator is inadequate for many sophisticated programs.

(3) Inadequate Speed. The transaction volume of some opera-
tions requires more card read, punch, calculating, and printing
speeds.

(4) Non- integrated Operation. A punched card system is a
series of separate, off-line, independent operations which can-
not be integrated into a single automatic operation.

(5) Bulky Storage. Punched cards are bulky and costly to
store in comparison with magnetic tape.

B. High Speed Card Processing Equipment

The unique characteristics of this type of equipment permit it
to "bridge the gap" between standard punched card equipment
and electronic data processing equipment. Input is limited to
punched cards and output to high speed printers and punched cards.

The UNIVAC 1004 Card Processor is a fine example of the
high speed card processor, (see Figure 3) . Listed below are
some of the outstanding features of the UNIVAC 1004.

178

(1) Faster input, output, and printing speeds:

Card Read - 400-600 CPM
Card Punch - 200 CPM

Print - 400-600 LPM.

( 2) Calculation at electronic speeds:

Compute - 400-600 CPM. J

( 3) Internal memory:

Core Memory 1000-2000 Characters
Simultaneous read - calculate - print - punch.

(4) More programming power:

62 Program Steps

Each step performs up to 9 different operations.

(5) Concurrent operations:

Simultaneous read - calculate - print - punch.

(6) Communications capabilities:

Can be linked to other types of data processing equip-
ment via standard communications facilities.

(7) Field expandable:

Additional modules of core memory, program steps,
and magnetic tape may be field installed.

( 8) Input flexibility:

Can process 80 and /or 90 column cards.

(9) Economical cost:

See Figure 3.

TheUNIVAC 1004 Card Processor is the only operable high
speed card processing system on the market at the present time.
Other manufacturers have recently announced new systems which
may fall in this category.

C. Electronic Data Processing Equipment (Computers)

Electronic computers are classified on a number of different ^
bases. The most common of these classifications is as follows: 1

(l) Special Purpose vs. General Purpose

The characteristics of these two types of computers are
obvious. Typical special purpose devices would include
computers designed for missile guidance, processing

179

control, numerical control, etc. General purpose com-
puters, by their characteristics, are multi-use systems
which can be applied to any business or general scientific
area.

( 2) Digital vs. Analog

Digital computers accept input and produce output in digital
form. Analog computers measure values and produce out-
put in various forms other than by digital representation.
An electric meter is a typical example of an analog device.
It measures the flow of current and represents consump-
tion on a readable dial.

Obviously, for library applications, the discussion will be
limited to general purpose digital computers.

(3) Batch Processing vs. Random Access

Batch processing computers usually process data in some
consistent sequence or arrangement. Data is usually
stored in memory in a given sequence, and input is pre-
sorted in the same sequence. Random access computers,
in contrast, permit the processing of data on a random or
nonsequential basis.

(4) Externally vs. Internally Programmed

Externally programmed processors or computers are
programmed by means of a wired plugboard. Programs
for internally programmed computers are recorded on
punched cards, paper tape, or magnetic tape and then
stored in internal memory.

All punched card equipment is externally programmed.
Likewise the UNIVAC 1004 Card Processor also utilizes
this programming technique. Although the UNIVAC 1004
has up to 2000 characters of internal core memory, the
program is actually wired on a plugboard. This frees the
internal memory for actual working storage. Most modern
electronic computers are programmed internally.

1. Basic Components of all Digital Computers

Generally speaking, all digital computers, regardless of type,
characteristics, speed, capacity, or cost, contain the same basis
components, namely: input devices, main processor, auxiliary
mass memory, and output devices, (see Figure 4).

a. Input devices: Every computer configuration must include one
or more devices which permit data to be entered into the system.

180

The more sophisticated the system, the greater will be the variety
of input devices. The larger the system, the greater will be the
quantity of each particular device. The most common input devices
are the following:

1 l) Console typewriter

(2) Punched cards

( 3) Magnetic tape

(4) Paper tape

(5) Communication devices.

b. Main Processor: Most main processors contain three sub-
components, namely: internal memory unit, program control
unit, and arithmetic unit.

1 l) Internal memory unit:

The internal memory is usually either magnetic core,
drum, or thin film. The amount of internal memory may
vary from 4000 characters up.

( 2) Program control unit:

This unit reads and interprets the program which has
been stored in memory, and conditions the computer so
that it will automatically complete the program.

(3) Arithmetic unit:

This unit performs all the arithmetic computations that
are required to complete a program. Computations are
limited to basic arithmetic functions: add, substract,
multiply, and divide.

c. Auxiliary Mass Memory: Most large scale digital computers
include auxiliary mass memory devices as optional features.
This mass memory is usually housed in separate cabinets but is
always under the control of the main processor. This memory
may be in the form of magnetic drums or discs, or in some in-
stances, magnetic tape. Mass memory capacity may extend into
millions of characters. Access time is usually relatively slow in
comparison with internal memory.

d. Output Devices: Every computing system must include one or
more devices to accept, transmit, or produce the output from a
computer operation. Here again, the larger and the more sophisti-
cated the system, the greater will be the variety and number of
output devices. The most common of these devices are:

(1) Console typewriter

( 2) Punched cards

181

(3) Paper tape

(4) Magnetic tape

( 5) High speed printer

( 6) Special visual display devices

(7) Communication devices.

