13 A BOOK CATALOG AT STAN FORD Richard D. JOHNSON : Stanford University Libraries, Stanford, Calif. Description of a system for the production of a book catalog for an under- graduate library, using an IBM 1401 Computer (12K storage, 4 tape drives), an expanded print chain on the 1403 Printer, and an 029 Card Punch for input. Described are the conversion of cataloging information into machine readable form, the machine record produced, the computer programs employed, and printing of the catalog. The catalog, issued an- nually, is in three parts: an author & title catalog, a subject catalog, and a shelf list. Cumulative supplements are issued quarterly. A central idea in the depiction of entries in the catalog is the abandonment of the main entry concept. The alphabetical arrangement of entries is discussed: sort keys employed, filing order observed, symbols employed to alter this or- der, and problems encountered. Cost factors involved in the preparation of the catalog are summarized. In November, 1966, a new library opened at Stanford University. De- signed primarily to serve undergraduates, the J. Henry Meyer Memorial Library is a major addition to the libraries on the University's campus. A four-story structure with 88,000 square feet of usable space, it has shelving for 140,000 volumes and seating for 1,900 readers. The new li- brary has numerous distinctive features. One is the subject of this account -the catalog. There is no standard card catalog in the building. Instead, copies of a book catalog are situated at eighteen locations throughout the library, easily accessible to all students and staff. In addition, copies 14 Journal of Library Automation Vol. 1/ 1 March, 1968 of the catalog have been placed at other points on the campus: the main and departmental libraries, offices of academic departments, and student dormitories. The literature now contains numerous accounts on the preparation of book catalogs in libraries ( 1,2). One may question the value of yet an- other narrative, but an account of the Stanford experience is valuable for several reasons. The genesis of the Stanford book catalog has been recorded, and a follow-up describing what happened subsequently is the next chapter in the story. The book catalog experience at Stanford is now sufficiently advanced that one may recount the undertaking both in depth and breadth-from its inception, through design, implementation, and first full year of operation. Such an account can give Stanford's approach to some still unsolved problems; for example, filing order, and the inno- vations it has made. The particular environment within which the book catalog was designed was conducive to innovation, because the entire University Library system was not itself committed. Finally, the approach here employed has been eclectic, and this report can record thanks to the many individuals .and institutions whose ideas and plans have been examined for possible use in the Stanford undertaking. Of particular im- portance to this project were the example and experience of Florida Atlantic University, the Ontario New Universities Library Project at the University of Toronto, and the Columbia-Harvard-Yale Computerization Project. ORIGINS The Stanford book catalog had its ongms in 1962. During planning for an undergraduate library it was felt a catalog in book form and avail- able in many locations would have immeasurable educational benefits for the students. Particularly was it felt that the subject portion of such a catalog would prove a valuable bibliography to students in the Uni- versity ( 3). Somewhat later, when the size and proposed layout for the new library indicated the desirability of at least three complete card catalogs as an adequate guide to the collection, further emphasis was given to the possibility of a book catalog in multiple locations. A grant in 1963 to Stanford University from the Council on Library Resources, Inc., permitted a study by Robert M. Hayes and Ralph M. Shoffner on the economics of book catalog production. This investigation compared the costs of the various ways in which a book catalog can be produced ( 4). Of the methods considered, Stanford selected the com- puter to study further. The computer was chosen not only because equip- ment was already available on campus but also because of the recent introduction of an expanded print chain with the capability of printing upper and lower case letters as well as necessary diacritical marks. In the fall of 1964 Stanford undertook further study, employing the Hayes-Shoffner report as a basis but now comparing refined costs of a A Book Catalog at Stanford/ JOHNSON 15 computer-produced book catalog with costs for three complete card cata- logs in the new library, as well as costs for two shelf lists and main entries in the University Libraries' union catalog. This second study was completed in December, 1964, and University officials approved the prep- aration of a computer-based book catalog for the library when it was determined that such a catalog would prove more useful, and for a few years less expensive, than the three card catalogs ( 5). While the autumn study was in progress, cataloging of the new lj- brary' s collection began. Plans were made for three card catalogs. Al- though the card catalogs were never prepared, the planning was of con- siderable value later in establishing field and record lengths for the ma- chine record, as well as in securing general agreement on the kind of information to include and the format of the final catalog. SYSTEMS DESIGN Preliminary systems design began in January, 1965. A systems engineer from IBM guided a team of University staff composed of librarians and personnel from the Administrative Data Processing Center in the Con- troller's Office. At the outset it was recognized that the assignment to produce a book catalog for the new library did not call for consideration of the other aspects involved in the library's operation~, such as acquisi- tions, circulation and reference. But as work proceeded, efforts were con- sciously made to design a system that could be integrated into a larger system at a later date. The basic object of the preliminary systems design was to refine fur- ther the cost estimates from the study of the preceding autumn. The system as it was being designed, however, called for increased machine time and corresponding increases in cost for processing as well as for programming. In retrospect, the major achievement of the preliminary systems design was to establish the environment for a meaningful dialogue between the librarian and systems and computer personnel. When the study began, the librarian requested a system that would have involved use of a large configuration of equipment with direct-access capability. The sys- tems and computer staff approached the design with knowledge of the equipment that would be used for the project (an IBM 1401 Computer, 12K storage, 4 tape drives) and thought in terms of fixed-length records and fixed-length fields. Through a program of mutual education, the li- brarian learned of the computer and what it could do and what it could not do; and systems and computer personnel learned of the library's re- quirements and desires. There evolved the basic design for a system capable of being implemented on the equipment at hand and acceptable to the library. As preliminary systems design drew to a close, necessary equipment was ordered. The principal element was the expanded print chain for 16 Journal of Library Automation Vol 1/ 1 March, 1968 the IBM 1403 Printer, containing 100 different characters and developed earlier by Florida Atlantic University, Yale University, and the University of Toronto. In addition, appropriate modifications were made to the cen- tral processing unit of the 1401 Computer to be used in the project. For the inputting of data the IBM 026 Card Punch was selected. It was available, and there was considerable local experience in its use. A modi- fication made to it simplifies punching of one character, the word-sepa- rator character, used to designate an upper-case letter. Delivery time on the 026 Card Punch was four months. Although it was realized that the newly announced IBM 029 Card Punch would be superior for our project, delivery time on it was one year. Even before the 026 Card Punch was received in July, 1965, an order was placed for an 029. The 029 replaced the 026 in August, 1966. The 029 Card Punch, designed for use with System/ 360, was considered superior to the 026, because it is possible to punch each of the characters specified on the expanded print chain without resorting to the multi-punch key. Appropriate modifications were ordered for the 029 so that desired characters would print at the top of the punched cards. Detailed systems design was completed by June, 1965, and the system may be described in the following manner. OUTPUT The design called for four basic outputs from the system: 1. An edit list to facilitate proofing of the items converted into machine readable form. This was considered essential be- cause of print-out in upper and lower case. 2. An author & title catalog listing items under their author and title entries. 3. An alphabetical subject catalog listing items under Library of Congress subject headings. 