MARCIVE: A Cooperative Automated
Library System

Virginia M. BOWDEN: Systems Analyst, The University of Texas Health
Science Center at San Antonio, and Ruby B. MILLER: Head Cataloger,

Trinity University, San Antonio, Texas.

183

The MARCIVE Library System is a batch computer system utilizing both
the MARC tapes and local cataloging to provide catalog cal'ds, book cata-
logs, and selective bibliographies for five academic libra1·ies in San An-
tonio, Texas. The development of the system is traced and present proce-
dures are described.

Batch retrieval from the MARC 1·ecords plus the modification of these
records costs less than twenty cents per title. Computer costs fo1' retrieval,
modification, and card production average six-ty-six cents per title, between
seven and ten cents per card. The attributes and limitations of the
MARCIVE system are compm·ed with those of the OCLC system.

In San Antonio, Texas, a unique cooperative effort in library automa-
tion has developed, involving the libraries of five diverse institutions:
Trinity University, The University of Texas Health Science Center at San
Antonio (UTHSCSA), San Antonio College (SAC), The University of
Texas at San Antonio (UTSA), and St. Mary's University. These institu-
tions are utilizing the MARCIVE Library System which was developed by
and for one library, that of Trinity University. The MARCIVE system is
a batch, disc oriented computer system utilizing both local cataloging and
the MARC tapes to produce catalog cards, book catalogs, selective bibliog-
raphies, and other products.

DEVELOPMENT

The Trinity University Library has been involved in library automation
since 1966.1 When the library reclassified its collection from Dewey to the
Library of Congress classification in1966, a simplified machine-readable for-
mat was developed and used for storage on computer. This format con-
tained the following bibliographic elements: accession number, call num-
ber, author, title, and imprint date. In 1969 the library decided to refor-
mat the computer data base into a MARC II compatible format in order

184 ] ournal of Library Automation Vol. 7 I 3 September 197 4

to build a data base of bibliographic records that could be the basis for all
future automated systems within the library. The resulting system,
MARCIVE, was designed jointly by the head cataloger, Ruby B. Miller,
and the library programmer, Paul Jackson, a graduate student in Trinity's
Department of Computer Science. Since in 1969 literature on completed
library automation projects was sparse, no other system was used as a guide.
The MARCIVE format was based on the designers' interpretation of the
1969 edition of the MARC manual.

The name, MARCIVE, evolved when the programmer facetiously
claimed that his format was so advanced he would call it the MARC IV
format. The computer room operating staff, ignoring the space between
the MARC and IV, combined the two, producing MARCIV. An E was
added later for ease of pronunciation.

The MARCIVE system was designed initially as a system for data stor-
age and retrieval. The UPDATE, SELECT, and ACQUISITIONS LIST
programs were operative in September 1970. The next month UTHSCSA
inquired as to the possibility of producing catalog cards as part of the
MARCIVE system. Within the brief span of three months, by January
1971, Trinity University Library produced 4,289 catalog cards and
UTHSCSA produced 1,719 catalog cards via MARCIVE. In February 1974,
the five participating libraries produced a total of 29,000 catalog cards,
with Trinity accounting for 10,740 cards.

Continued development of the MARCIVE system was delayed in 1971
by changes in Computer Center personnel and equipment. In 1972 new
programs were developed to incorporate the MARC tapes into the
MARCIVE system. The size of the MARC data base, which is now held on
three discs, was a major problem. Modifications were included to accept in-
put from magnetic tape and typewriter terminals using the APL language
as well as keypunched cards. The original restriction of the system to classi-
fications with one to three alphabetic letters followed by numbers, such as
used by LC and NLM, was modified to accept Dewey Decimal Classifica-
tion to accommodate San Antonio College. This restriction had been in-
corporated in an attempt to insure that the call number would be properly
formatted, thus simplifying retrieval in the SELECT program and group-
ing in the ACQUISITIONS LIST and UPDATE programs.

COMPUTER CONFIGURATION

The MARCIVE system is a disc oriented system which was programmed
for an IBM 360/44 using the MFT operating system. This computer
model was designed for scientific programming and was manufactured in
limited quantities. The programs were written in basic assembly language
since adequate higher level language compilers for the 360 I 44 were not
available at the Trinity Computer Center. In 1971 the programs were con-
verted to run under DOS, and in 1972 they were converted for processing
on the IBM 370/155 using the OS processing system. Since the initial pro-

MARCNE/BOWDEN and MILLER 185

grams were written in basic assembly language, the subsequent programs
have also been written this way.

MARCIVE FORMAT

The MARCIVE format is an adaptation of the MARC II format. The
definition of the MARC II format is a", .. format which is intended for
the interchange of bibliographic records on magnetic tape. It has not been
designed as a record format for retention within the files of any specific
organization ... [it is] a generalized structure which can be used to trans-
mit between systems records describing all forms of material capable of
bibliographic descriptions . . . the methods of recording and identifying
data should provide for maximum manipulability leading to ease of con-
version to other formats for various uses."2 Adaptation of the MARC II
format is common among users. An analysis by the RECON Task Force
found much variation among the use of the fixed fields, tags, indicators,
and subfields. 3 The OCLC system can regenerate MARC II records from
OCLC records although they contain only 78 percent of the number of
characters in the original MARC II record. 4

The developers of the MARCIVE system studied the MARC manual
and decided that the leader and directmy were not necessary for program
manipulation. Such information can be generated by a conversion pro-
gram. The MARC mnemonic codes were chosen instead of the numeric
ones because all bibliographic data were being coded locally and it was felt
that mnemonics would be easier to work with. The mnemonic codes are the
ones designated in the MARC manuals except that "SI" was substituted
for "SE." Rules for assigning indicators, subfields, and delimiters are those
described by MARC.

