College and Research Libraries


MICHAEL A. JENNINGS 

Optimizing Library Automation 

with a Central Dynamic Store 

Too many automated library systems are focused toward a mecha-
nized, batch-processing, card-oriented system. The technical processing 
areM of a large library (over 500,000 titles) are mainly business func-
tions applicable to a Central Dynamic Store (Data Bank) concept. 
The implementation of such a concept utilizing an on-line computer 
system is discussed with emphMis on the acquisition, serials, cata-
loging, and circulation functions of a library. 

Too MUCH has been said about the in-
creasing production of printed material 
in the world and the problems this mass 
of material will cause in our present li-
braries. Mechanization is thought to be 
the answer to this complex problem, but 
the inherent problems of mechanization 
have only added to the already existing 
complexities of the library environment. 

This paper assumes that library auto-
mation is needed and discusses a new 
approach in library automation as dis-
tinguished from library mechanization. 
complexities of the library environment. 
Library automation is the implementa-
tion of library procedures on a computer 
as opposed to punched-card equipment 
or mechanization. The approach taken in 
this paper is not new from a systems 
standpoint, but it is new in its applica-
tion to a library environment. Specifical-
ly, the type of library referred to here 
is a university library. This does not 
mean that the methods discussed are not 
applicable to other types of libraries. 

Mr. Jennings is Information Scientist and 
Library Systems Consultant with Arthur D. 
Little, Incorporated, Cambridge, Massa-
chusetts. 

The techniques of Business Informa-
tion Systems utilizing a "Data Bank" 
store are used to eliminate the single-
function machine concepts of the punch-
card systems. The library as a business 
environment is complemented by a Cen-
tral Dynamic Store (CDS) which hous-
es selected files and which supports the 
simple procedures of the library and 
does not pretend to deal with all of the 
exceptions of the library process. 

THE BusiNESS FuNCTION 

The library as a business has always 
existed, but the library profession has not 
satisfactorily approached it from this 
perspective. The librarian can be viewed 
as a businessman with an information 
service as the product he is furnishing. 
Within the library, there are common 
business functions such as purchasing, 
marketing, bookkeeping, and shipping-
receiving. These functions indicate a 
strong business attitude, and they should 
be treated as such for systems design 
simplicity. 

If these business functions are de-
fined, an approach for automating a 
business service environment can be 
taken. An initial step for definition is 

I 391 



398 I College & Research Libraries • Septe1nber 1969 

the analysis of the existing practices for 
more centralized procedures within the 
business or the library. The library, like 
many businesses, exists for the buyer or 
user; therefore, its foundation must be 
centralized for strength and yet fluid for 
diversity in order to meet the demands 
on it. 

The traditionalists have offered .a so-
lution to early business systems. This 
was the punched card. Without this ini-
tial media, large business systems would 
not have progressed to the more so-
phisticated on-line, time-sharing com-
puter systems of today and tomorrow. 
The punched card systems, however, 
are becoming dated, and more effective 
systems are now being implemented. 
From the experience learned through 
early systems, why should libraries adopt 
dated card-systems for libraries? Before 
discussing the CDS approach for li-
braries, some points should be clarified 
about the card-system in the library en-
vironment. 

THE CARD- SYSTEM LIBRARY 

A majority of the library automation 
projects (these are mechanization proj-
ects) revolve around the unit key-
punched card. 1 This approach to li-
brary automation is static. Card systems 
are not flexible for the selection and/ or 
changing of specific information. Al-
though card information is machine-
readable, the validity of the information 
is questionable because of the difficult 
verification procedures employed. These 
and other qualities made card-systems 

1 Ralph E. McCoy, "Computerized Circulation Work: 
A Case Study of the 357 Data Collection System," 
Library Resourc es & Technical Services, IX (Winter 
1965), 59-65 ; Carl C. Cox, "Mechanized Acquisition 
Procedures at the University of Maryland," College & 
Researc h Libraries, XXVI (May 1965), 232-36 ; Bruce 
W. Stewart, "Data Processing in an Academic Library," 
Wilson Library Bulletin, XLI (December 1966), 388-
95; " The Use of Data Processing Equipment in Cir-
culation Control," Library T ec hnology Reports (July 
1965), 24 p. (Widener Memorial Library, Harvard 
University Research Library at U.C.L.A., Research 
Library at the Thomas }. Watson Research Center of 
IBM Corporation). 

unsuitable for modern libraries as well as 
non-compatible with .any other library 
environment than the one for which 
they were designed. 