It is obvious, therefore, that the characteristics, the size,
and the price of general purpose digital computers may vary wide-
ly, depending upon the particular combination, variety, quantity,
and capacity of the individual components included in the system.

2. Types of General Purpose Digital Computers

Figure 5 is a tabulation of the basic characteristics of gen-
eral purpose digital computing systems presently marketed by
UNIVAC. These systems are classified into three general cate-
gories based on these characteristics:

a. Business Data Processing Computers— Batch Processing

b. Real Time Computers— Random Access

c. Scientific Computers.

The general characteristics of each of these three types of
computing systems follow.

a. Business Data Processing Computers— Batch Processing

Generally speaking, the average business data processing
job requires the processing of large masses of input data, rel-
atively simple calculations, and a large volume of output. For
example, large payroll, billing, accounts payable, sales analyses,
. and inventory applications involve the processing of large volumes
of individual transactions. However, the required arithmetic
calculations are very simple; i.e., hourly rate times hours worked,
unit cost times quantity, gross value times a discount factor, etc.
As output, the business data processing computer system must
produce large numbers of paychecks, vouchers, invoices, etc.
Consequently, fast output is required.

In order to meet these demands, the business data processing
computer usually has the following characteristics:

1 l) Fast input speed

( 2) Relatively slow computing speed

(3) Relatively small internal memory

(4) Little or no mass memory

( 5) Fast output speed.

It is usually practical and customary to process business
data on a batch basis, so there is relatively little demand for
random access capability.

182

b. Real Time Computers— Random Access

The unique characteristics of the real time, random access,
computer permit immediate, random access of memory when
required. The real time computer is used in situations where the
time requirements are particularly stringent. In most real time
situations, time will not permit the batching and pre-sequencing
of input data. In most instances, input is processed in random
sequence, memory is instantaneously searched, and the output is
delivered in seconds or fractions of seconds. To meet these re-
quirements, most real time, random access computers have the
following general characteristics:

1 l) Fast input speed

( 2) Fast access and fast computing speed

( 3) Moderately large internal memory

(4) Large mass memory

( 5) Fast communication devices

(6) Fast output speed.

One of the most common uses of the real time, random
access computer is in the general area known as information
retrieval. The information retrieval concept requires the ability
to interrogate the mass memory of the computer at will, usually
on a random access basis. The required information must be
delivered in the desired form almost instantaneously, and this
accounts for the term "real time."

The interrogation of the computer is, in many instances,
accomplished by the use of special communication devices. These
special devices may be located at remote points, and they are
linked to the computer via teletype or telephone lines. The inquiry
units may be special typewriters, paper tape units, or special
purpose devices. The inquiry devices may actually be other
"satellite" computers or processors, located at remote points.
For example, Figure 5 indicates that the UNIVAC 1004 and 1050
systems are capable of functioning as communication input and
output devices to larger UNIVAC systems if required. Their use
as communication devices in no way interferes with their utility
as "free-standing" systems.

In most instances, these special inquiry units are also capa-
ble of receiving and recording the information which was requested.
Hence, they are actually dual purpose input-output devices.

In order that the real time computer may efficiently fulfill
its mission in an information retrieval operation, it must be capa-
ble of instantaneously searching mass memory on a random ac-
cess basis, performing required computations at high speeds and
then transmitting the required output data to the remote output
devices.

183
c. Scientific Computers

The scientific computer must be capable of accepting a
modest amount of input data and then performing complex arith-
metic computations at high speeds. The ultimate output, in most
instances, is modest in volume. Since scientific programs are
generally quite complex, they require large memory capacity for
program and working storage. The characteristics of most
scientific computers are as follows:

1 l) Fast input

( 2) Extremely fast computing speeds

(3) Extremely large internal memory

(4) Modest mass memory

(5) Moderately fast output speed.

Since very few library applications fall into the pure scien-
tific category, for the purposes of this discussion we can dismiss
the scientific computer from any further consideration.

I. THE TOOLS OF THE TRADE-SUMMARY AND REVIEW

Before moving on to a discussion of where and how data
processing equipment can be used in the library system, the fore-
going comments on the characteristics of the various tools that
are available can be summarized as follows.

A. Punched Card Equipment

After reviewing the functions of the various component units
of a punched card system, it was observed that within the limits
of the transaction volume to be processed, punched card equipment
could be economically used by many library systems. The basic
limitations of punched card equipment as compared with more
powerful and sophisticated data processing systems have been
outlined.