The basic structure of the MARCIVE format is illustrated in Figure 1.
The differences between MARCIVE and MARC are as follows:

1. MARCIVE's leader consists of three fields: length of disc space,
status code, and length of record. In converting MARC the following
elements of the MARC leader are incorporated in the MARCIVE
leader fields: length of disc space, status code, and length of record.

2. MARCIVE does not contain the MARC record directory, but rather
places the tags and subfield codes in front of the actual data.

3. In the conversion from MARC II to MARCIVE, fixed fields such as
date of publication are omitted.

4. All data elements in MARCIVE are treated as variable tags even
though they contain fixed field data.

5. MARCIVE uses the mnemonic code names for the input of data
rather than the numeric MARC codes. For example "MEP" is used
for coding a person as main entry rather than "llO." The mnemonic
tag names are stored in the machine format and not the numeric
MARC tags.

,,
·'

186 J oumal of Libra1'y Automation Vol. 7 I 3 September 197 4

'""d
~

....
~ 0

i5 CJ " "' .OJ "' "' .OJ § ~ s s " '""' p:; Ql "' Ql ~ CJ "' FIN FIN-Data Data Elements ..sl 0"' ~ z z <:<:: "' 0 <:<:: <:<:: 1"'1 fJP< ~ Tag .g Elements bil .g bil ~ b!l"' fl "' "' .s "' bil
C/) E-<

C/) E-< "' "' " "' p >-1 "' >-1

Length of disc space. This identifies the number of seventy-two byte blocks a record
uses. The MARCIVE records average 350 characters or three to six blocks.

Blank. This field is used by the UPDATE program.
Length of record. Identifies the actual number of characters a record .contains.
FIN tag. This is the MARCIVE control tag and must precede each record. It contains

four subfields: accession number, type of material, location of material, and call number.
Tag name. After the FIN tag, any of the MARCIVE tags may be input as long as

they conform to the proper sequence (i.e., main entry must pi·ecede title). Each tag is
followed by its subfield codes and the data elements.

Fig. 1. MARCIVE Fo1•mat St1'uctu1'e.

6. All first indicators are input except for the first indicator in the con-
tents note.

7. Most of the second indicators are not input, except for the filing in-
dicators which are included in the MARCIVE format.

8. MARCIVE adds one variable tag to the MARC format called "FIN."
It serves the function of the MARC 090 local holdings tag. The FIN
tag must be the first variable tag in each MARCIVE record and must
contain four data elements: ( 1) accession number; ( 2) type of ma-
terial code (monograph, serial, etc.); ( 3) location of material within
library (reference, reserve, etc.); ( 4) local call number.

Even though MARCIVE is not a pure MARC format, there has been an
attempt to code most of the data elements into MARCIVE. A MARCIVE
to MARC conversion is being written by one of the MARCIVE libraries
in order to merge its MARCIVE data base with a purchased MARC data
base.

MARCIVE MASTER DATA BASES

Each of the M ARCIVE users maintains a separate data base of its hold-
ings, which is called its MARCIVE master. This master file contains a com-
plete bibliographic record for each title cataloged by the library, including
MARC cataloging and local cataloging. When a library modifies a MARC
record, the modified record is recorded in that library's MARCIVE master.
The various libraries' MARCIVE masters have not been merged, although
this is being considered. Each library has prefaced all of its accession num-
bers with a unique library code just in case a merged data base is desired.

MARC-CON DATA BASE

The largest data base in the system is the MARC-Converted data base,

MARCIVE/BOWDEN and MILLER 187

hereafter referred to as MARC-CON. This data base contains only pure
MARC data that have been converted into MARCIVE machine format. No
original cataloging or local modifications of MARC are contained in the
MARC-CON data base.

MARCIVE PROGRAMS

CONVERT-This program reformats the weekly MARC tapes into the
MARCIVE machine format.

MARC-UPDATE-This program merges the weekly converted MARC
tape with the MARC-CON disc file. An index sequential ( ISAM) file con-
taining LC card number, fifty characters of the title, and the disc address
of the MARC reoord is generated. The ISAM file is in LC card number or-
der. In 1974 the MARC-CON data base filled three 3330 disc packs. There
are three tape back-up files: one file consisting of original MARC records,
one of the MARC-CON records, and a third with the ISAM file. Deleted
records and replaced records are annually purged from the MARC-CON
files. A new set of back-up tapes for the disc packs is created every three
months in order to facilitate regeneration of the disc packs should damage
occur.

MARC-LIST-This program lists MARC records in title sequence from
the tape. Once every six to eight weeks the list is cumulated and printed.
These lists are used for searching until the annual cumulation of the
NUC is received. This provides current listings of records on the MARC
tapes that are not easily available in the National Union Catalog. This list-
ing will be eliminated in 1974, when access by title to the MARC-CON data
base is available.

MARC-SEARCH-This program searches for LC numbers on the
MARC-CON file using the ISAM file. A file of the matched records is pro-
duced on tape or disc as specified along with a listing of these records. This
listing contains the MARC-CON complete bibliographic entry (Figure 2).
Although access is currently only by LC card number, access by title algo-
rithm ( 3, 1, 1) is expected in 197 4.