4. A shelf list entering all items in call number order (the Li- brary of Congress classification was adopted in May, 1965), giving all tracings for a particular entry, as well as the num- ber of volumes and copies and their location in the library. A complete catalog was to be printed annually (author & title, subject, and shelf list) with cumulative monthly supplements to each. Output for the annual author & title catalog and subject catalog from the com- puter printer were to be photographically reduced, offset masters created, and fifty copies printed. The catalogs were then to be bound in reusable binders. Later it was decided to restrict use of the reusable binders to the shelf list, printed in four copies, and the supplements for the author & title and subject catalogs, to be printed in six copies, and to bind the basic annual catalog in standard book form. It was also decided to print ten copies of the author & title and subject supplements. A Book Catalog at Stanford/JOHNSON 17 It was originally proposed to divide the catalog in a slightly different manner: names (as authors and as subjects) and titles in one section, and topical subjects in the other. Although this seemed to have consider- able logical value, it proved impossible to implement during preliminary work with card files, given the time and staff available. Provision was also made to print the catalog in one section as a dic- tionary catalog if so desired, or on cards if the book catalog should be abandoned at a later date. INPUT To achieve the above output, the design called for four kinds of input into the system: 1. Entries for titles cataloged. A separate record was to be made for each volume or copy of a title cataloged so as to provide holdings information for the shelf list and for inte- gration into a circulation system at a later date. 2. Cross references to connect headings in the author & title catalog and in the subject catalog. In addition, the cross ref- erence format would permit the introduction of information notes into any of the catalogs. 3. Changes to entries that are in the catalog. 4. Entries for items that are on order, with a view to integrating this form of input into a larger acquisitions system at a later date. IMPLEMENTATION The systems design called for the preparation of eight different com- puter programs to transform the input into the various documents as specified above. The basic programs were written dlll'ing the six-month period of June-December, 1965. During the first part of 1966 the pro- grams were debugged and the very important change procedure pre- pared that enables revision or deletion of a record. Coincident with the preparation of the programs, library staff began in July, 1965, the input- ting of cataloging information. The expanded print chain was installed in June, 1965, and edit listings for proofing purposes were available in August. In order to test the programs and study the catalog's format, a first test catalog was prepared in January, 1966. A second test, incorpo- rating the change procedure, was undertaken in April; and a third, par- tial, test was run in June. THE MACHINE RECORD When Stanford first considered the costs of a book catalog in 1962, it was quickly discovered that the most expensive element was reproduc- 18 Journal of Library Automation Vol. 1/ 1 March, 1968 tion of the individual pages. This factor influenced many decisions in design: The more entries per page, the fewer pages and less overall expense. It became necessary then to consider which elements in a stand- ard catalog entry could be omitted or abbreviated. Decisions were fairly simple to make. The collection duplicates almost entirely material in the main research library's collection, with full bibliographical information given in that library's union card catalog. In addition, browsing is en- couraged among the open shelves of the new library. The books are readily available should further information be required. Along with the factor of cost another element appeared-the desire to make a book catalog that would be something more than reproductions of unit catalog cards. As this thought evolved, it was learned that more space could be saved in the catalog through abandonment of the unit card and main entry concept. Articles by Ralph H. Parker ( 6) and Wes- ley Simonton ( 7) were instrumental in developing this aspect of the system. · The Library was amenable to a short entry in the catalog, but the actual length was another matter. From a sampling of items cataloged, it was learned that more than 99 per cent of the entries would be less than 500 characters in length. There was considerably less certainty on maximum lengths for the individual units, or fields, composing each en- try. Computer personnel argued in favor of a fixed-length machine record in order to simplify programming, and a successful compromise was made: There was to be a fixed-length record composed of one fixed- length field and six variable-length fields. Each record is 570 characters in length. For the few catalog entries that are extremely long it is pos- sible to use two records for one catalog entry. The maximum length for any catalog entry is thus approximately 1,000 characters. It is possible to enter even longer units by dividing them into sections and entering each as an analytical entry. To speed input-output time and to conserve space on tape, the records are placed on magnetic tape in blocks of two records each. Each of the six variable-length fields in the record is individually tagged. It was learned during the preparation of a later program that it would be necessary to restrict the overall length of any one field, and it was agreed that the maximum length of any one of the variable-length fields would be 400 characters. Through a misunderstanding, the author did not realize that in tape storage an upper-case letter is equivalent to two characters, a factor not taken into account when record and field lengths were established. Fortunately, this minor error has occasioned no problem. THE MASTER TAPE RECORD The master tape record (Table 1) illustrates how all of the informa- tion appears on magnetic tape. (Figure 5 gives an example of the layout.) A Book Catalog at Stanford/JOHNSON 19 Table 1. Map of Master Tape Record. Position Type of Information 1-30: 31-35: 36-42 : 43-44: 45-46: 47: 48: 49: 50: 51: 52: 53-54: 55-57: 58-71: 72-77 : 78-83 : 84-89: 90-95 : 96-101: 102-107: Library of Congress classification Size and/or format of publication (e.g., folio, Mfilm) Volume number Part number Copy number Type (blank: monograph, no anal.; 1: monograph, anals. made; 2: serial received in unbound form; 3: serial, unbound, anals. made; 4: serial received in bound form; 5: serial, bound, anals. made; 6: analytic; 7: author-title cross reference; 8: subject cross reference; 9: item on order) Record indicator (program supplies ''1'' if there is an overflow record and "2" in second record) Special location in library (code A-Z) Change indicator (code C for revision; code D for delete) Title indicator (code T if entry desired under title) Shelf list indicator (code S if entry .is to appear in shelf list only) Year acquired (e.g., 67) Month and year reported missing (e.g., 117 for Nov. 1967). It is assumed a book will be removed from the catalog if missing more than· nine years. Future codings Address and length of main entry (Area 20) (three positions for address, three for length) Address and length of conventional title (Area 30) Address and length of title paragraph (Area 40) Address and lengh of notes (Area 50) Address and length of subject headings (Area 60) Address and length of added authors and added titles (Area 70) 108-570: Variable length fields THE FIELDS To simplify coding and keypunching, each field in the record is called an Area and numbered 10 through 70. As will be shown later, these num- bers are not transferred to tape. A description of the seven fields in each record can give a good idea of the elements included in cataloging and how the unit card/main entry concept was abandoned. Area 10 is the one fixed-length field in the record. It is 71 characters long and contains positions for call number, volume number, and copy l 20 Journal of Libra1·y Automation Vol. 1/ 1 March, 1968 number. In addition, it contains indicators for other elements: type of publication; record indicator (program supplied if there is overflow to a second record); special location in the library; change indicator; title indicator; shelf list indicator; year of acquisition; and date missing. Four- teen positions remain blank for future use. Area 20 contains the main entry, Area 30 the conventional title, Area 40 the title paragraph. The title paragraph includes: the title; author statement; edition statement; imprint, limited to publisher and date; and collation, limited to pagination. Area 50 contains notes. Subject headings are recorded in Area 60, entered one after another and separated one from another by a record mark, a symbol resembling a double dagger. Added authors and added titles are entered in Area 70, similarly separated one from another by the record mark. Only added titles are entered in Area 70. If a catalog entry is desired under title, then the title indicator is marked in Area 10. PERSONAL NAMES On the form of personal names in the catalog, it was decided to an- ticipate the