B. High Speed Card Processors

It has been determined that the high speed card processor
"bridged the gap" between the punched card and the electronic
computer systems. It had greater input and output speed than
punched card equipment and had a reasonable amount of internal
memory. It had more powerful programming capabilities and
could calculate at electronic speeds. Furthermore, it could con-
currently read, calculate, print, and punch data, and in most in-
stances, was quite comparable in cost to punched card equipment.

C. Electronic Data Processing Equipment (Computers)

Eliminated from consideration were the special purpose and
the analog computer, and it was agreed that for library purposes

184

only the general purpose digital computer should be considered.

It was observed that general purpose digital computers, by virtue

of their basic specifications, fall into three classes; namely, the

business data processing computer, the real time computer, and

the scientific computer. It was also concluded that there were

very few pure scientific applications in the library field; hence,

there was no need for further consideration of the scientific com- ^

puter. In summary, by process of elimination, it appears that ^

further discussion of the available tools of the trade should be

limited to four classes of equipment:

1. Punched card equipment

2. High speed card processors

3. Business data processing computers (Batch processing)

4. Real time computers (Random access) .

H. POTENTIAL AREAS OF APPLICATION

To facilitate discussion of the potential areas of library
application of these four types of equipment, see Figure 6 which
lists many of the commonly accepted applications in this area.
These applications are broken into five general categories:

A. Financial accounting

B. Book processing

C. Circulation control

D. Information retrieval

E. Research and statistics

This schedule indicates the areas of application and the capa-
bilities of the various types of equipment to handle them. Generally
speaking, the major factor in the determination of the proper
equipment to be selected and used will be the size of the system
and the transaction volume to be processed.

In the areas of financial accounting, circulation control, and
research and statistics, all four types of equipment are capable
of producing desirable end results. Quite naturally, the business
data processing computer can do a more complete job than can
the punched card system; but here again, the matter of economics
must be considered.

A. Financial Accounting

None of these applications is a newcomer to data processing. |
There is not an application listed in this category that is not
presently being handled by hundreds of government, commercial,
industrial, and financial users. You will not open new horizons or
chart new courses if and when you add these applications to your
equipment, regardless of type. Any manufacturer of data process-
ing equipment can supply proven, workable procedures and

185

programs for any of these applications. They are the orthodox
*bread and butter" applications that have supported punched card
and computer installations for years past. They are all proven,
economical applications; and, as previously stated, your equipment
selection will be determined by your volume.

B. Book Processing

This area of application is, of course, unique to the library
field. Generally speaking, punched card equipment cannot accept-
ably produce catalog cards or catalogs because of its inability to
print upper and lower case characters. Upper and lower case
characters are available on the UNIVAC 1004 Card Processor and
all other UNIVAC high speed printers. Consequently, acceptable
catalog cards and catalogs can be produced on any of this equip-
ment. While it is possible to produce continuous form book labels
on punched card equipment, here again the availability of upper
and lower case characters permits the production of a more ac-
ceptable label.

C. Circulation Control

Several libraries have for years been effectively handling
borrower registration, book charging and returns, overdue notices,
and fine accounting on standard punched card equipment. The
punched card has proven to be an excellent "turn around document"
for this use. For the large library system, the ultimate, most
sophisticated approach to this application is to store the entire
borrower and book inventory file in mass memory, and automat-
ically determine due dates, prepare overdue notices, compute
fines, etc.

D. Information Retrieval

Applications in this area can only be processed on large
scale systems with mass storage and real time capabilities. The
assumption would be that the entire book inventory would be stored
in mass memory. One or more input devices could then be used
to interrogate the memory to produce the output required. Re-
quests would refer to information regarding a single title, or if
required, extensive listings could be printed out containing com-
plete data in any category.

E. Research and Statistics

Generally speaking, research and statistics are usually de-
veloped as a free by-product of other applications. For example,
circulation statistics would be developed from the same documents
or data used in the book charging and return operation. Likewise,
analyses of title and subject usage would originate from the source.

186

Vendor analysis would be produced from data already developed
in the accounts payable application.

m. MATCHING THE TOOLS TO THE JOB

The ratio of the number of library data processing installa-
tions to total library systems in the United States indicates that
libraries in general are not utilizing modern data processing
techniques to the degree that is found in governmental, industrial,
and commercial organizations. In my estimation, this is the re-
sult of several factors and conditions:

A. Too few libraries and library administrators have sufficient
knowledge of the capabilities of data processing equipment to
visualize the potential economies of their use.

B. Generally speaking, librarians are ultra- conservative about
accepting modern data processing techniques. This is obvious
when one observes the hundreds of library systems which are
still preparing payrolls, vouchers, purchase orders, etc., by ob-
solete methods which have been discarded long since by other
organizations.

C. The library system, being a quasi-public service organiza-
tion, finds it difficult to obtain the necessary operating capital
funds to finance a modern data processing system.