REPLACE-The purpose of this program is to modify MARC-CON
records to fit the needs of the individual library. These modifications can
be done automatically to all records or on a single record basis by the li-
brary. The automatic changes are specified on a control card and include
twenty-two options such as assignment of accession number, usage of Dew-
ey class number instead of LC, and changing "U.S." in subject headings to
"United States." An example of a single modification would be the chang-
ing of a series entry from t~·aced to untraced. Most MARCIVE partici-
pants use a combination of automatic and single changes. The output
from the REPLACE program may be input to all other MARCIVE pro-
grams, such as EDIT, CATALOG CARD, UPDATE, etc.

EDIT-This program verifies the format of the input. Valid tags and
subfields as well as correct sequence of tags are checked. Multiple spaces

188 Journal of Library Automation Vol. 7/3 September 1974

LIBRARY CODE

T0000100FIN AB~PA3877.A1~D5~

T0000102LCN A~?0-022854 ~

T0000104LANO A~eng~

T0000106LANT A~enggrc~

T0000108DDC A~882j.01~

T0000110MEPF A~Aristophanes.~

T0000112TILN AC~Plays;~newly translated into English VBrsB by Patrie Dickinson.~

T0000114IHP AABC~London,~New York,~oxford University Pr~ss,~1970-~

T0000116COL AC~v. ~21 em.~

T000011HPRI AbLB.0.75 (v. 1)~{$2.95 U.s.)~

T0000120SIRU A~oxford paperbacks, 216-~

T0000122NOC A~1. Acbarnians. Knights. Clouds. Wasps. Peace,1

T0000124AEPS ADE~Dickinson, Patrie,11914-1tr.~

T0000200FIN AB1ND1097.W4~M613~

T0000202LCN A173-4J7272 ~

T0000204LANO A~enq1

T0000206LANT A~engita~

T0000210MEPS A~Monti, Franco.~

T0000212TIL AC~African masks;~[translated from th~ Italian by Andrew Hale].1

T0000214IMP AABC~London,~New York,1Hamlyn,~1969.~

T0000216COL ADC~J-157 p.169 col. illus.~20 em.~

T0000218PRI A,15/-~

T0000220SIRU a~cameo~

T0000222NOG A,Translation of Le maschere africane.~

T0000224SUT Az,Masks, African,- Africa, West.,

Fig. 2. SEARCH listing of MARC-CON data.

are compressed to one, implied subfields are added, and a limited number
of punctuation marks are generated. Actual bibliographic data are not
checked so spelling errors are not detected by the program. Those titles
which do not conform to specifications are rejected and an explanatory
message is generated. A library may choose one of three forms of listings
of output: (1) Full-Edit, (2) Mini-Edit, or (3) Error-Edit. The Full-Edit

MARCIVE!BOWDEN and MILLER 189

950564 FIN,CB6950564,M,RP, QS,4,jK49T,1961;,
950564 MEPS A, KIMBER, !DIANA jCLIFFORD,
950564 TIL AC,jANATOMY AND PHYSIOLOGY, (BY> JDIANA !CLIFFORD jKIMBER
950564 <ET AL,),
950564 EDN,14 TH ED. BY jLUTIE jC, !LEAVELL, <ET AL,),
950564 IMP ABC,jN.jY.,, !MACMILLAN, <1961),
950564 COL A,779 p,,
950564 NOG,IEARLIER EDS, HAVE TITLE: !TEXTBOOK OF ANATOMY AND PHYSIOLOGY,
950564 SUT A,IANATOMY,
950564 SUT A,IPHYSIOLOGY,
950564 AEPS A,jLEAVELL, ILUTIE jC.,

FIN ATLC,cb6950564,m,rp,QS,4,K49t,1961;,MEPS A,KIMBER, Diana Clifford,TIL
AC,Anatomy and physiology,[ by] Diana Clifford Kimber [et al. ],EDN A,14 th ed.
by Lutie C. Leavell, (et al. ],IMP ABC,N.Y.,Macmillan,[1961J,COL A,779 p,NQG
A,Earlier eds. have title: Textbook of anatomy and physiology,SUT
A,Anatomy,SUT A,Physiology,AEPS A,Leavoll, Lutie c,

QS
4 KIMBER, Diana Cliffor•l.
K49t Anatomy and physiology [by] Diana
1961 Clifford Kimber (et al.] 14th ed.

by Lutie c. Leave.ll, [ et al.] N.Y.,
Macmillan, [ 1961]

6950564

779 p.

Earlier eds. have title: Textbook of
anatomy and physiology.

1. Anatomy. 2. Physiology.
I. Title. II. Leavell, Lutie c.

MARCIVE

Anatomy and physiology

ANATOMY

PHYSIOLOGY

Leavell, Lutie C

Fig. 3. Full-Edit listing from the EDIT program showing the input, keypunched data, data
in MARCIVE file format, data in catalog card format, and tracings.

lists every title processed in the following forms: the input data, the data
as retained in the MARCIVE file, the data in catalog card format, and the
tracings (Figure 3) . The Mini-Edit lists the input for each title and any
error messages. The Error-Edit lists only the input form and the error mes-
sage of those titles which do not conform to specifications.

A library might use the Error-Edit for cataloging data from the MARC-
CON file as the proofreading has occurred after the SEARCH phase.
However, for original cataloging in which all data have been input local-
ly, the Full-Edit would be most beneficial.

ACQUISITIONS LIST-This program produces listings arranged by
classification in a format suitable for printing on 8~~-by-ll-inch paper.
These listings include the following data elements for each title: main en-

190 Journal of Library Automation Vol. 7/3 September 1974

try, title, imprint, collation, call number. The items are sorted by author
within classification.