Anglo-American Cataloging Rules, publication of which was imminent. In general, the title-page form of a personal author's name is used. On the one hand, this has meant a shorter record and greater sim- plicity in inputting data; on the other hand, it became necessary to main- tain a name authority file when the form adopted for the book catalog differed from that established by the Library of Congress or earlier cata- loging rules. The relator, the element that describes the relationship of a person used as an entry to the work being cataloged (e.g., ed., tr., comp., illus.), is omitted in the heading to save space. The relationship is shown in the title paragraph. A heading in the book catalog, either author or subject, is printed once before a group of titles and repeated only if the titles associated with it are continued in another column. In addition to not permitting use of the relator, the system does not permit in the author & title catalog "added" entries composed of an au- thor and a title. In standard cataloging such a technique may be used instead of a separate analytical entry. In the author & title catalog, how- ever, such a composite entry would establish a new "author" (name plus title of the work) and would file as a separate unit after all works by that author. In the subject catalog the author-title entry is permitted so that books about voluminous authors and their individual works may be better displayed. THE CONVENTIONAL TITLE The conventional title has been employed to assemble under an au- thor's name editions of a work with variant titles. Collected writings of an author, or selections, are given the conventional title [Works] or [Se- A Book Catalog at Stanford/JOHNSON 21 lections] ; Through a combination of coding and programming, they are entered first under an author's name before titles of individual works are listed. (See in Figure 8 the entry under Karen Horney for an example. ) The conventional title has meaning only as it is related to the main entry. For that reason it prints only in the catalog when preceded by the main entry. THE TITLE PARAGRAPH AND THE UNIT RECORD As summarized above, the title paragraph includes the title, author statement, edition statement, imprint, and collation. With one major ex- ception this involves the copying of, or truncation of, information present on a Library of Congress card. The exception is the author statement. As shown in a recent investigation ( 8), this element was present in but twenty-five per cent of the entries studied. Current cataloging rules per- mit in some cases the omission of the author statement when it is identical with the form used in the heading ( 9,10). These rules are based upon a cataloging system employing unit records on cards, the first element of which is the main entry. In unchanged form the author statement is used as the main entry; for added entries another heading, such as au- thor, title, or subject, is superposed on the card. In the Stanford system a new unit record was introduced. The first element of it is the title paragraph. All headings, rriain or added, are placed directly above it; and if entry under title is desired, a . title entry is made in hanging-indention form. The Stanford book catalog thus does away with the main entry con- cept completely. The necessity, or even wisdom, of setting apart one field in the machine record as main entry may be questi~ned. Why not group the main entry with the other added author entries in Area 70? There were two reasons: First, it is simpler to adapt the information from Library of Congress cataloging information if the form can be fol- lowed relatively closely. Second, we wished to allow for the possibility of printing standard catalog cards if necessary, and this would allow for a reinstatement of the standard unit card concept. A basic requirement of the system is that the author statement must be included in the title paragraph. If for any reason it cannot be listed there, then it is recorded in note position in Area 50. Although no formal study was undertaken, it was believed that works by single personal au- thors would constitute more than fifty per cent of the collection. The ad- dition of the author statement in the title paragraph for each such book could add considerable bulk to the catalog. Accordingly, through the use of record marks as coding symbols, the author statement is set off in the title paragraph for those works by single personal authors. Through pro- gramming, the author statement is suppressed when the work is entered in the catalog under the name given in the main entry; whereas it ap- pears under all "added" entries. 22 Journal of Library Automation Vol. 1/1 March, 1968 i }7({]1\_, '-. 0 ~' -oifOII j 1Y !.0. • "<• "'·><• ~~· ! :·~·;.~~~,, t: ,· DEC z Q 1966 1 1' fl ' 6'6 1 ·, . :··' ·""': '~~ ,,, ~~~ -·· ' ' ' ,. ':·.-:. }"'~:::;> ' '., • ., ..... NO. '/'---"'~' ' ' I . . ' ~ ' ? I .. · ·' .':' . ' . 1018~ Hdll~· ).;--' . ':.' _i''··-::-''~·· :;-_, . v ~·~" 101 i~- 0 ' . . ; ; r.~~- -Ickes, Harold L. • u • 1 " • u ~ ~O(]J ll .:l,'h•;j}secret diary ot Kt.rold L, I ekes, ' . ,t.-1 2 0 1, 1111:.: ' !4oiT.i- 8o6 Simon t.nd Schuster, 1954-~5- 3 v. II:J. '1' ~ ' :4.~~ :12 ~\eet.a, lc 'illae fire~ tbocuana OOOK No. OAT£ _l; I , . da;s, 1933 1936. ..e, 'l'he iulde n-t;~ 1 l1JJ : • · 11 1 c, . .,.,o.f_ atx oggh, 1936 1939 3 'Rio loun"'&- u.l.o · · : 1 • tl:Odth 1939 19'•1 1 OTH( '\ l.• O. sur&. h.._,,u,f ootS ! so[]] '<•'l-"·••,_, .. ' . { . ' V•';.,...,,,.;, J· . I llll I ~ ' u.s.~-Polltics and sover.,..ent--1933-1~' . · . ·~· Do~ I. Title . VOL. NO. u I'T. (f)I'Y s4 . ~ ' 11 J 11 J i. ·. ' l'{Jl () " . " ........ '1· .!' .:. ·. (\;,: ..... --·· · .. c.:T-·-··· OECZO~ _ ..... •· . I ·. 6J6' . ·- Fig. l. The Coding Sheet. A Book Catalog at Stanford/JOHNSON 23 The system as it has been established calls for works to be entered alphabetically by title under the heading in the catalog. In the subject catalog this means, too, that works are listed alphabetically by title under each subject heading and not by author. It was felt that this form of arrangement is quite satisfactory for a selective collection, such as the Meyer Library; and it offers the possibility of scanning a page of thirty or more entries. It has occasioned one problem-when a subject heading expresses form and not subject. For example, under "SYMPHONIES," works are arranged alphabetically by title and not by composer. CONVERSION The basic source document used is the Library of Congress catalog card. Although the card itself could be used by crossing out unnecessary information and adding other data, it was felt that a clearer document would result if the needed information were copied onto another catalog card. Examples of such cards are shown in Figure 1. (Subsequent figures depict the manipulation of the entry for the book by Thomas A. Bailey, Presidential Greatness.) As illustrated in Figure 1, an identification num- ber is assigned to each catalog card and four catalog cards are placed upon a coding board and a xerographic copy made. . The original cards are filed in a manual shelf list with the identification number as an indication that the information has been coded. The coding sheet is given to the coder, who enters Area 10 information in the blocks at the right and indicates to the keypuncher where other areas begin and what special symbols should be used. · To simplify the inputting of data and the scanning of punched cards, a special data-processing card was designed for input (see Figure 2) . Each title converted is represented by a decklet of punch cards averag- ing six. Each of the seven areas or fields begins on a separate card. Al- CN1'1oV 1. 0. !fQ tl.Ali :O ~VJS:ot\11 $ 118 fiN SOOt! ~!\ OAT I!: Oflll"!l 1. ()_ ~4/PI:IItMA....!l:!'OUJ!oi:E Ml'li!:R ~.:.y !' ~ " . 1\ /·· •o O.IIOEIIt£0 041[ OIIOt" IM*.ij" ; ,__ 1niT' ::-. ~ 0 0 0 0 D 0 DOOOODOOIIIDOOOOOOOOOOCOOODODIIIOOOODOOOOOOOOOOOOOOOOOOOOOOIDDOOOQOODO ~ '~ 1 t =. ~~Q~"~n ••Nwnn~ nuu~n~nnu~u~n~••ftdUM~u••u•l1t~li»~"~~~UMflHUMUHUhunnnnHn~n~na • 111111 I 11 I 1 I II 1111111 I 1.1 11 111 11 I 11 I 1 I 1111111 I I 11 11 11 1111 I ~ 2,0 MAIN ENTRY ~i 222222 30 222222222222222222222222222222222222222222222222222 CONVENfiON.AL TITlE• g~ 3 3 3 3 33 3333333333333333333333333313 33 33333333333 3333333333 z~ 40 TITLE Q~ 444444 444444444444444444444444444444444444444444444444444 ~~ $0 NOTE$ ~;;1 SSSSSS SS5555555555 55SS55 SSSSSSSSSSSSSSSSSSSSSSSSSSSSS6l55 c •o SUO JECT HEADINGS !ll sssiu iiiSISiiSIIIIGGii ii illiiiilllli l iillilliiilllllllll !:; TO IIII(DAUI~$-IGC£0 mm ~111117 11 I I I I I I 11111 111 111111 1J1111111JIII111111111111H 1771111111111711111111 IIIIIIIIIIIIIIIIII I IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII III IIIUIIIII I IIIIIIIIIIIIIII !9999919!9999999919999999999999!99999911119999!!99!99999!!! !!9! ! ! !! ! 