The mechanization of library proce-
dures utilizing the card-system is nor-
mally implemented to a specific area of 
library processing (i.e. , circulation, or-
dering, and/ or serials). Consequently, 
other areas of the libraries have suffered 
from the inherent problems of this 
mechanization. Considerations of the en-
vironmental controls and influences of 
the other areas within the library have 
not been recognized or have just been 
ignored. The area that is mechanized 
does not necessarily improve the total 
service or total organizational process of 
the library. It may, in fact, complicate 
or add needless problems to the library's 
other processing functions. 

The product of a card-system is a list. 
The list may reflect a status of material 
(i.e., on-order, charged circulation, or 
received) or financial statements for 
accounts. The list produced from cards 
is only an indication of status at the 
time the cards are processed, and does 
not reflect current or now status. The 
value of lists in a dynamic business en-
vironment is normally low. For purp<?ses 
of control on a physical inventory as 
well as the normal processes of ordering, 
receiving, etc., a card system is not the 
answer. 

The manual manipulation and stor-
age of cards has always represented a 
problem. The card, like a slip of paper, 
is a physical piece of information. It 
must be maintained, routed, and stored. 
The problems relating to a box or tray 
of cards being dropped or mutilated do 
not have to be enumerated. 

Cards can only be used as a tool to 
generate lists. Their use for conveyance 
of information to nonlibrary personnel or 
outside users is null. They are inter-
preted only on an internal library or 
processing nature. Because it is felt that 
library holdings must be accessible to 



outside users, information networks have 
been planned. 2 If this is to be a reality 
in the next few years, card systems 
will be incompatible in support of any 
information network of remote on-line 
nature. Current-status systems with ac-
cessible information to a remote user 
will be used in an information network. 
To ready a library for the eventuality 
of massive flows of information with ac-
cessible service to any user, the ap-
proach of establishing a Central Dy-
namic Store to support information net-
works must be planned. For the library 
card-systems in existence today, this 
means a repetition of initial analysis and 
planning to meet any on-line objective. 

THE CENTRAL DYNAMIC STORE 

IN THE LIBRARY ENVIRONMENT 

The CDS in a business as well as li-
brary environment is a working tool for 
the internal .and external organization. 
It is a flexible and dynamic medium 
which is as fluid as the environment de-
mands in which it is established. The 
characteristics of the CDS are expand-
ability, contractibility, compatibility, and 
.accessibility. With these characteristics, 
there is no question of its suitability in a 
library environment. 

Physically, the CDS is a large-hard-
ware, random-access memory device. 
This memory is a storehouse of any in-
formation desired to maintain the ob-
jectives for survival of the organization. 
The library business is maintained from 
a series of files of information that re-
flect the various functions or transac-
tions of a current and historical nature. 
This means the CDS system can yield 

2 U.S. F ederal Council for Science and Technology, 
Committee on Scientific and Technical Information, 
R ecomm endations for National Documents Handling 
Systems in Science and Technology, U.S. Department 
of Commerce, National Bureau of Standards, Institute 
for Applied Technology, 1965; Charles J. Austin , 
" Medlars Project at the National Library of Medicine," 
Library Resources & Technical Services, IX (Winter 
1965) , 94-99. 

Optimizing Library Automation I 399 

in-process, to be ordered, current hold-
ings, .and historical status information 
for forecasting and maintaining current 
ordering, processing, and receiving func-
tions. 

The CDS is accessible from any com-
munication-linked physical location in or 
out of the library facility. This is ac-
complished through a remote terminal 
device, preferably .a cathode ray dis-
play device. Any individual employed 
by the library can perform the various 
functions of the library process on the 
CDS without disturbing anyone else us-
ing the CDS (i.e., im1nediate access or 
no waiting). This is accomplished by op-
erating the CDS on a time-sharing com-
puter system. 

Changes to .any file or record with-
in a file can be made at any time to re-
flect new information or updated infor-
mation for any process within the li-
brary. 

In a particular library environment 
the CDS contains the major files of the 
library. The files or information from the 
files is accessed within a real-time or 
query-response mode. The response to a 
query is fast enough not to jeopardize 
any immediate need for the response. 
The information gathered fron1 the files 
depends upon the design of the files 
and the retrieval structures designed for 
the programming system. 

To .assist in the bibliographic verifi-
cation of incoming orders, searching an 
in-process or outstanding order file along 
with the shelflist initially eliminates 
any duplicate ordering. (This was found 
to represent an average of 30 to 40 per 
cent of the incoming orders at Oregon 
State University.) 