CATALOG CARD-This program produces catalog cards on one-up
stock The IBM TN train is used and the printing is eight lines to an inch.
The program has many options including number of cards produced by
type of entry, and whether or not the tracings will print on the tops of
the cards. The cards may be printed in filing order in the following arrays:

1. Shelflist arranged by call number.
2. Main entry-title-series-added entries in one alphabet arranged alpha-

betically by the first eight characters of the first word, excluding des-
ignated prefixes.

3. Subject entries in one alphabet arranged alphabetically by the first
eight characters of the first word.

The alphabetizing is intended to be a prefiling aid and not to be used as
the absolute filing arrangement since each library adapts the filing rules
for its own collection. The cards may also be printed in set order. Exam-
ples of typical catalog cards are shown as Figure 4.

BOOK CATALOG I-This is a modification of the ACQUISITIONS
LIST program in which the classification does not print. It is used to list
faculty publications.

BOOK CATALOG 11-This is a modification of the CATALOG CARD
program in which the cataloging information is printed with all blank
lines compressed. Since it is a variation of the CATALOG CARD, any
type of catalog can be created by specifying the type of entry and then
having it prefiled accordingly. Trinity produces an author-title book catalog
which is used in the interim period between the production and the filing of
the catalog cards. Thus, the public has a listing of new titles added to the
library.

BOOK CATALOG III-This program generates a book catalog ar-
ranged by classification. It is similar to BOOK CATALOG II but tracings
are not printed (Figure 5). Indexes by added entry, title, and subject are
also generated.

UPDATE-This program merges the additions and deletions to the
MARCIVE master file, producing an updated master and a listing in classi-
fication sequence of the additions. The MARCIVE master is in accession
number order. Duplicate accession numbers are rejected.

SELECT-This program generates bibliographies from the MARCIVE
master files (Figure 6). Any tag and its subfields can be searched. The out-
put can be sorted by call number, main entry, title, or any other biblio-
graphic element in MARCIVE. The program can be a powerful search
tool for a library. However, the program's method of retrieval is by com-
paring input data with each record on the file, which can be a very expen-
sive process. There have been discussions of building ISAM files for vari-
ous points of entry.

BUILD MISAM-This program builds an ISAM file to the MARCIVE

HF
5549 • .5
R3
F54

245701

MARCIVE/BOWDEN and MILLER 191

Finkle, . Robert,; B.,. 1922-... .
Assessing corporate talent; a key to
managerial manpower plann1ng, by
Robert B. Finkle and.William s.
Jones. New.York~ Wiley-
rnterscience, 1910~

ix, 248 p. illus. 22 em.

1. Employees, Rating of.
II. Jones, W1lliam s.

0 71-120702
I. Title.

MARCIVE

EMPLOYEES, RATING OF
HF
551Pl.5
R3
F54

Finkle, Robert. B., 1922-.
Assessing corporate talent; a key to
manaqerial manpower planning, by
Robett B. Finkle and William s.
Jones. New York~ Wiley-
Interscience, 1qro.

ix, 248 p. illus. 22 em.

245701

1. Employees, R~tinq of.
II. Jones, W1ll1am S.

0 71.-120702
I. Title.

MARCIVE

Assessing corporate talent
HF
5549.5
R3
F54

Finkle, ~obert, B., 1922-.
Assess1~g corporate talent: a key to
manager1al manpower planning, by
Robert B. Finkle an~ William s.
Jones. New York~ W1ley-
Interscience, 19f0.

ix, 248 p. illus. 22 em.

245701

1. Employees 1 Ratinq of.
II. Jones, W1lliam s.

0 71-120702
I. Title.

MARCIVE

Fig. 4. Sample set of catalog cards.

192 Journal of Libm1·y Automation VoL 7 I 3 September 197 4

VT
244 THE PHYSICAL examination (Video
-1 Tape). Paul Cutler, Dept. of

Medicine. San Antonio, The
University of Texas Medical School,
1972.

SO min. sd. color.

The complete physical examination
is presented and done on a patient
in an orderly precise method,
systematically covering the general
inspection of bead, neck, eyes,
ears, nose, mouth, glands, chest,
lungs, heart, abdomen, external
genitalia, rectum, and extremities.

248 MINIMUM dosage - local anesthesia
-1 (Video Tape). Astra Pharmaceutical

and Adrian Cowan, Dept. of Oral
Surgery. Houston, The University
of Texas Dental School at Houston,
1965.

45 min. sd. color.

Anatomical drawings and clinical
demonstrations are used to show
upper and lover incisor, premolar
and molar infiltrations as vell as
palatal injections. Infra-orbital,
palatine, mandibular, mental nerve
blocks together with anatomical
landmarks for each are demonstrated
clinically.

248 THE POSTERIOR cervical triangle
-2 (Video Tape). Yick Williams,

Department of Anatomy. San
Antonio, The University of Texas
Dental School, 1972.

7,1/2 min. sd. color.

The procedure for the dissection of

115

Fig. 5. Page j1·om UTHSCSA Video Tape Catalog produced by BOOK CATALOG III.

r
QH 431 1 SUBJECT GENETICS, TITLE GENETIC

G. MENDEL MEMORIAL SYMPOSIUM, 1865 1965, BRUE~N, 1965.
(SYMPOSIA CSAVJ 1. GENETICS CONGRESSES. 2.
III. PROCEEDINGS, G. MENDEL MEMORIAL SYMPOSIUM,
QH,43l 1 G66P 1 1965

PROCEEDINGS, EDITED BY MILAN SCSNA. P~AGUE, ACADEMIA, 1~66. 287 P,
GENETICS HIST. I. MENDEL, GREGCR, 1822 1864. II. SOSNA, MILAN, ED.
1865 1965.