1!!9911!1!9! I l 3 t I & ',.! ~~~~~~ 112 11 U I)J* ll)t IIJtJ\UU!cUan:UUOll Ublt ""31M··~·~ 0"4UUitiUSI$1'1" ~' "~Ui51~Hh1UUMIUUIMM71 n nn )4 Wltn Hltl Fig. 2. The Data Processing Card. 24 Journal of Library Automation Vol. 1/ 1 March, 1968 though this may be considered wasteful of cards and indicative of "SO- column mentality," it does have its benefits. A mistake in punching in one area requires repunching of the material in that area only, the area being the smallest unit for editing purposes. So that the cards are kept in correct order for processing, the first ten columns of the cru:d are used for identification numbers. The six-digit identification number assigned to the original card is punched in the first six columns. The first digit is the month, 1-9 being January through September and zero being October-December, one "month" of 92 days; the second and third digits are the day of the month; and the fourth through sixth a consecutive number assigned each day. It is possible thus to code 999 enhies each day. The year is omitted, because it was as- sumed data would be transferred to magnetic tape at least once a year and probably more often. The area number is punched in columns 1 and 8 and the sequence within the area in columns 9 and 10. Cataloging information begins in column 11. Figure 3 shows a decklet of cards for one title. Information in Area 10 is fonnatted somewhat differently for books on order or for cross references. To enter a book on order, the word "OR- DERED" is punched in columns 22 through 28, the date and order num- ber in columns 29 through 40, and 9 in column 57, the type indicator. ·. This information prints in the catalog as a call number, and books on order are listed first in the shelf list. To date, entering of books on order has been limited to a few sample cases only. A cross reference in the subject catalog has a "call number" composed of the first nine characters of the entry to which reference is made (punched in columns 13-21) and an eight-digit identification number (in columns 22-29). An 8 is punched in column 57, the type indicator. A cross reference in the author & title catalog has a similar "call number" composed of the first ten characters of the entry to which reference is made (punched in columns 12-21) and an eight-digit identification number (in columns 22-29). A 7 is punched in column 57, the type indicator. Subject cross references are listed in the shelf list in "call number" order following books on order; and author & title cross references follow subject cross referen.ces. The "call numbers" for cross references do not print in the catalog proper, and serve only as addresses to retrieve the cross reference from magnetic tape when a change or deletion is neces- sary. Added copies and volumes are entered by preparing Area 10 informa- tion only and punching S in column 61, the shelf list indicator. The first copy entered is never coded explicitly as copy 1, even though a second copy is being simultaneously added. The program automatically identifies the first copy entered as copy 1, and the number prints in the shelf list when another copy is added or a volume or location is shown. ~o long as cataloging information is on punch cards and not on mag- A Book Catalog at Stanford/ JOHNSON 25 netic tape, the 10-digit identification number is the device used to re- trieve the information. When the information is transferred to magnetic tape, the identification number is lost and the call number becomes the identification device. As shown in Figure 3, information in Area 40 (and in Areas 20, 30, 50, and 70 as well ) continues to successive cards. It is not actually neces- sary tb punch through column 80 before beginning a continuation card for an area, and experience indicates that corrections are simplified if blanks are left at the end of each card. The only requirement is that if 0 10'1000 1 0 0 00 1 08110001100800DOD8001000DGDGOO G~GOOOBIGODOOOOGOOGIOO nQQMBM ~ ·~Mnnn"aKnnH·~~n~ nXUM~~~U~~UUtt ~"M~U~~MMgMHUUVU~ »MUUUNftD n~n MnN ~ MAIN (HUY COHV( NTKlftAL. TI TlE> TITL E NOTE:S II 1("0 SUII..ICCT H£ /tDI,.G3 11 1 1111 1 1111 1 1111 1 111 11 11 1 1111 1 11 1 11 1 1111 1 11 1 111111 2 2 2 22 2 22 2 2 2 2 2 2 2222 222 22 2 22 22222 2 2 222 222f22 2 22222222 llll 3 J 3 3 3i 3 J 3 3 J J J 33 J 3 J l J J J J JJ J J J J J J 33 J J l J l ll lll J l J l 1 4411114 44111144114 4 4 1 44 1 1114 41 444 4 14 4 4 44 4444 4 4 14 44 I I S 5 IS IS II I I 51 I I 5 1 I IS I IS 55 5 fS I 1511151555 55 5 55 5 51 Ill AI:UO AUTII«Nez Perc6 lnd!ons end the opening of the Northwest+• by Alvin"· Josephy0 Jrto Yale Un lv. Press1 19€5. 705 P• Sub.! NEZ PERCE INDIANS--HISTORY NORT H~ EST0 PACIFIC-HISTORY r . o. o8I009 Call Auth Tltl Subj J.D. Call Auth Tl tl Sub j I.D. Call Auth 1111 Subj I. D. E17l .A43 INDEX Amer i can h eri ta ge typ Joe cllg u. sht yr 66 Ten year cu~ula t! ve Index, Amer!con heritage. Volu•e VI, Hu•ber 1-Volu~e XV• Nu•~•r 6. Dece•~•r• 195t - Octobe~, 1964. American Heritage, 1965. 167 P• U. S. --HIS TORY--INDEXES u.s.--CIVILIZ~TION--INDEXES 081010 El76.1.Bl7 typ loc chg tlo T aht vr 65 ::!!~~;ni~:~•:r~;tness: the Saage and th e • •n !toe George Washington to the presentt• By Tho••• Ao B•lleyte App l eton- Cen t ury, 196~ 368 ~· PRES IDEtiTS- -U .s. 081011 EB06.12 V.l tvp Joe chg I ckes, Harold t. secret dlar~ of Harold L• Iekel• Si~on 1nd Schuster, 1954-~5. 3 v. The pro9ra• ha; e ncountered a card sequence error. u. s. --POL !TI CS ANO GOVERN~ENT--1933-19~ 5 08101 2 u. T slit )lr 66 Page 35 FRANCE Call I.D. . E81!.§..I2 V.2 ·~~~i_:i) typ I oc chg tlo sht S yt 66 ( .... <;.. C'-<1 e'l/c'l3 Fig. 4. The Edit List. ~ <:r.l I .,... .a 1:'"1 .,.. ~ ~ ~ ~ B" ~ .... a· ~ ~ !""" ....... ........... ....... ~ II) '"t _?- ....... ~ ~ A Book Catalog at Stanford/ JOHNSON 27 THE EDIT LIST On a regular basis, generally once a week during activation, an edit list is run on the computer for cards punched since the last listing. An example of a page from this list is shown in Figure 4. This list is proofed against original cataloging data as represented on the coding sheets. In- formation still remains on punched cards, and errors detected are cor- rected on the cards. As an aid. to proofing, the edit program generates a number of error messages: cards out of order; absence of Area 10 information; call num- ber incorrectly formatted; an invalid character punched; information too long to fit into two machine records; information in one field more than 400 characters in length; an incorrect use of coding symbols (for use in determining filing order and to set off author statements) ; and incorrect use of record marks in Areas 60 and 70. A nagging problem encountered during proofing is the fact that it is done out of context. In the preparation of cards for a standard card cata- log, a second form of proofing is possible when cards are filed and entries compared with headings already in the catalog to insure that they are compatible. With a machine performing this function, this further check is not practical with existing equipment. Similarly, the machine does not recognize human errors and will file a misspelled word as it was entered and not as the word it was meant to be. The discipline required for the accurate inputting of cataloging information intended for machine ma- nipulation is at times frightening. Differences between a book catalog and a card catalog became more obvious as work progressed. In a card catalog one can: see from the typography, the stains on older cards, and the kind of card stock used, tl1at information was entered at different times. One is more willing to tolerate the differences that appear. Because it is not generally possible to compare a number of entries at the same time in a card catalog, one misses many inconsistencies. On the other hand, in a machine-produced book catalog, one scans numerous entries with one glance. Produced at the same printing, they appear of equal vintage even though they may have been entered at different times. The inconsistencies resulting from changing cataloging rules become very obvious. This is particularly an- noying with respect to the matter of capitalization, and the effort to pro- duce an internally consistent document is difficult. COSTS FOR INPUTTING The first year's experience ( 1965-66) has indicated a cost of $.40 per title for inputting of cataloging information. Indications are that this cost has remained constant for the second year. Included in inputting for each title is provision for all extra records needed for added volumes and copies and cross references. The cost does not include actual cata- 28 Journal of Library Automation Vol. 1/ 1 March, 1968 loging, preparation of the typed catalog card, overhead, or computer charges for the edit list. The cost may be broken down as follows: Table 2. Inputting Costs, 25,000 Titles (1965-66) Coding: 50 titles per hour @ $2.20 per hour Keypunching: 12 titles per hour @ $2.20 per hour Proofing: 72 titles per hour @ $7.40 per hour ( 2 staff members) Equipment : Keypunch rental ( $926.02); punch cards ( $312.34); and coding sheets ( $520.86) $ 1,100.00 4,583.33 2,569.43 1,759.22 Total $10,011.98 The Stanford experience indicates that over a period of time it is pos- sible to input 100 titles per eight-hour day on each card punch or ap- proximately 2,000 per working month. This figure is based on a shortened catalog record as described, but includes provision for separate records for added volumes and copies and cross references. The staff employed at Stanford had no previous experience keypunching and were instructed either in a formal school for five days or, as is now done, on the job. Three or four staff members are trained for keypunching at all times and have regular schedules. With a staff of this size, punching can proceed . on a steady basis in spite of vacations and illnesses. THE PROGRAMS Systems design called for the preparation of eight computer programs. In addition, two package sort programs from IBM are employed for the filing of entries in the shelf list and in the author & title and subject catalogs. As experience increased, it was found that the updating of en- tries in the first annual catalog when merged with new data for the sec- ond annual catalog could be simplified if three utility programs were used; these also were prepared. A locally devised assembly language, SOPAT, similar to Autocoder, was used for the programs. The first prog~am is the Edit Program ( LB001) which processes the cards to prepare the edit list described above. During the first two years of operation the basic pattern has been to prepare weekly the edit list described above using the Edit Program ( LB001), and transfer data to permanent storage on magnetic tape once every three months. (The punch cards are stored in another area on campus, as back-up.) The quarterly basis has coincided with the sched- ule for the program tests as well as the quarterly supplements and annual catalogs. In brief, the following happens: 1. ·Cataloging information is transferred from punch cards to magnetic tape through the Card to Tape Progr~m ( LB010). 