The ordering process is easily handled 
by the CDS through the introduction of 
the verified author-title and associated 
information to combine with a vendor 
file for the printing of the purchase or-
der from a CDS order file on a con-
tinuous-printed form order. The printing 
of orders is done either through a re-



400 I College & Research Libraries • September 1969 

quest to the system for immediate print, 
in the case of rush or direct orders, or 
hatched ordering by vendor, on a daily 
or weekly basis. Once the order infor-
mation is printed for mailing the original 
order information is transferred inter-
nally to an outstanding order file exist-
ing in the CDS. Because a historical-
financial file is maintained by the CDS 
for serial-ordering reference, the order 
process incorporates serials ordering as 
well. 

Serials processing has always been a 
problem in libraries because of their in-
consistencies compared with mono-
graphs. Title changes, irregular publica-
tion, and noncommunication between 
publishers and receiving agencies can 
be assumed always to exist. The CDS 
system with its flexibility in design han-
dles the majority of serial processing. 

The check-in process of serials with the 
CDS system is an update function. The 
date of issues received is recorded after 
searching the serials holding record for 
the particular issue received. The bind-
ing and claiming functions of a serial 
or cataloging area are by-products of the 
CDS system. Internal programs hold in 
check all serial records to conform to 
predetermined algorithms. These allow 
the CDS system to inform technical 
processing personnel that a claim should 
be issued for a serial or that a serial is 
ready for binding. 

Cataloging processes are a function of 
the CDS ordering process. If LC catalog-
ing is available from bibliographic veri-
fication, a catalog flag is noted when the 
title has been placed on-order. Upon re-
ceipt of the material ordered, the sys-
tem transfers all information for the 
shelflist record to the internal shelflist 
file. If a physical card catalog or book 
catalog is being maintained, generation 
of the appropriate entry is made or 
stored. If, during bibliographic verifica-
tion LC cataloging is not available, up-
dating of the shelflist, etc., is held for 
receipt and subsequent cataloging of the 

document and the files are appropriately 
updated at that time. 

The circulation function of the library 
business is accomplished without the 
standard unit card approach. 3 The shelf-
list, merged with a student personnel 
file, produces a complete charge record 
for the material. If the library CDS sys-
tem is used on a total university com-
puter system, the student personnel rec-
ords may be part of the registrar's sys-
tem. Maintaining these records would 
not be a function of the library. Over-
due notices are generated from algo-
rithms based on prearranged or stand-
ardized checkout periods for faculty , 
civilians, and graduate and undergrad-
uate students. 

Another by-product of the CDS sys-
tem is one of current operational sta-
tistics of the library. This by-product of 
the CDS is rare in today's libraries. In 
a manual environment, statistics are hard 
to retrieve because of the labor and time 
necessary for normal work loads. When 
collected, these statistics are of a his-
torical nature and are not a current re-
flection. The current statistics are gen-
erated by the CDS system in patterns 
which are used to forecast budgets, 
physical storage capacities, and addi-
tional needs for a growing library. 

Figure 1 illustrates a generalized ap-
proach of the CDS system for some of 
the processing and information needs of 
a library. With a little reflection, many 
more can be seen from the CDS con-
cept. 

CDS SYSTEM-
PROGRAMl\.UNG TECHNIQUES 

The CDS system is supported by a 
complex system of programming subsys-
tems that is an integration of fact re-
trieval and storage-retrieval techniques. 

3 Anne Flannery and James D. Mack, Mechanized 
Circulation System, L ehigh University Library (Library 
Systems Analysis, R eport No. 4 [Bethlehem, Pennsyl-
vania: Lehigh University, Center for the lnfonnation 
Sciences, November 1966]), 18 p. Appendices. 



Optimizing Library Automation I 401 

.. Serials File 

Shelf List 

Depa rtm e nt Fund File 

• Order File 

Vendor File 

Circulation File 

FIG. 1. A General CDS Library System 

To initiate an operational programming 
system or total operating system, a se-
ries of subprograms are employed to fol-
low a modular programming technique. 

Modular structuring of programming 
procedures and/ or instructions divides 
the total program into its logical parts. 
A logical part is a group of algorithms 
or instructions that performs one or more 
functions on the data or program input. 
The logical part is independent of the 
total program operation on the data. 
These logical parts that together com-
pose a total program are called mod-
ules. If for some reason a module or log-
ical part is revised or eliminated, other 
modules of the total program should not 
be affected. 

In a dynamic environment, the CDS 
system must adapt to changing environ-
ment. For example, a programming sys-
tem may normally experience changes 
because of experimentation, policy 

changes, or new design strategies. To 
meet this changing environment, the 
system designers must create programs 
that can be quickly understood and · eas-
ily modified to reflect changing condi-
tions and whose effect on related pro-
grams is understood. 