GALTON, FRANCIS, SIR, 1822 1911. HEREDITARY GENIUS AN INQUIRY INTO ITS LAWS AND CONSEQUENCES. LnNOCN, MACMILLAN, 18~9. 390
P. TABLES (PART FOLD.). SVO. GARRISON AND MORTON, NO. 226. REYNOLDS HISTORICAL LltiPARY CATALOGUE, NG. 1593.
1. GENETICS. 2. CREATIVENESS. 05 72.

BF,418,Gl81H 1 1S6S

GARDNER, ELDON JOHN, 1909 •
12.

PRINCIPLES OF GENETICS. 4TH ED. ~EW YORK, WILEY, Cl972 527 P. ILLUS. 1. GE,ETICS. 07

QH,431 1 G226P,1S72

GATES, REGINALD RUGGLES, 1882 •
07 72.

HEREDITARY IN MAN. NEW YORK, MACMILLAN, 1930. 365 P. ILLUS. DIAGRS. 1. GENETICS, HUMAN.

QH,431,G25SH 1 1930

GIBLETT, ELOISE R. GENETIC MARKERS IN HUMAN BLOOD. PHILADELPHIA, DAVIS, 1969 629 P. 1. BLOOD CELLS. 2. PHENOTYP2.
3. PLASMA. 4. POLYMORPHISM IGENETICSl

QH 1 431,G446G,1969

GRUBB, RUNE. THE GENETIC MARKERS OF HUMAN IMMUNOGLOBINS. NEW YORK, SPRINGER VERLAG, 1970. 152 P. ILLUS. (MOLECULAR
BIOLOGY, BIOCHEMISTRY, ANU BIOPHYSICS, 9J 1. AMINO ACID SEQUENCE. 2. ANTI ANT!RODIES. 3. GAMMA GLORULIN.
4. IMMUNOGENETICS.

QW,504 1 G885G 1 1970

HAMRERT, GUNNAR. MALES WITH POSITIVE SEX CHROMATIN AN EPICEMIOLOGIC INVESTIGATION FCLLOWEO BY PSYCHIATRIC STUDY CF SEVEHTY
FIVE CASES. GOETEBORG AKADEMIFCERLAGET, 1966 98 P. 1ST. JOERGEN'S HOSPITAL, GOfTEBOFG. PSYCHIATRIC HESE4~CH
CENTRE. REPORTS, 1) 1. MENTAL DISORDERS ETIOLOGY. 2. SEX CHROMOSOME ABNCRMALJTIES.

QH 1 431,H199M,1966

HARRIS, HARRY. HUMAN BIOCHEMICAL GENETICS. CAMBRIDGE ENG. UNIV. PRESS, 1962 C1959 310 ?. 1. GE"JETICS.
z. BIOCHEMISTRY.

QH,43l,H314H 1 1S59

HARRIS, HAR~Y. PRINCIPLES OF HUMAN BIOCHEMICAL GENETICS. N.Y., AMERICAN ELSEVIER, 1970. 328 P. (fRONTIERS OF BICLGGY 1 V.
19J 1. GENETICS, BIOCHEMICAL. 2. GENETICS, HUMAN. I. HUMAN BIOCHEMICAL GENETICS.

QH 1 431 1 H314P 1 1S7C

HARRIS, MORGAN, 1916 • CELL CULTURE AND SOMATIC VARIATION. N.Y., HOLT, RINEHART & WINSTON, 1964 54 7 p. 1. GCNE"ICS.
2. TISSUE CULTURE.

QH,401 1 H315C 1 1964

Fig. 6. SELECT listing arranged by main entry. Retrieval was based on call number QH 431, subiect "genetics" and "genetics" as first word in
title.

~
::tl
a
~
~ -c:;
~
u
l;tj
z
8.
~
J-j

t""
r
l;tj
~

1-'

194 ]oU1'nal of Library Automation Vol. 7/3 September 1974

data bases by accession number. This ISAM file is then used by the
MARCIVE SEARCH program. This ISAM file is built only on demand.

MARCIVE SEARCH-This program searches for accession numbers on
the MARCIVE ISAM file. A disc file and a listing of the MARCIVE rec-
ords is generated. This file can be corrected by using the REPLACE pro-
gram, thus creating a corrected disc file which can be input into the various
MARCIVE programs.

USING THE MARCIVE SYSTEM

The original MARCIVE programs used only keypunched card input.
The cataloged data were coded with the MARC mnemonic codes and then
keypunched. The resulting decklet of cards was the input to several pro-
grams. If in the EDIT program an error was detected, the appropriate
punched card could be corrected and the decklet resubmitted to the EDIT
program, or that program could be bypassed and the decklet input into the
CATALOG CARD program. These same decldets were saved for the
monthly acquisition list. They were not stored on disc until recorded on
the MARCIVE master disc file in the UPDATE program. Although this
may seem awkward, it was easier for the library staff, who thus did not
worry about change codes and deletes, but could work with their prime in-
put, the keypunched cards.