2. Through a Call· Number Sort ( IBM Sort 7) the above rec- ords are arranged in call number order for a basic shelf list. A Book Catalog at .Stanford/ JOHNSON 29 3. Through the Format and Update Program (LB020), all the necessary entries for the author & title catalog and subject catalog are generated from the above records and the shelf list is updated. 4. Through an Alphabetical Sort (IBM Sort 7 or IBSYS 7090 Sort depending on the magnitude) the entries for the author & title and subject catalogs are arranged in alphabetical order. (Longer sorts have been run on an IBM 7090 Computer. ) 5. Through the Author-Title and Subject Update Program ( LB050) , the new entries created by LB020 are merged with existing entries and existing records are deleted as speci- fied by LB020. 6. The Author-Title and Subject Split Program ( LB060) sets up the entries on magnetic tape (line length, indention) as they will appear in the catalog and establishes the two columns for each page. 7. The Author-Title and Subject Printout Program (LB070 ) prints the pages of the author & title and subject catalogs. 8. The Shelf List Split Program ( LB030) performs the same function for the shelf list as undertaken by the Split Program for the author & title and subject catalogs ( LB060). 9. The Shelf List Printout Program ( LB040) prints the pages of the shelf list. THE CHANGE PROCEDURE It is possible to change information in a preceding supplement by fol- lowing tl1e change procedure: To change a call number, the entire entry is deleted, employing a record consisting of Area 10 and a delete symbol in the change indicator, and a new entry inserted in a separate record. To change an area, a change record is prepared consisting of Area 10 information with the change indicator marked plus card( s) for any area( s) to be changed. For example, if there is an error in a subject heading, it is necessary to prepare only an Area 10 plus an Area 60 change record showing the subject headings desired. The smallest unit for editing purposes between supplements is the Area. CHANGES IN AN ANNUAL CATALOG Dm;ing systems design it was realized tl1at machine time could be saved if a different procedure were followed to change information in the preceding year's catalog when merging it with fresh data to form a new annual catalog. The procedure used is to delete through three utility programs ( LB075, LB080, and LB090) entries that are to be changed and then enter them anew. The first of these programs ( LB075) is a card-to-tape program. Special delete cards (essentially Area 10 information giving call number ) are 30 Journal of Library Automation Vol. 1/1 March, 1968 prepared for each volume and copy to be deleted. They are transferred to magnetic tape and sorted into order by call number. In the second program, the Shelf List Delete Program ( LBOBO), the last annual shelf list tape is read and entries as specified by LB075 are deleted and a new shelf list tape written. Since the author-title and subject files are not in call-number order, a table look-up technique is used in the third program, LB090, the Author- Title and Subject Delete Program. The table consists of call numbers (in proper sequence) for all records to be deleted. Each entry in tl1e author-title and subject files is checked against the table and deleted if the call number for the entry is listed there. A revised author-title and subject tape is thus prepared. RECORDS ON TAPE The basic tape record follows the format of the master machine record described earlier. Figure 5 shows a tape dump of the master shelf list 501 601 701 801 901 1001 1101 1 lOt 201 301 ~01 501 601 701 801 ,01 1001 1101 1 101 201 301 401 501 601 701 801 901 1001 1101 1 101 201 101 ~01 501 601 701 801 901 1001 1101 1 101 H 174 0563 12 1 C S46 + 1 10 20 30 40 50 60 70 80 ~0 100 e 1761 ~17 1 r 66 108021 129151> 27901 7 BAILEYo THOMAS A.+ PRESIDENTIAL GREATNESS THf IMAGE ANO ne MAN FROM GEORGE WASHING TON TO THE P~ESE~T+o 5Y THOMAS Ao 5AILEY?. "PPLETON- CENTURY, 1966o 368 PotPRESIOENTS•-u.S.t S66 + 1 10 20 30 40 50 60 70 80 9 0 !CO e 1761 n 1 c r 66 1oao23 t31nnrco3510501 7 KANE, JCSEPH NATHAN+ FACTS ABOUT THE PRESIDENTS A CCMPilATICN OF 810GRAPe l "l _.C HI STOR !CAl OAU. 5Y JOSFPH NATHAN KANE. Ho W. WilSON, 1960. 348 P.H S$H$ESLSVSESO CNLY I ~ ARU Z 30o+PRESIOENTS-U. S.+ +E 1762 "4 1 T 66 108018 126153 279CH ~EANSo ~lRIA~~E+ 7~E WC~AN IN THf WHITE ~OVSE THE liVES, TIMES AND INFlUENCE OF TWElVE MTABlE fi~ST l~CIES+. 8Y II ARIAMNE ~EANS+. RANOCH ~OUSEo 1963o 299 P.+PRESIOENTS--U.S.--~IVES+ + 1 10 20 30 40 50 60 70 80 90 100 E 178 A2~ 1 I K T 66 108C24 1321C1233 C9933204 'I AOl~S, JAHES 7~USlCW+ ALBUM OF ~MERICAN HISTORY. JAMES ·TR~SLC .. ACAMS, ECITOR I ~ CHIH . . C. SCA.Ifthf.R• l91tlt-60. 6 V.+- VOle 5 EClTED BY J. G. E. HOPkiNS.• VOL. 6 (S l"'CEX.4$ SS H$ E HlVS ESO IN REFERENCE ALCOVE HO.+U.s.--HISTCRY+U . S.--SOCIAL LIFE AND CUSTO~S? +E 178 A24 2 1 K 566 + 10 2 0 30 50 60 70 8 0 9 0 10 0 178 .~24 l K S66 Fig. 5. Tape Dump of Master Shelf List. A Book Catalog at Stanford/ JOHNSON 31 as formatted in the Card to Tape Program (LB010) and after having been sorted into call-number order. As stated above, two records are placed in a block of 1,140 characters, each record with 570 characters. Printed with a limited print chain, some characters do not appear as they will finally. The record mark prints as a plus sign, and in the entry for the work by Kane the symbol indicating underscoring prints as a dollar sign. The word-separator characters and some other characters do not print at all; spaces are left to indicate their presence. Once the information has been processed through the Format and Update Program (LB020), however, the machine record is somewhat different. New records, one for each entry that will appear in the final catalog; are generated. A listing of the elements in each of these records is shown in Table 3. Figures 6 and 7 show tape dumps for author a11d subject entries. Table 3. The Author-Title and Subject Tape Record. Position 1: 2-81: 82-101: 102-131: 132-136: 137: 138: 139: 142: 145: 148: 151: 154-618: Type of Information Catalog indicator ( 1: Author & Title Catalog; 2: Sub- ject Catalog) Major sort key Minor sort key Library of Congress classification number Size and format Record indicator (Program supplies ·"1" if there is an overflow record and "2" in second record) Delete indicator (program supplied, for use in change procedure) Address for main entry (Area 20) or added author/ added title (Area 70) Address for title paragraph (Area 40) Address for conventional title (Area 30) Address for notes (Area 50) Address for subject heading (Area 60) Variable-length fields THE SORT KEY The Format and Update Program ( LB020) generates the sort key for each entry. The sort key determines the characters that will be consid- ered when the entry is to be alphabetized. A succeeding sort program does the actual alphabetizing. Mter some study, experimentation, and conjecture a 100-character sort key was selected. It is in two parts: a major sort key of 80 characters and a minor sort key of 20 characters. The major sort key is formed from the first 80 characters of the ele- 32 Journal of Library Automation Vol. 1/ 1 March, 1968 1 10 1 20 1 30 1 401 501 6 0 1 1 10 1 20 1 JCI 40 1 SO l t01 1 1 01 201 301 4 CI SC I 6 01 1 10 1 201 30 1 40 1 501 6 01 1 101 20 1 30 1 4 0 1 50 1 f 0 1 I 10 \ 20 1 301 1 10 zo 10 1BA1CWI ~ T ~ RPR 3095 83 L OF I~E S~AKESPEAREAN COMPANY • 40 s o 60 70 eo 90 100 CRG ANIZHICN ANC PE 154 171 BALDWIN, T. W,+ THE ORCANllATICN A~O PERSONNE RUSSELL - RUSSELL, 1961. 463 P.+ + I 1 10 zo 30 40 50 60 70 80 90 100 POLl T ICAL PART! ES 8AllEY, STEPHEN KEHP+ POliTI CA L PARTlf. St U. EOJIEO BY ROBERT A, CCLCW JN. RAN C MC HAL 1BAILEY STEPHE N KEHP UJK 226S G59 1 54 178 s. A. ESS.V S tY STEPHEN KENP BA ILEY M/0 OTHERS. lY, 1965 . 158 P. + + 1 10 20 30 40 50 60 1"AI LEY TPO~AS A EE 1761 Bl7 . 1 541 75 BAILEY, P. E !~AGE ANO THE "AN FROH GEORGE WA SHI NGTON TO THE PRESENT • + ·1 10 20 30 40 ! BAI LEY TPO~AS A TO 6 4 1 A7828 1 54 175 RE4T BETRAYAL, H ICOUNTER PAP ERBAC KS, 1963 . + 50 6 0 BAILEY , lt29 P.