College and Research Libraries ROB MCGEE Key Factors of Circulation System Analysis and Design Li~rarian_s must frequ~nt.ly iudge circulation systems on the basis of w>tdely d>tsparate de~cnptwns that. make comparisons difficult. A way is needed to place varwus systems mto a common perspective or frame- work, so that their similarities and differences can be readily under- stood. This paper explains basic (and largely familiar) concepts and components that are common to manual, machine-aided, and com- 'f!uter-based systems, and documents their significance as key factors m the analysis and design of academic library circulation systems. Cost factors are not discussed. INTRODUCTION LmRARIANS ARE FREQUENTLY put into the position of judging circulation sys- tems-at professional meetings, in sales presentations, upon reading library liter- ature. The variety of systems (e.g., man- ual, machine-aided, computer-based) and the varying qualities of their de- scriptions often make understanding and comparing systems difficult, particularly since proponents of specific systems may only emphasize selected features. What is needed is an explanation of funda- mental concepts and components that provides a general framework for view- ing systems. This paper therefore at- tempts to explain key factors in the anal- ysis and design of academic library cir- culation systems. Let us begin by mak- ing several assumptions and definitions. The basic purposes of a circulation Mr. McGee is library systems analyst, University of Chicago library. This work was supported by the Library's Systems Development Office through grants from the National Science Foundation, the Coun- cil on Library Resources, and the National Endowment for the Humanities. system are to record, regulate, and con- trol the movement ( absences) of library materials from their designated locations. Consequently, a circulation system for an academic library may become an intri- cate combination of policies, procedures, data processing, equipment, and staff, the operations of which may demand a significant portion of the library budget.1 To fulfill its purposes a circulation system must contain records for items such as books, journals, and microforms that have been removed from the loca- tions indicated for them by library cata- logs. It is assumed· here that a circula- tion system contains a file of these rec- ords, called an absence file. The term absence file is preferred to charge file, since a file of this nature will inevitably hold records for items that are absent for . reasons other than library charges of items to patrons. It is important to rec- ognize this in basic terminology and con- siderations, since the cause of an item's absence determines how its absence rec- ord is processed to secure the item's re- turn. A circulation system may also contain other kinds of records. The library may maintain files of user records (e.g., a I 127 128 I College & Research Libraries • March 1972 file of registered borrowers) and item records (e.g., an accession book or com- puter-held shelflist). Different combina- tions of files like these may be used for circulation control, depending upon ali- brary's circumstances. Special user-held and item-held rec- ords such as borrower identification cards and manual or punched book cards may also be employed in a circulation system. Since these records serve as sources of borrower and book infor- mation for charges, it is convenient to call them source records. There are also transaction records which may be created for a variety of transactions that can occur between the library and its users: charging and dis- charging items, placing reserves on cir- culating books for users who have re- quested them, settling fines, etc. Charge records may be updated for subsequent transactions such as reserve requests and discharges, although some transactions such as fines handling may require new records and special files. For a library to ensure that items carried by exiting users have been properly charged, some form of transaction evidence-typically cards or slips-is required. This evidence may contain return dates upon which items are due, for the borrowers' con- venience. Together these elements furnish a set of key factors by which circulation sys- tems may be described and typed: the media and data of source records and absence records; the techniques of data transfer by which book and borrower data are combined to create charge rec- . ords; the kinds of transaction evidence; · and the existence, storage characteris- tics, and processing of library files of user, item, and absence records. This pa- per documents the significance of these and other factors in the analysis and de- sign of academic library circulation sys- tems. It has been useful in the following dis- cussion to introduce still more nonstan- dard terms for concepts which are usu- ally referenced in circulation system lit- erature by the names of particular ex- amples. The use of such terms may inconvenience some readers, but that is the cost of attempting to deal in gener- alities rather than specifics. As the large number of circulation systems described here and elsewhere indicates, a range of possibilities rather than a single an- swer exists for many aspects of circula- tion system design. It is therefore im- portant to think in terms of general concepts rather than of particular exam- ples. Different terms have thus been liberally used to supplement standard ones, to avoid conceptual limits that might otherwise be imposed. Where new or different terms are used, the familiar