The UTHSCSA library still uses the keypunched card method of input,
since its cataloging is based on the National Library of Medicine's catalog-
ing. Citations are manually coded with the M1ARC mnemonic codes at a
rate of one to three minutes per title. Keypunching takes approximately
five minutes per title and is much easier than typing a catalog card because
placement on the catalog card is not a consideration. The keypunched data
are input to the EDIT program and the Full-Edit listing is used for proof-
reading. Correct data are input to the CATALOG CARD program, then
saved for input to the monthly ACQUISITIONS LIST and the UPDATE
programs.

The other libraries in the MARCIVE system use the MARC-CON por-
tion of the MARCIVE system. With the addition of the MARC tapes it
became possible to bypass the coding and keypunching steps, thus. saving
both time and effort and reducing the chance of error.

At Trinity University Library when books arrive for processing, a li-
brary clerk keypunches the LC card number for each book published after
1968. These keypunched cards are then submitted to the MARC SEARCH
program. The books are matched against the SEARCH listing of MARC
data and any cataloging changes in addition to certain automatic changes
are noted on the listing. When the books on a SEARCH listing are checked
and classified, the books are sent for further processing for the shelves.
Unmatched books are held for later input to the SEARCH program.

The corrected SEARCH listing is sent to the IBM 2741 typewriter termi-
nal operator. The operator types an input file of changes to the MARC

MARCIVE/BOWDEN and MILLER 195

records, such as a series entry change. Trinity uses the following automatic
options in the REPLACE program: ( 1) the call number used is the LC
call number without the period; ( 2) a sequential accession number is gen-
erated; ( 3) the date entered on the file is added to the record; ( 4) library
location code is generated; ( 5) "U.S." and "Ct. Brit." in subject headings
are spelled out.

The change file plus the automatic options control data are transmitted
via telephone line to the computer and submitted to the REPLACE pro-
gram. The REPLACE program modifies the records on the SEARCH file
and creates a file of records which can be used by any of the MARCIVE
programs to produce catalog cards, book catalogs, or updates to the library
MARCIVE master. Since many automatic options have been included in
the REPLACE program, the correction required on MARC SEARCH rec-
ords is minimal and a great many catalog cards can be created with little
or no input required from the library.

Books published before 1968, unless one of the MARC popular titles,
must be fully coded and input by the Trinity University Library staff.
This represents about 25 percent of the books cataloged by the library,
with the percentage varying from month to month depending on the cata-
loging priorities in the department.

Trinity University Library is the only MARCIVE user that inputs via an
on-line terminal. This is a much more expensive though versatile method
of input. All of the other MARCIVE users who modify MARC SEARCH
records follow the same procedure except all input is via punched cards. ·

One potential MARCIVE user is experimenting with producing catalog
cards without any interim editing of the MARC records other than the
automatic options available through the REPLACE program. This. method
of catalog card production is quicker and less expensive and would be use-
ful for many libraries.

BENEFITS

The benefits of the MARCIVE system must be compared to the manual
system which it replaced. Clerical staff time to type tracings and call num-
bers has been eliminated. Trinity has effected major time savings by elimi-
nating the maintenance of proof slips. UTHSCSA has reduced the typical
three-week period during which the unit card was reproduced by a com-
mercial firm to one week. The production of the monthly acquisitions list
which formerly took days is now accomplished in a matter of minutes.
Production of bibliographies is also now an easy task, whereas in the man-
ual phase it was done only by manually searching and copying the card
catalog.

UTHSCSA has also utilized the MARCIVE system for the cataloging of
local products-the videotapes produced by its television department, the
citations of publications of its faculty and staff, and most recently the
cataloging of computer assisted instruction programs, a project funded by

196 Journal of Libmry Automation Vol. 7/3 September 1974

the National Library of Medicine. The MARCIVE system was used to pro-
duce book catalogs for each of the above.

COSTS
The costs of developing the MARCIVE system were borne by Trinity

University Library and the Trinity University Computer Center. There
was no outside funding for the development. No records were kept of the
computer time used to test the MARCIVE programs since all computer
time was charged to a university-wide academic budget. During the various
development periods of the MARCIVE system there were never more than
1.5 full-time employees engaged in the project. Table 1 is an estimate of
the manpower and time spent in the various phases of the system. The li-
brary paid the salary of the librarian and the Computer Center paid the
salary of the programmer. MARCIVE evolved as the result of the coopera-
tion of both departments within the university and would not have been
possible if the administration had not supported the project.

Table 1. Estimated Time and Manpower Involvement

Months

1969-1970
1971
1971-1972
1972-1973
1973-

Development 12
Hiatus 6
Conversion to DOS & OS 9
Addition of MARC capability 12
Maintenance

PRODUCTION COSTS

Librarian Programmer/ Analyst

.25

1.25
1.00
.25

Charges to a MARCIVE library are presently determined by the number
of programs that the library uses and the method of input. A program's
cost is based on CPU time, cards read, number of lines printed, and on-
line data storage. Commercial rates which reflect overhead and salaries are
used. Method of input can be keypunched cards, typewriter terminals, or
magnetic tape.

The computer costs for producing a set of catalog cards depend on the
method of input, whether the MARC tapes are searched, which EDIT list-
ing is used, and the length and number of cards produced. An additional
$0.02 per card is charged for the cost of card stock and the maintenance
of the system. In 197 4 this charge will be increased to $0.03 per card to
cover the rising costs of paper.