+ ro eo " too PRES IOH T I AL CREATH THOMA S A.+ PR ES,! CENT I AL CRE6TNESS T APPLETON- CEN TURY , 1~66. 168 P,+ 70 80 90 10 0 WCCC ROo WILSON AND TP.OHA S A.+ WC CCRCW oiLS CN AND T~E C 1 10 20 30 40 50 60 70 80 90 100 LIFE AND EXPl OR AT 10 1 6 AJ N J A~THUR N~ 70018~38) 1 54173 BAJ~ , J , ART J.tVR+ LIFE AND EXPLORATIONS OF f RICTJCF NANS EN . • lt'i 9 P.+ ~EW EO• RE V. AND f.ONSIOERABL Y ENL. , WtTH NUMEROUS !LlUS. ANC MAP. w, SCCTT , N. O + 1 10 20 30 40 50 ! BA I NVIllE JACCUES 7C C 335 832 1 54 175 70 -1 935 , TRANS lAT ED, WITH AN INTROOUC TDRY NCTE, BY 60 70 80 90 10 0 FRE~CH REPUeLIC 18 BAINVILLE , JACOU ES + T~E F~fNC' RfPUBLI C, 18 HAMIS H HILES. J, CAPE , 1916. 25 3 P. + Fig. 6. Tape Dump of Entries in Author & Title Catalog. ment that will serve as the entry in the catalog-the author, the added author, the title (for entry under title), the added title, or the subject heading. The minor sort key is formed from the first 20 characters of the element that follows the heading-the title or conventional title. The conven- tional title can never be a major sort key. The title, on the other hand, can be either a major sort key or a minor sort key. During the course of the Split Program ( LB060), major sort keys are compared. If two or more are identical, the entry words for the second and subsequent identical headings are suppressed and do not print. Un- der the heading that does print, entries are arranged in alphabetical order through the first twenty characters of the element generating the minor sort key. In a sense, there is a third sort key. If both major and minor sort keys A Book Catalog at Stanford/JOHNSON 33 are identical, items will print in call-number order. In no case will the element generating the minor sort key be suppressed. During the first test of the programs one mistake was discovered with respect to the generation of sort keys and the suppression of entry ele- ments. In a few cases the library possessed multiple copies of the same book cataloged under different call numbers (for example, one as a separate and one as part of a series). The problem arose when there was to be an entry under title, with two major sort keys identical. A similar situation arose for periodicals. A periodical might be entered itself under its title (Area 40) and also appear in the catalog as an author (Area 20) for a book it might issue. To eliminate this problem, a minor 1 10 20 ...)O,L 40 50 60 70 80 90 ICO 1 2P~F.~ERVATI~N CF 2COLOGICAL SPEC t•ENS SE E" ZC CLCGTI'l!L SPF.C 1 01 I lCCLCCICA12ll~616 191 154PRESERV4TICN GF ZCCLCGICAL SPECI"E'S+\ SHSE Z 201 CClCGJCAL SPECI"ENS-- COllECTIO~ 4NO PE PECPLE S CHOIC E 201 HC" THE VOTER ~AKES UP HIS Ml~lO I~ A PRESIOENT!Al CAMPAIGN, eY PAUL F. LAZARSFELC, BERNARC 8 3CI EltELSCN, AND' HAlEL GAUOET. ZO EO. COLUMBIA UNtV. PRESS, 1965, 11e P o+ 4 01 501 t:Cl + 1 10 20 30 40 50 60 70 BO 9C ICO 1 2PRF~ICENTS U S ELECTION STATISTI CAL ~ ISTCRY 1CI JK 1967 P4 tal 34q[54PRESIOENTS--U.s.--ELECTIC•• A SHTIS IICAL tt lSTC 2CI RY CF THE AHRICA~ PRESIOF tHIAl ELECTIONS, P.V SVF.NO PET~RSEN, INTRCC. C"~ ·I·TIC,A L ELECTic•s 301 AY LCUIS Fill ER: • F. UNCARt 1963. 247 P.+S S,H.E'ItlVSE'O 1~ REFERfNCE ALCCVE 'HO.+ 4CI '3Cl ttl + 1 tr.l 201 1 10 2.0 30 40 2PHSIDENTS U S ElECTION 1696 CF 71C J6 11!7 tAL HECTIC~ CF 1•~6, 8Y STA~LEY lo JO>;fs, so 60 70 eo 9C teo PRESlt£•TIAl ElECT! 154PR£S I OE~TS--U. S .--FLECT tc•--1~96• T~E PRES I CENT lJNIVo OF wtSCO'SIN PRFSS, 1964. ~3~ Po+ Fig. 7. Tape Dump of Entries in Subject Catalog. 34 Journal of Library Automation Vol. 1/ 1 March, 1968 change was made in the program: If a title generates a major sort key, it is never suppressed if identical with a preceding major sort key. FILING ORDER Formation of sort keys leads immediately to a discussion of the un- solved problem of alphabetization. In the Meyer Library catalog the aim has been to duplicate as closely as possible the arrangement of entries found in the University Libraries' union card catalog. Basically, this means a word by word alphabetization. In addition, we have attempted to pre- serve as many of the currently used typing conventions as possible in preparing entries for the card catalog. For example, two or more initials separated one from another by periods have no spaces following internal periods. Thus, the abbreviation for the United States is typed as U.S. and not as U. S. Abbreviations filed as they are spelled and Me's and Mac's in separate sequences are two of the major differences from stand- ard manual library filing. It was recognized that in generating the sort key the computer will scan the entry words character by character and space by space. Thus, it is important that each character and each space be positioned accu- rately. The computer checks a character and either interprets it as a blank, a letter, a numeral, a symbol, or else ignores it. In alphabetizing, · this basic rule is followed : A blank files before a letter (A through Z), and a letter files before a numeral ( 0 through 9). . Certain marks of punctuation are interpreted as a space. They are: period, comma, colon, semicolon, hyphen, and question mark. Some marks of punctuation are ignored. They are : parentheses, brackets, dol- lar sign, virgule or slash ( / ), equal sign, number or sharp ( #), per cent, asterisk, apostrophe, Hat sign, and ampersand. It was believed that the presence of a space on either side of an ampersand would place entries in correct order, but in some cases this did not happen. Some diacritical marks change the value of the character with which they are associated. For example, an umlaut over an "a," "o," or "u" changes that character to "ae," "oe," or "ue" respectively. NON-FILING SYMBOLS If an alphabetical order is desired other than that explicitly given in the entry words, special symbols are employed at the time of coding. Since language of publication is not coded, it is necessary to place sym- bols around introductory articles for them to be ignored. The less-than ( < ) and greater-than ( >) signs are the symbols used to set off a sequence of characters to be ignored. The placement of these symbols is important. For example, to eliminate the article from the title, The century of sci- ence, it is necessary to place the symbols in this manner: century of science. In this way the sort key would be generated starting with A Book Catalog at Stanford/JOHNSON 35 the letter "C" in the word "century"; and a space would be left between the words "the" and "century" in the printed heading. Use of the non-filing symbols internally in an entry is limited and must be strictly controlled through recording of decisions in authority files. So that names filed under prefixes and written as two words will be filed in the same sequence as names written as one word, the non-filing sym- bols are employed internally. For example, in order that Van Buren and Vandenburg will file in the same sequence, Van Buren is coded as Van< >Buren. In this manner the computer is instructed to ignore the space when forming the sort key. The use of non-filing symbols has proved quite useful in subject head- ings to arrange period subdivisions in chronological order when there is a word or words intervening between the heading and the date. Thus, with non-filing symbols employed as shown below, these particular sub- ject headings are arranged chronologically: CT. BRIT.-HISTORY-1042-1066 CT. BRIT.-HISTORY-1066-1154 CT. BRIT.-HISTORY-1066-1485 CT. BRIT.-HISTORY-1135-1154 CT. BRIT.-HISTORY-1154-1189 CT. BRIT.-HISTORY- 1154-1216 THE FILING SYMBOL The less-than and greater-than signs are provided on the expanded print chain purchased for the book catalog project. To date it has not been found necessary to use these signs as symbols in titles, and so their use is restricted to their role in forming non-filing elements. As work proceeded, need was felt for another symbol-one that would set off a field that would not print but which would be filed upon. For example, we wished to file the title, 1848: chapters of German history, as though it were written, Eighteen forty-eight; chapters of German history; yet we did not wish to violate the form of the title as given in the book. An examination of all characters in the print chain led us to sacrifice the symbol @ for use as a sign in its own right. It is used solely as a filing symbol. Thus, any characters or spaces placed between two @'s will generate a sort key as specified by those characters, but the information will not print. The title, 1848: chapers of German history, will be coded in this manner: @eighteen forty-eight@< 1848 >: chapters of German history. It will be filed as though it were written: Eighteen forty-eight: chapters of German history. The use of the filing symbol has been especially useful in arranging period subdivisions in chronological order when treating years in the pre- Christian era or before the year 1000 A.D. For years in the pre-Christian era, coding permits chronological arrangement of years beginning with 36 Journal of Library Automation Vol. 1/ 1 March, 1968 9999 B.C. The following procedure is observed: The year in question is subtracted from 9999 and the resulting difference, preceded by the letter Z, is entered inside the filing symbols. The year will thus file after all letters but before years in the Christian era. For years before 1000 A.D. the leading 0 is simply placed inside the filing symbol, for example, @0@476. To illustrate further, here are three subject headings as manually filed: ROME-HISTORY-REPUBLIC, 510-30 B.C. ROME-HISTORY-REPUBLIC, 365-30 B.C. ROME-HISTORY-AUGUSTUS, 30 B.C.