Certain approaches make understand-
ing somewhat easier: 

Modular flowcharting. This gives a 
picture of the programming logic for 
each module. The module How charts 
.are tied together to present the total pro-
gram's logic How chart. 

Program standardization. Standardiza-
tion of module symbolism, labeling, and 
descriptive documentation gives a total 
program an easier means of understand-
ing. 

Detailed documentation. A step-by-
step description of the modular routines 
and their logic is necessary for quick 
understanding. 



402 I College & Research Libraries • September 1969 

Because each modular routine is a 
separate entity, it is independent of the 
total program when writing, coding, and 
de-bugging. If modular or logical rou-
tines are intertwined as in batch-proc-
ess programming, it is very difficult to 
modify a portion of the program with-
out .affecting the whole program. Any-
one who has ever tried to modify a pro-
gram written by someone else knows the 
difficulties of dissecting and patching 
another person's programming logic. 

Modular programming incorporates 
the following methods or techniques: 

To.tal (Main) Program. The main pro-
gram executes a complete number of op-
erations that constitute a task on input 
data. The main program may be an ex-
ecutive control program or an applica-
tion program. It must make decisions 
governing the flow of all data to the 
modular routines it requires. 

Multiple-Use Routines. If the same 
sequence of instructions is used two or 
more times by the main program, these 
instructions constitute a modular rou-
tine. The modular routine after execu-
tion of its instructions must return con-
trol to the main program. 

Modulatr Routine Processing. A mod-
ular routine should not be dependent 
on any decisions made outside the rou-
tine, nor should any decisions made in-
side the routine determine processing 
outside the routine. The modular rou-
tine is responsible for all ·'housekeep-
ing" for its own existence. 

The construction of .a modular main 
program in the CDS system uses an 
.. overlay" technique and provides an ef-
ficient way for programming computer 
core memory programs. It conserves 
memory by the use of a physical part 
of memory for more than one modular 
routine. Sippi defines "overlay" as: 

The technique of repeatedly using the 
same blocks of internal storage during dif-
ferent stages of a program .... The use of 

one area in storage (core memory) to suc-
cessively store more than one different sub-
routine or program parts.4 

The main program calls the modular 
routine. After the modular routine has 
completed its sequence of instructions 
control is returned to the main progra~ 
to call the next module, etc. A nlathe-
matical description below shows this 
concept. 

A main program resides during execu-
tion in core or main memory of a com-
puter. The main program, P, is com-
posed of a series of call statements ) 

• • • Cpn• 

A list or library of modular routines, 

is stored on some auxiliary memory such 
as magnetic tape, disk file, or in sec-
ondary core storage. Each of the modu-
lar routines has a return statement 

' 
Rq, 

where 

q = n-1. 

During main program execution, each 
call statement solicits from auxiliary 
memory the desired modular routine. 
After system execution of the routine 
program execution control is returned t~ 
the main program, P, for the next call 
statement in sequence (i.e., 

C M ~M ~c M ). 
p2 35 35 p3 m 

R 
3 

Thus, a general equation for the total 
programming system (or main program) 
is 

4 Charles }. Sippl, Computer Dictionary and Hand-
book (Indianapolis: Howard W. Sams and Company, 
Inc., 1966) , p. 224. 



Given: A main program, P 
A series of modular routines, 
M1, M2 ... Mm 
A call statement, Cv 
A return statement, Rq 

The structural configuration follows: 

Auxiliary M emo1·y 

M 
1 

R 
1 

M 
2 

R 
2 

M 
3 

R 
3 

M 
m 
R 

q 

Where generally, 

P = (Cv1Mm, Cv2Mm ... CvnMm), 
m < n > m; n,m ~ 0 

and a list, L, of modular routines, 

L = (M1R2, M2R3 ... MmRq) 

Thus, 

P = Cvn (L) 

It has been found that programs writ-
ten employing the modular programming 
technique are efficient both from the 
standpoint of core memory utilization 
and of program execution times. With 
the complexity of on-line real-time sys-
tems programming, the modular concept 
furnishes comprehensive and detailed 
documentation that may be understood 
by programmers at all levels of pro-
ficiency. 

The CDS system is a new concept, 
and it is of an experimental nature to-
day. Consequently, change and modi-
fication are characteristic. The modular 

Optimizing Library Automation I 403 

technique in programming structures 
meets the changing design and is neces-
sary for the CDS system where pro-
gramming logic becomes complex. 

FILE 0RGANIZA TION 

The file structures of the CDS must 
be planned to allow a maximum of re-
trieval flexibility. This can be accom-
plished economically by fixed field for-
mats. This strategy allows any piece of 
information within any file or record to 
be addressed for retrieval by the pro-
gramming system. The retrieved param-
eter can be displayed either for re-
ferral type of information or updating 
operations. 