UTHSCSA has kept records of the cost of each computer job since 1971.
The average cost per card for 65,217 cards in finished form produced be-
tween April 1972 and August 1973 was $0.024 per card. Considering that
the average set had ten cards and a surcharge of $0.02 per card was added,
the cost for producing a set of cards was $0.44. For the average 400 titles
cataloged by UTHSCSA in each month, the MARCIVE system charge
would be .$246.00. The same input can be used to produce the monthly ac-

MARCIVEIBOWDEN and MILLER 197

quisitions list at a cost of $0.015 per title, or $6.00 per list. The addition
of a title to the MARCIVE master disc file is an additional $0.03 per
title. An average monthly bill for MARCIVE computer services is $262.00.
To this should be added the $40.00 prorated cost of the UTHSCSA key-
punch which is also used for other projects, giving a total of $302.00, or
$0.755 per title. A library assistant codes and keypunches the data as one
half of the job assignment. An average of fifteen minutes per title is in-
volved in the coding, keypunching, proofreading, and handling of data.
At a salary rate of ·$4.00 per hour, this amounts to $1.00 per title. Trans-
portation costs for delivering data average $17.00 per month at $0.10 per
mile.

Trinity University retrieves 75 percent of its cataloging data from the
MARC tapes via the SEARCH program. The per title cost of this retrieval
varies according to the number of items searched and reflects the fact that
the more records a computer processes, the less expensive the process be-
comes. During September 1973 a search of 723 LC numbers resulted in a
$0.025 computer charge per retrieved title. A search of 10 LC numbers cost
$0.05 per retrieved item. Trinity edits each retrieved title to make local
changes and to add an accession number. The average cost for this using
the REPLACE and EDIT programs is $0.12. Thus batch retrieval from the
MARC tapes combined with the modification of these records costs from
$0.145 to $0.17.

THE MARCIVE SYSTEM AND THE OCLC SYSTEM

It is useful to compare the attributes and limitations of the MARCIVE
system with those of the Ohio College Library Center ( OCLC) system.
The two were developed fully independently of one another during the
same period of time. 5

OCLC is an on-line system with access to the cataloging input of its
member libraries in addition to the MARC tapes. This access is by author-
title and title algorithms in addition to LC number. MARCIVE is a batch
system with access to the MARC tapes by LC card number. Even though
it is a batch system, the libraries in the MARCIVE system enjoy high pri-
ority which enables immediate usage of the computer, and jobs are execut-
ed throughout the day. The turnaround time for jobs other than
SEARCH is one to two hours. Because of the large size of the MARC-
CON file, SEARCH programs are executed only at night, so the turn-
around could be as much as twenty-four hours.

The importance of the access to the OCLC member libraries' original
cataloging in addition to the MARC tapes has not been evaluated. OCLC
reports that 71 to 76 percent of new requests run against the MARC file pro-
duce hits. An eventual success rate approaching 100 percent was predicted. 6
It would seem that the MARC tapes would be sufficient for all but the
larger or more esoteric libraries.

Printed copy is generated from a MARCIVE search of the MARC tapes,

;.,I
~~ I ,,

198 Jou1'nal of Libm1'y Automation Vol. 7/3 September 1974

thus allowing a library which does not accept LC cataloging unaltered to
do its checking and revising off line. OCLC displays the retrieved item on
a CRT screen. Printed copy would be generated only if a special attach-
ment at additional cost could be hooked to the terminal. If a library ac-
cepted LC data as displayed, only one retrieval would be necessary. If,
however, the procedure was similar to that described by Walsh College Li-
brary, two retrievals would be made. In this procedure, some manual tran-
scribing from the OCLC CRT screen is made in order to check the entry
in the library catalog. 7 However, OCLC does not presently charge for are-
trieval with no other transaction.

The costs for catalog cards produced by OCLC and MARCIVE are sim-
ilar. Kilgour has reported that OCLC cost for formatting but not printing
catalog cards is $0.0221 using commercial rates. A printing cost of $0.0033
was also given. 8 UTHSCSA reports a cost of $0.024 for the average catalog
card, including formatting and printing. OCLC charges a fixed price of
$0.035 per catalog card. 9 MARCIVE takes the commercial cost of produc-
ing the cards and adds an additional $0.02 per card for card stock and sys-
tem maintenance, thus causing the average card to cost $0.044. In 197 4
the surcharge will increase to $0.03 per card to cover the increased cost of
card stock, thus causing the average card to cost $0.054.

It is in retrieval charges that a significant difference occurs between the
MARCIVE and OCLC systems. It has been difficult for the authors to
know the precise costs of OCLC to participants as these vary according to
the structure of the agreement and the location. We have chosen to com-
pare the MARCIVE costs to those of the IUCjOCLC system which has re-
cently negotiated a contract. These costs do differ in structure from the
OCLC member libraries.

In the agreement between OCLC and the Interuniversity Council ( IUC)
of the North Texas Area, the charge for service is based upon the calls
made upon the OCLC system for catalog ca1'd p1'oduction whe1'e the cata-
loging data 1'equested m·e found within the OCLC data bank. Such a call,
referred to as a "hit," is charged at the rate of $0.875. To this must be
added such items as leased line costs, terminal hardware and maintenance
costs, local training and administrative costs, etc. The total of these costs
is approximately ·$1.70 per hit. 10 No charge is made for retrieval from the
OCLC system unless catalog cards are requested, or for material input by
the requesting library. If a library were to catalog 1,000 books using re-
trieved data from the OCLC system, the charge would be $1,700 plus the
cost of catalog cards and postage. If a library were to catalog an additional
200 titles not found on the OCLC data bank and input these titles into
OCLC, the charge for the 1,200 titles would still be $1,700 plus the cost of
catalog cards and postage.