-14 A.D. They are coded in this manner : ROME-HISTORY- @Z9489-Z9969@ ROME-HISTORY-@Z9734-Z9969@ ROME-HISTORY-@Z9969-0014@ The following sort keys are generated: ROME-HISTORY-Z9489-Z9969 ROME-HISTORY-Z9734-Z9969 ROME-HISTORY-Z9969-0014 The headings will file in correct chronological order and print as origi- nally shown above. OBSERVATIONS ON FILING ORDER With but few exceptions, the filing order as designed has proved a very satisfactory arrangement. It has been felt advisable to place notes at various points in the catalog to link together headings which are filed separately. For example, the abbreviation Mr. is filed as mr, and the word Mister is filed as mister. Here a note refers from one to the other. In the subject catalog it was discovered that if a country or local heading is abbreviated, two different alphabets are established. So far this has occurred for the United States (U.S.) and Great Britain ( Gt. Brit.) The terminal period generates a space when the sort key is established. Thus subdivisions separated from the heading by a dash (two hyphens equiva- lent to two spaces) are in fact separated by three spaces and file before jurisdictional or form subdivisions which do not require the dash. For example, U.S.-HISTORY files before U.S. DEPT. OF STATE. A note in the catalog gives instructions on the filing order in such a case. Less fortunate is the situation of the author who chooses to use a name with a first initial and a complete middle name. Because of the period and space separating the first initial from the middle name, there are established two spaces. Thus, the following "incorrect" alphabetical order is established. Smith, J. Russell Smith, J.A. Smith, J.C. A Book Catalog at Stanford/JOHNSON 37 As may be expected, situations such as those described above do not occur often. It is hoped that through scanning of the open page before him the reader will find the correct heading. It may be argued that the coding required to achieve the alphabetical order in this catalog is too demanding for a project based upon use of a sophisticated electronic computer. Possibly, programming should have taken care of all of this work. It has been our belief that we have achieved, in terms of the present state of the art, a good balance between what the machine should do and what the human should do. In the process we have been able to keep the form of the information as it appears in the source. As examples, introductory articles have not been eliminated from titles, and Library of Congress subject headings have been retained ( 11). Most important, it has been possible to implement these rules consistently with a relatively inexperienced staff. PAGE CREATION With each entry created and alphabetized, the Author-Title and Sub- ject Split Program ( LB060) is called upon to generate the lines for the final catalog and create the two columns of each page on magnetic tape. The final program ( LB070) prints the pages of the catalog from the tape. The computer line printer permits the use of 132 print positions in each line. The type size is the same as pica type-ten characters to the horizontal inch, six lines to the vertical inch. It was decided that the completed page size for the book catalog should be 8~" · x 11". With an allowance for an adequate margin on all four sides of the page, it was believed that the reduction necessary to employ the 132 characters in a line probably no longer than seven inches would be too great. Ex- perimentation led us to accept a reduction to 68 per cent and use of 98 of the 132 print positions. This can, however, prove expensive, as the printer takes as long to print 98 characters as it does 132. The catalog page as designed calls for two columns, each 45 characters in length, with an eight-character margin between them. The text is 80 lines, and the page is 84 lines in length because of the heading at the top and the page number centered at the bottom. Catalog entries are not split between columns, so that the bottoms of the pages are rarely even. To simplify programming it was decided not to attempt programmed hyphenation of words or to require right and left justification of the lines in the catalog. The first words of a catalog entry are set flush left, and all successive lines are indented two spaces. The call number is set flush right on the last line of the entry if there are three spaces separating it from the last word of the entry; otherwise, it is set flush right on the following line. As stated earlier, entry words (authors, subject headings, added titles) are suppressed if they are the same as those found in a preceding entry 38 journal of Library Automation Vol. 1/ 1 March, 1968 and are repeated only at the head of a new column if the entries are continued there. Entry words are so clearly shown in the catalog that it was not considered necessru:y to use keys at the top of each page indicating which letters are included on that page. Because an expanded print chain is employed, speed on the printer is considerably reduced, actually to 250 lines per minute. The printer requires eighteen seconds to print one page. The page image is approxi- mately ten inches by fourteen inches. Through use of the Itek Plate- master, this image is reduced to 68 per cent and an offset master cre- ated for reproduction on offset equipment. Figures 8 and 9 show repre- sentative pages from the 1967 annual catalog. The foregoing account has emphasized the preparation of a page in an annual catalog. Except for the size of the page and the kind of paper used, the identical process is followed for the preparation of the supple- ment. Through a switch setting, a forty-line page for the supplement is printed. The supplement is printed on ten-ply paper ( 8~" x 11"), kept in unburst form, and bound at the top in post binders. Figures 10 and 11 show representative pages from the January, 1967, supplement and illustrate how some of the entries earlier depicted in Figures 6 and 7 appear in final form. Through similar programs the shelf list is prepared and printed in . essentially the same format as the supplement, a 98-character line and a page of forty lines. A key at the top of each page indicates the first call number on that page. The shelf list is printed on four-ply paper, and copies distributed to important staff service points in the Main and Meyer Libraries. Figure 12 shows a copy of a page from the January, 1967, shelf list, depicting how the information from Figure 5 appears when finally printed. The lone call number, E176.1.B77, copy 2, is for an item, of which copy one is represented in the 1966 annual shelf list. THE FIRST ANNUAL CATALOG AND ITS SUPPLEMENTS The first annual catalog was prepared during the summer of 1966, list- ing the 25,000 titles cataloged as of the end of June. The catalog was 2,804 pages long, 1,569 pages in the author & title catalog and 1,235 pages in the subject catalog. Each page from the printer was first scanned by library staff and seri- ous errors masked with white tape. In consequence, the user of the cata- log will encounter an occasional blank on a page. The pages were then sent to the University's Photo Reproduction Service, where offset masters were created and fifty copies of each page reproduced. The Stanford University Press prepared the binding. Each set of the catalog was bound in red buckram in seven volumes, approximately 400 pages in each, four volumes for the author & title catalog, three volumes for the subject catalog. There is a title page in each volume and several pages of expla- nation on the use of the catalog. Letters included in each volume are A Book Catalog at Stanford/ JOHNSON 39 AUTHOR ~ TITLC CATALOG -- JUNE 1967 Ho rnbeln, Thomas r. Everest: th e west ri dge. Photop raphs fr o• the A•er le~n Hount Eve rest EKped ltl on and bv tts leader, Nor~ an G. Dyhrentur th . Introd. bw William E. S!rl. Edited by David Br.ower. Sierr a Club, 1965. 198 P•, Ill us. DS486. E8H54 Foil o Hornberger, Theodore · Benj~mln Fra nklin. Univ. of Minnesota Pre ss, 1 962. 48 P• PS75l oH6 lfo rnbJow, Ar thur The ca p tive, b~ Edoua rd Bourdet. Translated by Arthur Ho r nb l ow, Jr. lntr od . by J. Brooks Atk inson. Bre n t ano's , 19 26 . 255 p. PQ2603.0777P7Z A. hi sto ry o( the theatre in Afl'l.e rlca (t' om Its beg lnnln n s t o the pre sent tt-e. J. D. Li ppin cot t, 19 13. 2 v. PN2221.H6 The trlu~ph of dea th, by Ga briele d'Annunz io . Tra n, lated by Arthur Ho rn blow. lntrod. by Burton nasc oe . 8oni a nd Llverl ght, 192~ . 412 P • PQ4 803.Z3T7 Hornb y , A. S . A nuid e t o patterns and usage in English. Oxford Untv. Pre ss , 1962 . 26 1 p. Hot"n e, Alistair The fall of Pari s; Commune , 1870-71. 1966. 458 P• PC14 60.H5 4 the Siege and the St. Hartin's Pr e ss, OC311.H6 The price of ~ J o ry! Verdun 1 91 6 . S t. Ma rti n*s Pr ess , 196J . 371 P• D545.V3H6 Return t o po wer; a r epor t on th e new Germa ny . Pr a ege r, 1956. 