ones to which they are related are usu- ally mentioned as examples. INFORMATION REQUIREMENTS Two basic information needs are usu- ally considered minimum requirements for academic library circulation systems: R\ the provision of locational informa- tion about items absent from their prop- 1 er locations, and R2, the provision of processing and control information nec- essary to regulate these absences. An optional requirement is R3 , the provision upon request of information about spe- cified users' transactions. There is an- other, more general need for managerial or statistical information about circula- tion system operations which will not be analyzed here. BASIC FILE PROCESSING CAP ABILITIES Beyond updating the absence file by adding new records and changing and removing old ones to record new trans- actions and changes in absence status (e.g., recording that books have been loaned or returned or reserved by a patron), other, special capabilities are needed to satisfactorily process it for R1 and R2 requirements: the capability to -- ___ ______________________________________________________________________ _. answer locational queries with reason- ably current information, within a satis- factory response time; the capability to discharge returned items rapidly and to identify special categories and pieces and generate appropriate notifications (e.g., identify overdue books and items that have been reserved by users, and send fines bills and reserve notices) ; and the capability to identify and proc- ess records for overdue loans and for other kinds of absences periodically (e.g., to search for lost books). Even a quick comparison of how systems pro- vide these capabilities and what their methods imply for R1 and R2 perform- ance in a particular library is one way of evaluating alternative designs. ABSENCE CATEGORIES AND PROCESSING PHASES The term item absence is used to indi- cate the physical removal of a library item from its assigned location. Item absences may be classified into catego- ries like those shown in Figure 1: lost items, one-day loans, etc. Here absence categories are primarily distinguished as controlled or uncontrolled, based on ac- cessibility by call number of correspond- ing absence records. If a record for a missing item can be looked up by call number, then the library knows about or has control of the absence. Whether in- dividual absence categories are elimi- nated (rather minimized), allowed but conb·olled, or ignored defines circula- tion system scope and sometimes other library requirements as well. Consider, for example, trying to minimize some of the uncontrolled categories shown by Figure 1. Denying public access to book- stacks to reduce thefts and misshelving, or increasing staff to speed shelving of returned items may have been significant policy and cost implications. To control a category of absences through a circulation system requires that special (absence) records be cre- ated and processed for them. Depending Key Factors I 129 upon the category, these requirements may become major system design and cost factors. For example, if special lo- cations of compactly stored items are shown by circulation system records in- stead of updates to library catalogs, then a large number of records may be added to absence files for an indefinite time. The penalties of creating a special ab- sence file for storage items, or of en- larging existing files, may become cru- cial when file storage and processing costs or system response times are si a- nificantly increased. 1 To exercise R1 control requires ac- cessing absence records by locational keys of missing items-in the example of Figure 1, by finding , records through call number lookups. To have R2 control requires that each absence record be periodically retrieved and processed according to its age and membership within a given absence cate- gory. When an absence record is created , some time factor such as current date or transaction due date is usually r~corded. Within each category records may there- fore be subcategorized according to their time factor associations, or ages. Thus at any point in time each subcategory may be scheduled for a unique processing action: some students receive first over- due notices while others receive second notices; some missing books are de- clared lost and are reordered, while others are simply searched for again. Because it is generally expected that absent items can be returned or found , the circulation system may actively seek these returns by periodically retrieving and processing absence records through such actions as these. Multiple process- ing phases are generally prescribed for each absence category, on the order of: ( 1 ) first record retrieval and processing phase (e.g., first overdue notice, first follow-up search for a missing item, first interdepartmental notice to preser- vations department about an item sent there for repair, etc. ) ; ( 2) one or more Some Possible Absence Categories* Controlled (by absence records accessible through call number) circulation system Rl file non-charges lost items reordered, lost items missing items being traced to individuals 1-day loans overnight loans student loans faculty loans charges loans of reserved items other files compact storage files binding records for items in process to nonindividuals photoduplication binding interlibrary loan acquisitions cataloging reserve department Uncontrolled (not represented by records accessible through call number undetected missing items stolen items misshelved items specially located items whose whereabouts are not shown by library catalogs uncharged items in use within library · discharged items which have not folios microfilm new books on display been reshelved 0 These examples illustrate p ossible a b sen ces, a nd a re not intended to represent those of a n y sp ecific libra ry. F ig. 1 ~ ~------~----------- ' ... Given: 1. A category of absences for "regular loans," which are made daily and fa ll due exactly one week from given charge dates. 