IV. RECENT TECHNOLOGICAL DEVELOPMENTS

Fortunately, some recent technological developments should
make it easier for the librarian to justify the investment in modern
data processing equipment. Here are a few of them.

A. The "Intermediate" Equipment

This equipment, of the high speed processor class, permits
the library system to start small and build. It gives the inex-
perienced librarian an opportunity to develop an understanding of
basic data processing techniques on equipment that is simple in
concept but efficient in its operation.

B. The Inexpensive Remote Input Device

The development of these devices permits the larger library
system to install a central computer facility and interrogate it
with a series of inexpensive inquiry devices which can be installed
at remote points.

C. The "Time Sharing" Concept

In the near future, every manufacturer will offer computer
"time sharing" services. This would permit a library to install
a modest "free-standing" system on its own premises to handle

187

its routine housekeeping operations. This same system could then
be linked by communication lines to a large, real time, mass
memory computer installed in a manufacturer's service center.

In conclusion, modern data processing equipment, systems,
and techniques have much to offer the librarian and the library
system. Efficient, economical tools are "on the shelf." Many of
the major applications have already been developed and pro-
grammed. Finally, the equipment manufacturers are ready,
willing, and hopefully, able to supply any assistance that may be
required.

188

Figure 1
THE STANDARD PUNCHED CARD

CARD FORMAT AND PUNCHING CODES

yfMt MOST PUKX I 0I.C O^ Al_l_ ACCOUNT INO OOCUMCNTI

000 M X0>«014XK0U0I<U0M0XI|I<M1<I<U1<I40M N 00 N X V00 N X X 0 X <
kkkk0i.0i.kkk k k00k kkkkk0kkk0kk V>00k000k0k k00k0
00* ** 100101 10111111011111 10010101 1011 1001 * 100

ABC oe *«** Ji KL.RN o*a* STOW wx** 13343
I •••

H N M H00M N •• •••><0X k M ><000X XU0u0l<0X0HkN I.00H * * N M I
4 >.00k0ki.k0<<00k0k00kk<.0kfc0k0i.i>kkk«kkkkk0k«kkl

;•.**•*%*•••*********•••*•••**••*•** »*****««i

The UNIVAC Punched Card: Data appears in this card as round holes in 45
vertical columns, divided horizontally into two fields, for a total of 90 columns.
Numbers are represented by a one or two hole code, letters by a two or three
hole code.

4BU-

I""— fiimun "" p...—*-.

Illllllll III 1 If

ii|iiiiiiii>|>Hii!iin»ij|iiiiiiii|iiiiiii|iiiiiiiiiiiii||||iiiininiiii

44444441 44444444 44|4444< 4 {^4444414444444414444444144444444444444444411114444444
llfRlllflllllllllllllllllllllllllllllltlllllllltllfllll

IIIIIIIIIIIII|IIIIIIII|IIIIIII|IIIIIIII||||II||||IIII4II

;t!2!!!!!!!!!!!!!||!!!!!!!!!!!!!!l!!!!!!!!l!!!!!!!l!!!!!!!!!!!!!!l!!!!!'-!!! '

The IBM Punched Card: Data appears in this card in the form of rectangular
holes in 80 vertical columns. Numbers are represented by one hole, letters
by two hole code.

189

Figure 2
UNIVAC OPTICAL SCAN AND IBM MARK SENSE CARDS

S A09 6 SM I TH W K

f u**t BUMM« I mtotn M^

UNIVAC Optical Scan Card: Marks may be made with any pencil, ball point,
or regular pen. Each card will record up to 40 columns of marked informa-
tion. The marks are converted to standard holes by the UNIVAC Optical Scan-
ning Punch.

IBM Mark Sense Card: Marks may be made with any soft lead pencil. Each
card will record up to 27 columns of marked information. The marks are
converted to standard holes by the IBM Mark Sense Reproducer.