The fixed field format of a title rec-
ord allows future expansion of the sys-
tmn for selective dissemination of infor-
mation ( SDI) purposes. This type of 
system function is an alarm system to 
announce to personnel receipt of ma-
terial that they may have interest in 
reading. A university research environ-
ment has a high usage potential for this 
type of by-product from the CDS system. 

For standardization, the author-title 
record formats for the CDS or any li~ 
brary system should be built around the 
MARC (Library of Congress) format. 
If file structures conform to the MARC 
formats, compatibility with the new-ti-
tle magnetic tapes from the Library of 
Congress will exist. 

The files in the CDS environment are 
accessed by main entry, LC cataloging 
entry, or any other parameter of infor-
mation that is used for a look-up key. 
These parameters can be combined in 
a ~oolean logical search request. 5 Dy-
namic file structuring maintains a link-
ing organization of files that allows re-
trieval of series of titles, volumes, etc. 

5 Michael A. Jennings, " Construction of Boolean 
Search Operations in a Coordinate Indexing System," 
Proceedi ngs of the American Docume ntation Institute 
1966 Annual M eeting, October 3- 7, 1966, Santa 
Monica, Cali fornia , Adrianne Press, 1966. p. 17-27. 



404 I College & Research Libraries • September 1969 

HARDwARE CoNSIDERATIONS 

To support CDS approach, an on-line, 
time-sharing computer configuration is 
needed. This is understandably a cost 
that would stop most libraries from im-
plementation of this type of system. But 
the concept of time-sharing lends itself 
toward a community or cooperative 
ownership among a group of libraries.6 
The shared concept helps to dissolve 
the total costs in maintaining a com-
puter facility by a single library. If the 
library is in a university environment, 
the Computer Service Center of the uni-
versity may be already operating a time-
sharing system. The computer does not 
necessarily require a large core mem-
ory to support the CDS system because 
most operations are input and output 
( I~O) file processing and large compu-
tatiOn programs are not involved. 

The storage unit for this system is of 
~rimary concern. Random access storage 
IS costly today for the capacity to hold 
a total file system or CDS approach for 
a. large library environment ( 500,000 
titles) . The IBM data cell is the only 
mass storage equipment on the market 
today that approaches a reasonable stor-
age-cost-bit. For example, the CDS sys-
tem for a library of approximately 
500,000 titles must employ at least one 
data cell of maximum capacity or ap-
proximately $35,000 per year for stor-
age costs. 

Remote consoles in the form of both 
teletypes and cathode ray displays are 
needed to support the remote on-line 
inquiry and output printing-processes of 
the system. The number of terminals is 
a function of the system's environment. 
. It should be mentioned that in design-
Ing a system as discussed here, hardware 
backup should be considered. The crit-
ical importance of the service rendered 

• 6 "New England Interstate Agency Plans Regional 
L1brary Center," Scientific Information Notes, VIII 
( October-November 1966) p. 5, 6. 

by a library utilizing the CDS system 
approach must be protected with back-
up systems. If the main system were to 
fail, backup storage and master files 
should be maintained for recovery from 
a failure. Backup hardware in the form 
of magnetic tapes is one solution. Other 
solutions can be employed depending on 
the critical importance of the function-
ing environment and monies available. 

CoNCLUSIONS 

Improving manual procedures (in 
formulating the manual processes) with 
the thought of eventual automation 
must be of initial concern. The comput-
er and the CDS system is the means of 
assisting in library procedures of a 
clerical nature. By no means does this 
paper and its concepts offer a cure-all 
to all complexities of a library business. 
But, if properly implemented, the sys-
tem frees librarians from routine tasks 
in order to concentrate more fully on 
"snags" or the decision-making processes 
for which they are trained. 

The costs are never reduced through 
library mechanization or automation. A 
leveling of costs may be felt if an auto-
mation system is allowed to run for about 
five years. After this, costs could settle 
to a palatable level. Expanded user serv-
ice is a normal rationalization for high 
costs. Until deeper subject categorizing 
is employed this service can be practical 
only in a small specialized subject area 
or from the broad LC subject categories. 
Lately the expanded user service hy-
pothesis has resulted in the olde practice 
of book catalog issue. 

The notions discussed in this paper 
are not impractical from a standpoint of 
technology. Future automation projects, 
if they are to be meaningful and prac-
tical, will or should utilize the CDS con-
cept. With these considerations, the com-
patibility with any future information 
network hookup on a state or nation-
wide level can be met. • •