2. A five -day week. 3. A seri es of R2 p rocessing phases ( A, B,C ) for items not returned b y their scheduled due d ates, the first phase of which occurs one week after a given item's due date, and the otl1er ph ases of which occur at one-week int ervals thereaft er. DAY Week 2 Week 3 Week 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 LOAN PERIOD I I 1 A A2 -LOAN PERIOD A3 LOAN PERIOD A4 -LOAN PERIOD As -LOAN PERIOD - LOAN PERIOD LOAN PERIOD LOAN PERIOD I LOAN PERIOD LOAN PERIOD LOAN PERIOD LOAN PERIOD Week 4 16 17 18 19 20 1 B B2 - B3 - B4 - Bs - A6 - A7 - As - A9 - AlO - - Week 5 21 22 23 24 25 1 c c2 - c3 c4 - cs - B6 - B7 - B8 - B9 - Bl - All - 12 A - 0 11 ~ 12 c The n umb er of subcategories extant within a single absence category depends upon the maximum lengtl1 of each absence subcategory ( in this case 2 1 days), and the int ervals at which new subcategories are begun (here the int ervals all equal 1 day). It can be shown for this case that the maximum number of absence subcategories is the length ( 21 days) divided by the interval ( 1 day), plus 1. Thus at any t ime after day 21, twenty-two absence subcategories may be extant. Each subcategory goes through three processing phases ( A,B,C), so tl1at d aily after day 21, three separate H~ retrievals mu . t be made to process one absence subcategory each through one of tl1ree phases. Fig. 2 Absence Subcategories and Processing Phases 132 I College & Research Libraries • March 1972 subsequent phases (e.g., send a second overdue notice and then a bill for a lost book, conduct a second follow-up search for missing item, etc.); and ( 3) a termi- nal phase (e.g., order a replacement copy for an unreturned item, or update library catalogs to show its permanent loss). The number of processing phases and the nature of actions taken during each depend of course upon the category of absence records being processed. The point is that items become absent for different reasons, and that the processes necessary to secure their returns vary accordingly. Enforcing a processing schedule and especially .a terminal phase will ensure that absence records do not remain in files indefinitely, and that ap- propriate remedial action will be taken. This benefits delinquent borrowers and those with reserves on absence items, and may simplify .and reduce circulation system R 2 activities. Within a given absence category each absence subcategory undergoes its own series of processing phases. The number of subcategories is not necessarily re- lated to the number of processing phases prescribed for a category. Figure 2 illus- trates this. Specifications of absence categories and subcategories and processing phases give a basic definition of circulation R2 requirements, since the number of re- quired R2 retrieval categories is equal to or greater than the maximum number of absence subcategories extant at any time. That is, the system must be able to retrieve and process each subcategory of absent records. It should be realized that the number of processing phases and the number of subcategories for a single .absence category may vary through time. Their dynamic nature may be attributed to changing library use patterns and circulation policies, in- creased processing loads, etc., which em- phasize the importance of flexible sys- tem R 2 capabilities. How or if present and future R2 requirements for a speci- ned set of absence categories can be met is critical to circulation system design. This is especially important for manual systems, where record encoding methods and file review techniques may effective- ly limit the frequency of retrieval passes on large files, as well as the number of absence subcategories which can be sep- arately retrieved during a single file pass. CIRCULATION SYSTEM RECORDS Circulation system records are cre- ated and . accessed to control users' priv- ileges and item absences. Two general types of users' privileges are the priv- ilege to enter the library and restricted areas within, and the privilege to initi- ate transactions. Records of item absences may be ac- cessed for several purposes. Accesses by primary item identification key (usually call number) are made to satisfy R1 re- quirements for locational information. Periodic retrievals of absence subcate- gories by attributes such as due dates are made for R 2 processing. Accesses by a user identification key such as name or identification number are needed to pro- vide information about specified users' transactions. Records of charge . transactions be- tween the library and its users are cre- ated by combining user data, item data, and time data. Media (source records) and methods by which user data and item data may be initially presented to the system are indicated by Figures 3 and 4, which distinguish between the active participation of users in making their identifications, and the passive na- ture of items is being identified. The process of identifying users to the system may be designed for differ- ent levels and qualities of user participa- tion. The user's role may range from simply presenting an identification card to writing or speaking varying quantities of