190

Figure 3

COMPARATIVE SPEEDS, CAPACITIES, AND COSTS
UNIVAC 1004 CARD PROCESSOR VS PUNCHED CARD EQUIPMENT

COMPARISON OF SPEEDS AND CAPACITIES

Operation

UNIVAC 1004

High Speed
Card Processor

Standard

Punched Card

Equipment

Card Read

Card Summary Punch

Memory Capacity

Computing Speeds

Printing

Magnetic Tape-Read & Write

Paper Tape-Read & Write

Concurrent Read-Compute-Print-Punch

400-600 CPM

200 CPM
2000 Characters
400-600 CPM
400-600 LPM

Yes

150 CPM
100 CPM
150 Characters
150 CPM
150 LPM

COMPARISON OF COSTS-TYPICAL EQUIPMENT & CONFIGURATIONS

Equipment

IBM 407 Printer

IBM 514 Summary Punch

IBM 604 Calculator

Total Monthly Rental

$900
125
600

$1,625

150 CPM Read-Print
100 CPM Punch
100 CPM Calculate

UNIVAC 1004 Card Processor

UNIVAC 1004 Card Punch
Total Monthly Rental

$1,150

300
$1,450

400 CPM-Read,

Calculate, Print
200 CPM Punch

Figure 4
BASIC COMPONENTS

GENERAL PURPOSE DIGITAL COMPUTERS

191

AUXILIARY MASS MEMORY

Discs - Drums - Tapes
Special Devices

1 J>
^

INPUT DEVICES

MAIN PROCESSOR

Typewriter

Internal Memory

Punched Cards

Core - Drum - Thin Film

Paper Tape

Program Control

Magnetic Tape
Communication

External - Internal
Arithmetic Unit

Devices

Add - Subtract
Multiply - Divide

OUTPUT DEVICES

Typewriter
Punched Cards
Paper Tape
Magnetic Tape
H. S. Printer

Communication
Devices

Figure 5
UNIVAC DATA PROCESSING EQUIPMENT

COMPARATIVE SPEEDS, CAPACITIES AND OTHER CHARACTERISTICS

BUSINESS DATA PROCESSING EQUIPMENT
(Batch Proceaalng)

REAL TIME COMPUTERS
(Random Access)

SCIENTiriC

COMPUTER

Punched Card
Equipment

1004 11

Processor

1050 IV
Computer

UNIVAC III
Computer

418

Computer

490

Computer

1108

Computer

Card Read- Card* per Minute

150

600

900

700

600

900

Paper Tape Read - Characters per Second

400

1000

500

too

400

Maximum Capacity • vorde

150

65H

33M

65M

33M

131K

Word Site - >lta

Total Bits - (appro..)

150

65M

198M

1170M

990H

4716M

Auxiliary Mas. He»ory - Dru-a

Capacity per Drum - Million characters

132

Haxlaui Number of Unite per Controller

Average Acceaa Time - Milllaeconda

Add Tl»e - Hlcroieconda

400

3/4

tamjatm

100

300

200

150

3/4
300

High Speed Printer - Line* per Minute
Communications Eauip»ent

150

110
600

110
133H
922

110
133M
922

110
120M
600

110
12SH
922

110
120M
922

Via Data Line Terminal to:

1004

1050

U III

418

490

1108

not

192

Figure 6
ELECTRONIC DATA PROCESSING EQUIPMENT

PARTIAL LIST OF REPRESENTATIVE LIBRARY APPLICATIONS

Punched

Card

Equipment

Bullnei
ta Proce

Compute

Revenue Accounting

Inventories: supplles-equipattnt -visual aide

InventorleSrBooks: ovned «nd svailsble

location-shelf llit>

Budgetary Accounting: fund accounting
Purchasing: requisitions-purchase orderi
Accounts P«ytble: vendor re«ltt«r,cei-voucher« rtglitt
Billing: accounts rtcelv«bU-<t«c«««nti-trt«l balancing
Payroll: paychocks-registors- labor costs-parsonntl

ooh Processing Applications
Catalog Cards

Catalogs: union catalogs-book catalogs
Serial Lists: -union lists
Book Labels

rculatlon Control

took Chirglng: (lie •elntenanee-book returns
. Overdue Notices
rine Accounting
Inter-Llbrary Loans

Subject LUtlngi
. Author and Title Listings
. Book Location Lists: departmental, etc.

Aged Title Lists

etearch snd Statistics
. Circulation Statistics

Tltl- and Subject Usage

Borrower Analysis

Vendor Analysis

Reference Analysis

thar Apjllcatlons

LIBRARY APPLICATIONS OF DATA PROCESSING:
A SELECTED BIBLIOGRAPHY, 1963-64*

Compiled by James Krikelas

The scope of this bibliography has been limited to actual ap-
plications of data processing machines for the mechanization of
library routines. General survey articles have been included only if
they deal, wholly or in part, with operational or pre-operational
programs. Published material dealing with information retrieval,
such as reports of the programs at Western Reserve University and
the National Library of Medicine, has been excluded.

The increasing interest of librarians and information special-
ists in convening and discussing their problems in relation to auto-
mation has produced a number of significant meetings during the
past two years. In addition to the annual proceedings of the Clinic
held at the University of Illinois, the published reports of the Arlie
Foundation Conference* and the 1964 ALA Pre-Conference Institute
at the University of Missouri^ may be of special interest.

*Articles published in 1963 that appear in the McCormick bibli-
ography (cited below) do not appear in this list; one item (no. 13)
published in 1962 has been added.

McCormick, Edward Mack. "Bibliography of Mechanized
Library Processes." In Herbert Goldhor, ed., Proceedings of
the 1963 Clinic on Library Applications of Data Processing.
Champaign, 111., Distributed by the Illini Union Bookstore, 1964,
pp. 157-176.

iMarkuson, Barbara Evans, ed. Libraries and Automation.
Proceedings of the Conference on Libraries and Automation held at
Arlie Foundation, Warrenton, Virginia, May 20-30, 1963 under Spon-
sorship of the Library of Congress, National Science Foundation, and
the Council on Library Resources, Inc. Washington, Library of Con-
gress, 1964.