In the MARCIVE system there is a charge each time the MARC tapes
are accessed, regardless of whether catalog cards are produced. Retrieval
from the MARC tapes via the MARCIVE system costs between $0.025 and

MARCIVE!BOWDEN and MILLER 199

$0.05 per title, depending upon the quantity and percentage of matches in
a hatched search. Costs vary according to the mode of input from the
$80.00 monthly rental for a keypunch machine to the $120.00 rate for an
on-line typewriter terminal. Using the maximum $0.05 per title retrieval
cost and the ·$120.00 terminal cost and adding another $0.12 for the
REPLACE and EDIT programs, the cost for 1,000 titles is $290.00 plus on-
line data costs. Using the minimum $0.025 per title and $80.00 for the key-
punch machine and adding the $0.12, the cost for 1,000 titles is $225~00.

Although all the factors which account for this charge variation are not
known, the following appear to be significant. First, the costs inherent in
an on-line system may be higher than in a batch procedure. Second, the
size of the OCLC data base, which in 1972 had 181,209 member records in
addition to the 229,807 MARC records, may increase search time and thus
costs. Third, overhead costs of the OCLC staff and their developmental
projects in other areas may be a factor, although the substantial grant sup-
port OCLC has received should have offset some of such costs.

MARCIVE presently serves five libraries within a twenty-minute driving
area that are responsible for picking up their own output. MARCIVE has
minimal overhead costs. The present pricing structure of the Computer
Center has covered its overhead costs. All new development is separately
funded. Mrs. Miller's salary has been totally absorbed by Trinity Universi-
ty Library even though she is the MARCIVE liaison for the Computer
Center in addition to her responsibility as head of Trinity Library's cata-
loging department. It is probable that if MARCIVE were to expand the
number of participating libraries, additional administrative support would
be required. The present pricing structure used by MARCIVE would then
be reevaluated.

CURRENT DEVELOPMENTS

Continued development of the MARCIVE system is in progress. One of
the MARCIVE libraries is programming a MARCIVE to MARC conver-
sion which will allow it to merge its MARCIVE holdings with a purchased
MARC data base. Programming is underway to create book labels. Addi-
tional access indexes to the MARC files are also being programmed.

Some procedures which are feasible now were not feasible when the
original programs were written. For example, it would be logical for the
UPDATE program to generate both an updated master and an ISAM file,
instead of the latter requiring the BUILD MISAM program. The CATA-
LOG CARD program will be rewritten for output on two-up card stock
using the ALA print train. A union catalog and a joint circulation system
for the participating libraries are also possibilities.

In January 1974 the MARCIVE Users' Group was formed as a special
interest group of the Council of Research and Academic Libraries. Its
purposes are: ( 1) to share information among users regarding problems,
procedures, and needs pertaining to the MARCIVE Library System; (2)

200 Journal of Librm·y Automation Vol. 7/3 September 1974

to establish guidelines, standards, and manuals for input, output, and re-
porting procedures of user libraries; ( 3) to maintain a sound financial pol-
icy for the MARCIVE system; ( 4) to explore and develop new ideas for
programs, techniques, and procedures to further enhance the MARCIVE
system; ( 5) to recommend changes and/ or new programs for the
MARCIVE system; and ( 6) to assist libraries in the installation of any
MARCIVE programs. A major project of the Users' Group has been the
writing of a detailed procedure manual for the MARCIVE system.

CONCLUSION

The MARCIVE system is an excellent example of library cooperation
in which the sharing of facilities, interest, and expertise has had great
benefit for all concerned. In 1971 only Trinity University and The Univer-
sity of Texas Health Science Center at San Antonio used the MARCIVE
system. With the addition of accessibility to the MARC tapes, San Antonio
College joined in 1972 and The University of Texas at San Antonio and
St. Mary's followed in 1973. One factor which made this cooperation pos-
sible was the formation in 1968 of a consortium of libraries in the San
Antonio area, the Council of Research and Academic Libraries
(CORAL). Its membership is comprised of the academic, public, and spe-
cial libraries in the area. Another factor has been the Trinity University
Library and Computer Center staffs' enthusiasm and graciousness, which
have been of great importance in fostering the desire to cooperate. Too
many projects fail because of lack of communication between partici-
pants. Such communication can be as important a consideration as econom-
ic benefit in the decision to enter into a consortium.

REFERENCES

1. Ruby B. Miller and Robert A. Houze, "New Horizons in Computer Applications
at Trinity University Library," Texas Library Journal 48:227-29, 254-55 (Nov.
1972).

2. "USA Standard for a Format for Bibliographic Information Interchange on Mag-
netic Tape," Journal of Library Automation 2:53-54 (March 1969).

3. RECON Working Task Force, "Levels of Machine Readable Records," Journal of
Library Automation 3: 124-25 (June 1970) .

4. Frederick G. Kilgour et al., "The Shared Cataloging System of the Ohio College
Library Center," Journal of Library Automation 5:157-83 (Sept.1972).

5. Kilgour et al., "Shared Cataloging System," p.159, 164.
6. Frederick G. Kilgour, "Evolving, Computerizing, Personalizing," American Li-

braries 3:146 (Feb. 1972).
7. Patricia Lyons and Margaret Northcraft, "OCLC: A User's Viewpoint," Catholic

Library World 44:265-68 (Dec. 1972).
8. Kilgour, "Evolving, Computerizing, Personalizing," p.146.
9, Ohio College Library Center, Newsletter 53:2 (Sept. 29, 1972).

10. Interuniversity Council of the North Texas Area, Memorandttm (Aug. 24, 1973).