information (e.g., full data such as name, address, telephone number; or partial data such .as an identification alphanumeric data embossed/debossed printed data that are also machine- readable optically readable fonts magnetic-ink character recognition fonts Possible Methods of User Identification identification card machine-readable coded data punched optically readable bar coded optically various combinations of alphanumeric and coded data readable (mark-sense) codes magnetically readable no identification card unverified user stipulation of identity user stipulation of identity and system confirmation (e.g., use of passwords) staff recognition emerging machine-aided identification systems fingerprint recognition voice recognition Except for "staff recognition," each of these methods of user identification requires initial or responsive user action. It is important to recog- nize which user card data may or may not be visually readable or otherwise meaningful to users. Fig. 3 item cards machine-readable punched magnetically coded optically readable mark sense codes readable fonts Methods of Item Identification machine-aided replication of item record data removal of manual item card(s) for duration of item absence reprographic transfer embossed/debossed transfers machine readable item/label magnetically coded optically readable no item card reprographic transfer of item/label information photocopying manual transfer of item/label information quick copying (e.g., xerography) Item identifications are passive, in the sense that items do not initiate or enact their own identifications. Users or library employees must therefore do the implied tasks. Item card data may not be visually readable, since the same data may be printed elsewhere inside or outside the item. Fig. 4 number).: What the user must do to identify himself is critical to the sensi- tive user-system interface, and may sig- nificantly influence system performance. Similarly, the complementary roles of user and system in making item identi- fications can be variously designed for minimal user participation (e.g., present .a book for. charging) to extensive par- ticipation (e.g., fill out a transaction form). It is thus important to evaluate the method and ease with which item data can be transferred, and the extent of the user's participation beyond simply designating wanted material. TIME DATA Time factor data are made part of a transaction record to associate it with some point in time such as a transaction date or due date, or a time clock read- ing. There are two basic types of time data for lending transactions: fixed time data, which are the same for all charges within a given time frame (e.g., all books charged today are due in one month); and variable time data, which differ for charges made within a given period, according to particular transac- tion criteria such as user and item attri- butes (e.g., student loans, faculty loans, loans of unbound periodicals ) . USER CATEGORY TRANSACTION ATTRIBUTES Faculty X X X Student X X X ITEM CATEGORY LOAN PERIOD monograph X bound serial unbound serial 1-day 2-week 1-month X Fig. 5 X X X X X X Complex Loan Periods in a System Using Variable Time Factor Data X X X X Key Factors I 135 Since variable time data may be de- termined by either a user's or item's at- tributes or by a combination of both, a decision-making or computing capabili- ty is required that is not needed for fixed time data. Variable time data can thus be distinguished as simple or complex. Simple data are determined by a single transaction criterion such as a single user or item attribute: e.g., two-week student loans, one-month faculty loans, overnight loans of unbound periodicals. Complex data are determined by a com- bination of transaction criteria, as illus- trated by Figure 5. TRANSACTION EviDENCE Evidence of valid charges is required to inspect items that users carry from the library building. This follows from the basic R2 need for information to con- trol item absences. The proliferation of library security systems in recent years emphasizes the necessity and quality of transaction evidence. Two types of transaction evidence can be asserted. Simple evidence contains no more user and item data than are in- put to the system at transaction time, such as user numbers and item call numbers. Complex evidence involves user and item data other than transac- tion time inputs. For a system to re- spond with additional data for complex evidence may require special files, proc- essing, and hardware not otherwise needed. This can be particularly de- manding in online computerized systems. Consider, for example, the requirements of providing transaction time outputs of full user (e.g., name, address) and item information (e.g., call number, author, title) upon inputs of just user and item identification numbers. Real-time access to files of user and item records would be needed, in addition to terminals that can rapidly display (e.g., printout) the evidence. Think also of security systems that use special book labels or embed- ded book plates, which are switched 136/ College & Research Lib1·aries • March 1972 and detected to indicate "valid charge" or "no charge." Transaction evidence usually contains time data to remind borrowers of item return dates. Since time data are typical- ly assigned during the creation of charge records, for both the borrowers' con- venience and system R 2 needs, the inci- dence and quality of these data are not questioned or used to classify types of transaction evidence. TRANSACTION DATA STORAGE AND TRANSFER In examining how charge