2 "Introduction to Data Processing," Library Resources &_
Technical Services, 9:5-103, Winter 1965. (These are revised ver-
sions of the papers presented at the ALA Pre-Conference Institute.)

193

194

Catalogs

1. Bromberg, Erik, Dubinski, G. A., and Remington, Bonn.
"Preparation of a Book Catalog," Special Libraries, 55:611-614,
November 1964.

2. Cline, Catherine. "Procedures for Developing Timberland's
Book Catalog." PNLA Quarterly, 28:127-132, January 1964.

3. Fasana, Paul J. "Automating Cataloging Functions in Con-
ventional Libraries," Library Resources & Technical Services,
7:350-365, Fall 1963. "

4. Henderson, John D. "The Book Catalog of the Los Angeles
County Public Library." In Herbert Goldhor, ed., Proceedings of the
1963 Clinic on Library Applications of Data Processing. Champaign,
Illinois, Distributed by the lUini Union Bookstore, 1964, pp. 18-33.

5. Highum, Clayton D. "Cataloging for Document Retrieval at
Florida Atlantic University," College and Research Libraries, 25:
197-199, May 1964.

6. Moreland, George B. "Montgomery County Book Catalog,"
Library Resources & Technical Services, 8:379-389, Fall 1964.

7. Perreault, Jean M. "The Computerized Book Catalog at
Florida Atlantic University," College and Research Libraries, 25:
185-197, May 1964.

8. Richmond, Phyllis A. "Book Catalogs as Supplements to
Card Catalogs," Library Resources & Technical Services, 8:359-365,
Fall 1964.

9. Weinstein, Edward A., and Spry, Joan. "Boeing SLIP:
Computer Produced and Maintained Printed Book Catalogs," Ameri-
can Documentation, 15:185-190, July 1964.

10. White, Herbert S. "Use of Mechanized Equipment in the
Production of Library Records for Manual Handling or Computer
Manipulation," Sci-Tech News, 18:23-26, Summer 1964.

11. Wilkinson, W. A. "A Machine-Produced Book Catalog:
Why, How and What Next?", Special Libraries, 54:137-143, March
1963.

12. Wilson, C. W. J. "Use of the Friden Flexowriter in the
Library of the Atomic Energy Research Establishment, Harwell,"
Journal of Documentation, 20:16-24, March 16-24.

195
Circulation

13. Chosen, Lawrence I., and Kodroff, Bernard. "Automation
of a Document Library," Bulletin of the Special Libraries Council
of Philadelphia and Vicinity, 28:25+, February 1962.

14. Gibson, R. W., and Randall, G. E. "Circulation Control by
Computer," Special Libraries, 54:333-338, July-Aug. 1963.

15. Havlik, Robert J. "Automatic Journal Routing Letter,"
Special Libraries, 54:175-176, March 1963.

16. Haznedari, I., and Voos, H. "Automated Circulation at a
Government R and D Installation," Special Libraries, 55:77-81, Feb.
1964.

17. Jordan, M. P. "The Electrowriters in the Vancouver Public
Library," Canadian Library, 19:419, May 1963.

18. Pizer, Irwin H., Anderson, Isabelle T., and Brodman,
Estelle. "Mechanization of Library Procedures in the Medium-sized
Medical Library: II. Circulation Records," Bulletin of the Medical
Library Association, 52:370-385, April 1964.

19. Southern, Walter A. "Information Services at Abbott Lab-
oratories," JUinois_ Libraries_, 45:493-498, Nov. 1963.

20. Weyhrauch, Ernest E. "Automation in the Reserved Books
Room," Library Journal, 89:2294-2296, June 1964.

Serials

21. Brown, Jack E. and Wolters, Peter. "Mechanized Listing
of Serials at the National Research Library," Canadian Library,
19:420-426, May 1963.

22. California. University. San Diego. Final Report: Serials
Computer Project. La Jolla, Calif., University Library and Com-
puter Center, University of California, San Diego, 1964.

23. Pizer, Irwin H., Franz, Donald R., and Brodman, Estelle.
"Mechanization of Library Procedures in the Medium-sized Medical
Library: I. The Serial Record," Bulletin of the Medical Library
Association, 51:313-338, July 1963.

24. Srygley, Ted F. "Serials Record Instruction for a Com-
puterized Serial System," Library Resources & Technical Services,
8:248-256, Summer 1964.

196

25. Vdovin, George. "The Serials Computer Project, University
of California, San Diego." ^n_Wesley Simonton, ed., Information Re-
trieval Today. Minneapolis, Center for Continuation Study, University
of Minnesota, 1963, pp. 109-118.