records are created, basic questions are how much transaction information is required in the system, and where data are stored and in what forms. Important factors are the media and data of source records, whether there are pretrans.action system- held user or item records, the amount and methods of transaction time data transfers, and the quality of transaction evidence. Table l illustrates relation- ships among data storage and transfer, and requires a somewhat detailed ex- planation. Table l is concerned first with relative quantities of transaction information- how much data must the system record for a charge, how much identification data are carried by user and item source records (for example, in punched identi- fication and book cards), how much user and item data must be input to the sys- tem at transaction time, how much in- formation must transaction evidence contain? We are dealing here with rela- tive amounts of data-partial data and complete data-since the individual re- quirements of each circulation system specify for it the exact data elements that are needed. When we speak of com- plete data we mean all the data of a given type-e.g., complete user data. The term partial data is used when only part of the complete data is stored or trans- ferred. It should be realized that what may serve as partial data in one system (e.g., the use of item call numbers to access full descriptive records of books) may be regarded as complete data in another system (e.g., the use of only the call number to describe the item- without citation of author or title). The table assumes for charge records that enough data (i.e., complete data) about users and items must be held by the system for it to send sufficient de- scdptions (e.g., telephoned or mailed overdue messages) of unreturned items to delinquent borrowers. In some sys- tems, call numbers alone may suffice as item data, whereas in other systems au- thor, title, and so forth may also be used. Given that the system must hold com- plete charge information, we can now go back and examine the possible quan- tities of descdptive data that may be distributed prior to transaction time among users, items (i.e., stored in source records), and system-held files. From here we can see how much data of each type need to be transferred into the sys- tem to create a charge record. We also know from this how much data are available for simple transaction evidence, and what further data would be in- volved in supplying complex evidence. Looking at Table l, the first column is headed "Source Record System Types," which are simply reference num- bers that are used in following para- graphs to descdbe prototype systems. The columns headed "Pre-Transaction Data Storage" show where and how much data for each system type are held by user and item source records and by the system, prior to transaction time. The symbol "N" shows that no data are held by a given component. By defini- tion partial data are inadequate for ad- dressing borrower notifications or iden- tifying items, and must therefore be matched to system-held records contain- ing complete data. Examples of source records containing complete data are magnetic user cards and punched item cards that each contain full information. Key F actm·s I 137 TABLE 1 Data Storage and Transfer for Charge Transactions Pre-Transaction Data Storage Post-Transaction Source Record System Types User Data Item Data Transaction-Time Sys-Held User- Sys- Item- Sys- Transfers Transaction Evidence Chrg Records Held Held Held Held User Data Item Data Simple Complex User Data Item Data 1 2 3 4 5 6 7 p p c c c N p c c N p N N c c N p c c N c N N N N N N N p c p c c c p p p c c c c c c c c c c c c c c c c c c c c c c c p c p c c Symbols: C: complete; N: none; P : partial ; -: either not applicable, or complex evidence is not required to hunish complete data. Printed user cards could also contain complete data. The columns headed "Transaction- Tin:w Transfers" show for each system type how much user and item data must be transferred to create a complete charge record. The columns headed "Transaction Evidence" indicate the qualities of possible transaction evi- dence, using the meanings of partial and complete data, and simple and complex evidence given previously. The following comments are made to illustrate possible influences of source records upon system design and per- formance. Type 1 is representative of systems with user cards that contain Hollerith-punched identification num- bers, and Hollerith-punched item cards containing complete item information. The only user data contained by simple transaction evidence are the identifica- tion numbers, which are regarded as partial data. More user data could be provided by implementing a special sys- tem response to furnish complex trans- action evidence. Type 2 represents sys- tems in which transaction time inputs of user and item identification numbers are made as links to complete system- held records. Using embossed/ de bossed cards as source records would yield sim- ple transaction evidence that contains partial data. In .an on-line system rna- chine-readable cards or manual keystrok- ing could be used to input identification num hers, and a system response could provide complex transaction evidence containing complete user and item data. Type 3 suggests a system with mag- netic user and item cards encoded with complete data. Another type 3 system is one that quick-copies printed user and item cards to produce a record (perhaps multiply copied to provide simple trans- action evidence) containing complete user