Other or Combined Processes

26. Bauer, Charles K. "Practical Application of Automation

in a Scientific Information Center— A Case Study." In_ Harold S. Sharp,
ed., Readings in Information Retrieval. New York, The Scarecrow
Press, Inc., 1964, pp. 569-61 IT"

27. Bauer, C. K. "Practical Applications of Automation in a
Science Information Center— A Case Study," Special Libraries, 55:
137-142, March 1964.

28. Burns, Lorin R. "Automation in the Public Libraries of
Lake County, Indiana." In Herbert Goldhor, ed., Proceedings of the
1963 Clinic on Library Applications of Data Processing. Champaign,
Illinois, Distributed by the Hlini Union Bookstore, 1964, pp. 9-17.

29. Croxton, Frederich E. "Alpha, et al," Sci-Tech News,
18:79-82, Fall 1964.

30. Croxton, Frederich E. Automation Progress at RSIC: The
Status of Alpha L Redstone Arsenal, Ala., U.S. Army Missile Com-
mand, 1963.

31. Culbertson, Don S. "Data Processing for Technical Pro-
cedures at the University of Illinois Library." Jn Wesley Simonton,
ed., Information Retrieval Today. Minneapolis, Center for Con-
tinuation Study, University of Minnesota, 1963, pp. 99-107.

32. Felter, Jacqueline W. "Initiating a Mechanized Union
Catalog for Medical Libraries in Metropolitan New York," Special
Libraries, 55:621-624, Nov. 1964.

33. Griffin, Hillis L. "Electronic Data Processing Applica-
tions to Technical Processing and Circulation Activities in a Tech-
nical Library." In_ Herbert Goldhor, ed., op. cit., pp. 96-108.

34. Griffin, Hillis L. EDP Procedures in Technical Library A
Operations. Washington, Office of Technical Services, U.S. Depart-
ment of Commerce, 1963.

35. Griffin, Marjorie. "IBM Advanced Systems Development
Library in Transition." Jn_ Herbert Goldhor, ed., op. cit., pp. 79-95.

36. Hinkle, Elizabeth. "Computers and Books at the Manned
Spacecraft Center Library, "SLA Texas Chapter Bulletin, 15:11-13, 1964.

197

37. Lipetz, Ben-Ami. "Labor Costs, Conversion Costs, and
Compatibility in Document Control Systems," American Documenta-
tion, 14:117-122, April 1963.

38. Mehler, John S. University of Alaska Library Book Budget
Accounting Program. College, Alaska, University of Alaska Library,
1963. (Multilith)

39. Minder, Thomas, and Lazorick, Gerald. "Automation of the
Pennsylvania State University Acquisitions Department." In Automa-
tion and Scientific Communications, Part 3_. Washington, American
Documentation Institute, 1963, pp. 455-459.

40. Nicolaus, John J. The Automated Approach to Technical
Information Retrieval; Library Applications. Washington, Bureau
of Ships, Department of the Navy, 1964.

41. Parker, Ralph H. "Development of Automatic Systems at
the University of Missouri Library." In_ Herbert Goldhor, ed., op.
cit., pp. 43-55.

42. Randall, G. E., and Bristol, Roger P. "PIL (Processing
Information List) or a Computer-Controlled Processing Record,"
Special Libraries, 55:82-86, Feb. 1964.

43. Sievers, P. T., and Fasana, P. J. Automated Routines in
Technical Services. Washington, Office of Technical Services, De-
partment of Commerce, 1964.

44. Van Wazer, John R., Logue, Paul, and Wilkinson, William
A. "Information Retrieval at the New Monsanto Information Center,"
Journal of Chemical Documentation, 3:174-177, July 1963.

45. Vertanes, Charles A. "Automating the School Library:

An Advance Report," Wilson Library Bulletin, 37:864-867, June 1963.

Surveys

46. Becker, Joseph. "Automatic Preparation of Book Catalogs,'
ALA Bulletin, 58:714-718, Sept. 1964.

47. Becker, Joseph. "Automating the Serial Record," ALA
Bulletin, 58:557-560, June 1964.

48. Becker, Joseph. "Circulation and the Computer," ALA
Bulletin, 58:1007-1010, Dec. 1964.

49. Melin, John S. Libraries and Data Processing— Where Do
We Stand? (University of Illinois Library School Occasional Papers,
No. 72) Urbana, University of Illinois Graduate School of Library
Science, 1964.

198

50. Stein, Theodore. "Automation & Library Systems,* Library
Journal, 89:2723-2734, July 1964.

Cost Surveys

51. Cox, James R. "The Costs of Data Processing in Univer-
sity Libraries— In Circulation Activities," College and Research
Libraries, 24:492-495, Nov. 1963.

52. Culbertson, Don S. "The Costs of Data Processing in Uni-
versity Libraries— In Book Acquisition and Cataloging," College and
Research Libraries, 24:487-489, Nov. 1963.

53. Griffin, Hillis L. "Estimating Data Processing Costs in
Libraries," College and Research Libraries, 25:400-403+, Sept. 1964.

54. Hays, R. M., and Shoffner, R. M. The Economics of Book
Catalog Production. Sherman Oaks, Calif., Advanced Information
Systems Division, Hughes Dynamics, 1964.

55. Spiro, Herbert T., and Kotin, Allen D. "A Cost Analysis
of An Automated System for the Library of Congress." In Automa-
tion and the Library of Congress. Washington, Library of Congress,
1963, pp. 27-88.

56. Voight, Melvin J. "The Costs of Data Processing in Uni-
versity Libraries— In Serials Handling," College and Research
Libraries, 24:489-491, Nov. 1963.

Experimental or Pilot Studies

57. Courtright, Benjamin. "The Library as an Inventory Sys-
tem." In Progress Report on an Operations Research and Systems
Engineering Study of a University Library. Baltimore, Md., Johns
Hopkins University, 1963, pp. 123-144.

58. Gore, Willis C. "A Study of Circulation Control Using
Electronic Data Processing Equipment." In, ibid, pp. 109-122.

Author Index

Anderson, I. T.

Bauer, C. K 26,27

Becker, J 46,47,48

Bomberg, E 1

Bristol, R. P 42

Brodman, E 18,23

Brown, J. E 21

Burns, L. R 28

Chasen, L. 1 13

Cline, C 2

Courtright, B 57

Cox, J. R 51

Croxton, F. E 29,30

Culbertson, D. S 31,52

Dubinski, G. A 1

Fasana, P. J 3,43

Felter, J. W 32

Franz, D. R 23

Gibson, R. W 14

Gore, W. C 58

Griffin, H. L 33,34,53

Griffin, M 35

Havlik, R. J 15

Hays, R. M 54

Haznedari, 1 16

Henderson, J. D 4

Highum, C. D 5

Hinkle, E 36

Jordan, M. P 17

Kodroff, B 13

Kotin, A. D 55

Lazorick, G 39

Lipetz, B 37

Logue, P 44

Mehler, J. S 38

Melin, J. S 49

Minder, T 39

Moreland, G. B 6

Nicolaus, J. J 40

Parker, R. H 41

Perreault, J. M 7

Pizer, I. H 18,23

Randall, G. E 14,42

Remington, D 1

Richmond, P. A 8

Shoffner, R. M 54

Sievers, P. T 43

Southern, W. A 19

Spiro, H. T 55

Spry, J 9

Srygley, T. F 24

Stein, T 50

University of California

San Diego 22

Van Wazer, J. R 44

Vdovin, G 25

Vertanes, C. A 45

Voight, M. J 56

Voos, H 16

Weinstein, E. A 9

Weyhrauch, E. E 20

White, H. S 10

Wilkinson, W. A 11,44

Wilson, C. W. J 12

Wolters, P 21

199

Library Index

Air Force Cambridge Research Laboratory Research Library ( 3J (43;
Atomic Energy Research Establishment, Harwell, England ( 12)

Boeing Scientific Research Laboratories Library ( 9)
Brentwood Public Schools, Long Island, N.Y. (45)
Brooklyn College Library (20)

Florida Atlantic University (5) (7) (24)

General Electric Missile and Space Vehicle Department ( 13)

IBM Advanced Systems Development and Research Library (35)
IBM Kingston Library ( 10)
Itek Corp. (37)

Johns Hopkins University (57) (58)

Lake County (Ind.) Public Libraries (28)

Lockheed -Georgia Scientific and Technical Information Center (26) (27)

Los Angeles County Public Library (4)

Medical Library Center of New York ( 32)

Monsanto Chemical Comp. Information Center, St. Louis (ll) (44)

Montgomery County (Md.) Department of Public Libraries (6)

NASA Manned Spacecraft Center Library (36)

National Reactor Testing Station Technical Library, Idaho Falls,

Idaho (33) (34)
National Research Council Library (21)

Pennsylvania State University Library (39)

Picatinny Arsenal (Dover, N.J.) Technical Information Section (16)

Redstone Arsenal (Ala.) Scientific Information Center (29) (30)

Timberland (Washington State Library) Library Demonstration (2)
Tonawanda (N.Y.) Laboratories, Linde Comp. (15) ^

U.S. Department of Navy, Bureau of Ships Library (40)

U.S. Department of Interior (Portland, Ore.) Library (l)

U.S. Library of Congress (55)

University of Alaska Library (38)

University of California, San Diego Library (22) (25)

University of Illinois, Chicago Library (31)

200

201

University of Missouri Library (41)
University of Rochester Library (8)

Vancouver Public Library (17)

Washington University (St. Louis) School of Medicine Library (18) (23)
(Thomas J.) Watson Research Center Library (14) (42)

UNIVERSITY OF ILLINOIS-URBANA

30112082902039

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Books

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Papers presented at the ... Clinic on Library Applications of Data Processing, .."

See other formats