and item data. Type 4 1·epresents a system containing partial information for users with in- valid privileges, which if matched at transaction time, triggers a system re- sponse to prohibit the charge. This fea- ture can be incorporated by manually checking a blacklist, by building this capability into an on-line system, or by adding to any circulation system special negative authorization hardware now used in some commercial applications. A type 5 system could be one using embossed/ de bossed user cards to trans- fer complete data to transaction forms , which are filled in manually with item data. A type 6 system is one in which complete user and item information is transcribed. In a type 7 system a user identification number and full item data could be transcribed to a transaction form , perhaps from a printed user card i38 I College & Research Libraries • March 1972 and the item itself. If the user number were mark-sense coded using a special form, it could be read by an optical reader to access a complete computer- held user record. FILE MEDIA AND ORGANIZATION Absence record media and file organi- zation are mutually influential system characteristics which affect the response times of file-dependent activities such as querying and discharging. They are defi- nitely related to the media and methods of data storage .and transfer, as dis- cussed above. So far the concept of an absence file has been discussed in terms of a single file. It should now be recognized that several differently ordered files or even a single file partitioned into sepanite sections may actually serve the absence file functions, which are essentially to support R1 and R2 processing. Table 2 illustrates possible combina- tions of file media, organizations, and functions. Since the R 1 function is viewed as essential to academic library circulation systems, systems which can- not provide locational information for specific, missing items upon request are not considered. This figure does not at- tempt to give comprehensive descrip- tions of possible file organizations. Rath- er, it only illustrates the significance of file characteristics by defining sample types of systems. These are explained below. Type 1 is representative of systems in which absence records are coded for due date and put into a single manual file ordered by call number. Examples of this .are some notched-card and tabbed-card systems. Examples of a type 2 system are traditional do11ble- record (e.g., two book cards, multiple- part transaction forms) systems which have an R 1 file ordered by call number, and a separate R 2 file primarily ordered by date and secondarily ordered by call number. A type 3 system could have a single file like this R2 file-a call number file partitioned into sections that each contain records with the same date due. Some R1 call number queries require multiple lookups, but R 2 identifications are simple. A type 4 example is a sys- tem with a single machine-held file that supports R\ R2, and R3 functions ( Ra is the optional requirement to provide in- formation about specified users' transac- TABLE 2 Descriptions of System Types by Absence- File Characteristics File Characteristics System Types Media Organiz ations Functions 2 3 4 5 6 manual records single file, call number order Rl. X manual records single file in call number order, Rl. X with records coded for due R~ X dates manual records single file by due date, Hl. X secondary order by call R2 X X number machine-held sequentially single file, ordered b y call Hl. X accessible records number R2 X Ra X machine-held directly various files in various orders Rl. X X accessible records R2 X X R3 X X computer printout sequential listing in call number Rt X X order compi1ter printout lists of specified users' R3 X X transactions tions). R2 and R3 outputs are made by automatic, batch processing runs that generate overdue and other notices, and lists of users' transactions. R1 quedes are answered from periodic printouts of the file in call number order. In a type 5 system transaction data are di.J:ectly processed to update a randomly accessed machine file. Printouts of the file sup- pmt R1 querying and R3 requests, as in the type 4 system. A type 6 system would be one in which R\ R2, and R3 functions are directly performed through on-line, real-time access to computer- held files. CoNCLUSION This paper has attempted to explain basic concepts and components common to many types of circulation systems. Classifications of systems according to their source records and files have been given only as illustrations. These are not comprehensive, and other criteria and Key Factors I 139 classifications are possible. To evaluate and design specific kinds of systems such as manual systems and on-line systems will require special considerations be- yond the general ones that have been made here. Nevertheless, an understand- ing of the factors that have been dis- cussed should benefit library managers and system designers alike, who in the process of circulation system develop- ment must establish requirements and choose among alternative designs. Once library requirements are known and a system design is well understood, ques- tions of managerial information and costs can then be examined. REFERENCES 1. Herman H. Fussier, A Report on the Oper- ations of the University of Chicago Libmry for the Fiscal Year 1967/68, University of Chicago Library ( 29 Aug. 1968), p.5. ]. Robert Adams, ''A Systems Study of the Circulation Department of The University of Chicago Library," unpublished M.A. diss., The University of Chicago (Aug. 1968).