41 5 P• 00259.4 .H65 Ho r ne, C. Sl l vest ~r Put"lt.t nlsD .tn d art ; an l nq ulr11 int o a popular rat lacy. B~ Joseph Crouc h. Introd. by the Re v. C. Si lv es ter Horne. C as s e llt 19 10. J8 1 P• N7Z.C 8 Ho r ne d Moo n ; an a ccount of a jou r ne y throu g h Pakis t an, Kash• ir, a nd Af gha nist a n. BV I an S te phen,. . Indillna Unlv. Press, 1955. 2.88 P• 05377.58 Horner, Har l an Ho~t Lin coln and Gre:ele!l• Un l v. ot 11tlno t s Press, 1953. 43 2 P• E45 7. Z. H79 Horne~ , tlia ren rwork o . 196 4.] The coll ec t~d work s o f K~ren Horney. W.W. No rt on, 19~2-6~. 2 v . Conte n t s . - v.l. Th e neurot ic personality ot our t i ~e.- v.2. Self- analy s i s . RC435 .ff6 Neurosis and hu man Qrowth; the struggl e t o ward se lf-reall z otlon. w.w. Norton, 1950. 391 P• RC 343. H5 48 The neuro t i c pe r sonality of our tf •e • W.W . Norton, 1937. 299 P• RC343oH75 N~w Wll l/S ln psychoanalysts. w.w • . Nof'ton, 1939. 3 13 P• BF1 73.H762 Ou r inn e r confli cts: a co nstructive the o ry of neurosis. w.w. Norton, 1945. 25 0 P• RCJ43,H56 115Z J. HENRY MCYCR MEMOR I AL LIBRARY Horne~ , Karen Self-an al ysis. W. ~. Norton, 1942. 309 P• BFI73.H"1625 Ho rn gren, Charl es T. Cost accountin g ; a managerial e~pha~ i s. Pre ntice-Hall, 1 964. 80 1 r • HF~5~6. C8H59 Hornik, He nri Le t emp le d'honneur et de ver tu~, par Jean Lenalre de Belges. Ed. critique pub ll ~e p4r Henri Horni k. Oroz, 1957. 136 p. PQ1628 . LST4 Horns, stri nn s and har•ony . fty Art hu r H. Be na::le. Doubleday, 1960. 271 P• ML380S.B33 Das Ho rnunge r Helm we h, a nd o ther 3to rt es . By Wern e r Bergen~ ruen. Edi t ed by W. I. Lucfts. To Ne ls on , 1963. 1 17 P• PT260J.ES9H6 Ho rodls ch • Abraha m Picasso ~s a book a rti s t. ~orld Pub• co •• 1962. 136 P• NC247.PSH63 H~ ronj~rr, Robert Th e p lanning and design ot airports, HeGraw -H lll, 1962. 464 P• TL7 2S . 3 .P5H6 Horowit z, David He•ls phe res no rt h and south; econo~ lc di s parity among na ti ons . John s Hopkins Press, 1966. 11 8 P• HD82 . H617 Stu:fe n t . R~lla~tt n~ Books, 1962 . 160 P• Ho r owitz, Irving Lou l5 The anarchists, edited Irvin g Louis Horowitz. P• L076 0.1i6 vll h an intr od . by 0• 11, 1 96 4. 6 40 HX 82 6.H6 The I dee o f var end p eace in conte mpora r y p h ilos op hy. W~th an in tr oduc tory essay by Roy Wood Se llar s. Pai ne-~hit~an , 1~57. 224 P• J X1952 . H72 The ne ~ socio l ogy; e$says In ~oc ial sc ie nce and socia l th eo r y in honor o ( C. Wri ght Mi ll s . r.di ted by Ir ving Lou is Ho rowitz. Oxfor j Un i v. Press , }g64 . 5 1 2 P• H35.H68 Radi ea 1lsm and t he rev o lt B3alnst reason; the socia l th eori e s d f Georges So r e l, wit h a tra ns lati on o ( hls ess a y on The deco•posl ti on of Ma r xi s m. Hu• en t tir.s PreSS t 1961. 264 P• HX26J.S6H5 Revolution In Brazil; polit i cs a nd soc iet y In a d eve l opi ng nation. E.P. Dutton , 196 4 . ~30 P• f2538 . 2 . H6 Th r ee .,. orl ds o t deve l o pMen t; the t heory and p ra c t ice o f international stratl!icatlon r Oxfor d Univ. Press , 1966. 4 75 P• 0640.H6 Horrabin, J.r. An atlas or Afrtca. 2d, rev . ed . F.A. Praeger, 1961. 126 P• G2445.116 Mathe ~atics Cor th e mlllton, by Lance l o t Hogb en. I llustrations by J .f. Horrahtn. W.~. Nort on, 1 937. 647 P• QA36 .H6 Horrobln , Dav id P. The com~unlcat lon systems o r t he body . Basic Bo oks, 1964. 214 P• QH508 .H6 Fig. 8. A Page in the Annual Author & Title Catalog. 40 Journal of Library Automation Vol. 1/1 March, 1968 SUBJ ECT CATALOG -- JUNE 196? CROMYELL, OLIVER Ullver Cro~well, by John Morley. Centurv Co., 1900. 48G p. DA4Z6.K86 Oliver Cro•well, by c.v. Wcdgwood . Mac~illan, 1956. 14~ P• DA426.W4 Oliver Cro• wetl and the rule or the Puri tans In En~ l•nd. By S ir Charles firth . With an lntrod. by G.H. Youn". Oxro~d Unlv. Press, 1~61. 486 P• OA~26.F52 CROHYE LL, THOMAS Tho•as Cro•wel l and the En~lish Refor••tlon 1 b~ A.C. Dickens. English Unive rsities Pr e1 s 1 1959. 192 P• DA334.C9D5 CRONIN, A.J. Adventur e s in two worlds, by A.J. Cronin. Little, Brown, 1952. 331 P• PR6005.R68A4 CROP R£PO RTS ~.!! Agrleulture--Statlstlcs. CROPS--STATIST IC S ~~~ Agrlcu1ture--Stotiotlcs. CROPS AND CLIKATE--U.s. Cli•ete and •an, the yeerboo~ ot agriewlture, 1 ~41. U.s. Dept. or Agr iculture. U. S . Govt. f>riot. orr., lO•H. 1248 P• S21oA35 1941 CROSS, "ARI~N EVAN S §!!El iot, George, pseud. CROSS-COUNTRY RUN~ING §:!~ RIJnning. CROW INDIANS Cro w Indian beadwork: a descriptive and historical studS~• By Willie"' Wildschut and John C. Ewe rs. Museu~ of the A•e rlean lndlan, Heye Foundation, 1959. SS p. , il1uo. £99.C92W5 The Crow Indians, by Robert Lowie. farr•~ ~ Rinehart,. 1935. 350 P• E9!1.C92l913 The lite •nd a dventu res or Ja~es P. Ueckwourth. Edited by T.O. Bonner. A.A . Knopr, 1931. 405 P • f592.B393 The re11glon of the Crow Indians, by Robert H. lovte. AMerican Museue ot Natural History, tnz. 30>1-4H P• E99.C92L6 CROIIDS The crowd: Gu stave Le 23~ P• a studv or the popular ~I nd. By Bon. T. Fisher Unvln, 1917. HM28t.L 5 The cro wd In history: a ~tud~ ot popular disturbances In france and Eng land, 1730- 1848· 8y George Rude. J. Wiley, 1964. 2&1 P• HMZ83,R8 The erowd In the French Revolution, by Ceo r o• Rud~. Cl•rendon Press, 1961. 267 P• DC158.8.R& The psychology or social movement•, by Hadley Cantril. J. Wlley 9 1~41. 274 P• H"291o C3 418 CRUSADES An Arab-S~ rl•n "entJem•n and wlrrlor In the period or the crus•des: •e•olrs or Usa11ah lbn-Hunqldh. Trans l ated fro~ the original Manuscript by Philip K. Hltti. Colu~bla Univ . Preos, 1929. 265 P• DS!I?.U5 Background to the Crusades. a 8BC publication. British Broadc asting Corpore tlon, n.d. 38 P• 015 9. 87 The Crusades, by Ernest Barker. Oxford Unlv. Press, 1923. 11 2 P• 0158.83 The crusade3t by Richard A. Newhall. Rev. ed. Holt, Rinohort and Winston, 1964. 136 P• Dl58.N~ fhe crusades, by Zo~ Old~nbourg . Translated by An ne Cdrter. Pantheon Books, 1966. 650 P• 0158,04 The Crusades: Iron •en and saint~. By Ha r old La•b• Doubleday. Doran, lVJO. 368 P• D15?.L3 The Crusadest the ~tory ot the Latin Kingdo• ot Jeruaale~. By T.A. Archer and Charles L• Kingsford. c.p. Putn••• 1936. 467 P• Dl58.A67 A hlatorv ot the Crusades, by Steven Runc:l•an. Cambridge, Eng., Unlv. Press, 19~?. 3 v. D1 57.R8 The klngdo• of the cru•eders, by Dana Carleton Munro. o. Appleton, 1935. 216 P• D18 Z.M8 The recovery ot the Holy Land, by Pierre Dubois. Translated wlth an Introduction and not~• by Walth e r 1. Brandt. Columb ia Un lv . Press, 1956. 2Sl P• DI~Z.D813 CRUS ADE S-- HI STORY A history ot the Cru~ade~. £dttor-ln-chiet, Kenneth H. Setton. Unlv. or Pennsylvania Press. 1 958 - Librory has v.~-2. D1S?.S48 CRUSADES--FIRST, 109 6-1099 The first erusede; the accounts witnesses and participants. Pe 1958. 299 P• or eye- S~Ith, D161.1.A3~7 Gest• Francoru• et e11orum Hlerosollai t anorum. The deeds of the Franks and t~e other pllgrl~s to Jerusalem. Edited bl( Rosalind Hilla; lntrod. by R.A.B. "ynors. T. Nelson. 196Z. 103 9 103 P• In Lati n and En~ iish. D16l.l.G4 CRUSADES--SECOND, 114?-1149 De protection• Ludovlcl VII 1ft oriente•• edited, vlth an English translation by Virg inia Gingef"lck. Berry. Coluabla Uniy. •Preos, 1948. 154 P• Dl62oloU3 CRUSADES--THIRD, 1189-1192 The Crusade ot Richa r d Llon-hea'f't, by A~brolse. Translated by "erton Jero~e Hdbert. With notes and documentation by John L. L• Monte. Columbia Univ. Press, 1~41. 4?8 P• D163.AJA52 Fig. 9. A Page in the Annual Subject Catalog. 0 • • • ~ • • • ~ ~ oe ~ ~ ~ • /\UniOII E TITLE CATALOG -- JANUAI!Y 1967 SIJPPLf.HE:NT Bailey, Stephen Ke~p Political pnrtles, U.S./I,; essays by Step~en Kemp Bailey and others. Edited by Robert A. Goldwin, Rand McNally, 1965. 158 P• JK2265.G59 Bailey, Thomas A, Presidential greatness; the l10age and the man trcm George ~ashlngton to the present. Appleton-Century, 1966, 368 P• El?6.l.B17 Woodrow ~llson and the great betrayal , Encounter Paperbacks, 1963. 429 P• D543.A7B28 Baln, J. ATthur Lite and exploratlcns of Fridtjot Nansen. New ed. rev. and considerably en!., with numerous illus. and map. w. Scott, n.d. 449 P • G700,l893.B3 Bainville, Jacques The French Republic, 1870-1935. Translated, with an Introductory note, by Hamish Miles. J, Cape, 1936, 253 P• DC335,B;JZ Baird, A, Craig ArAumentatlon, discussion, and debate. McGraw-11111, 1950, 422 P• PN418l,B29 General speech; an Introduction. By A. Cral!l Ealrd and Franklin H. Knower. 3d ed. McGraw-1111 I, 1963. 44 8 P• PNH21. 8314 Speech criticism; the development ot ~tandar.rls Cor rhetorl cal aoo••al9al. Bv J, HENRY ~EYER MEKORIIIL LIBRARY Baird, Donald The English novel, 1578-1956; a checklist o! twentieth-century criticisms. By Inglis f. Bell and Donald !laird. A. Swallow, 1958. 169 P• ~h~l~ed only In Reference Alcove 280, Z20l4.F4B4 Baird, Wlllla~ Ralmond Baird's manual ot American college fraternities. 17th ed. G. Ba~ta, 1963, 834 p. ~h£1~£4 only In Area 2~0. LJ31.B2 Baird's manual ot American college fraternities. 17th ed , G. Banta, 1963. 834 P• ~h£1Y£4 only In Area 230. LJ3l, B2 La baja Edad Media, por Enrique Bag~ e y Juan Petit. s. Barra!, 1956. 412 P•t lllus. DP99.B3 Bak, B-rge Elementary Introduction to molecular spectra. 2d rev. ed. Intersclence, 1962. 144 P• ~C451.B16 Baker, Carlos American Issues: The social record. Edited by Merle Curti, Willard Thorp, and Carlos Baker. 4th ed. J.a. Llpplncctt, 1960. 1160 P• PSSO?.T54 Ernest He10ln9w~y: critiques of four major novels, F.dlted by Carlos Oaker. C. Fig. 10. A Page from the January 1967 Author-Title Supplement. C) • u uu 0 • • • w wu 8 • • • p uv • • w u~ ol > tl:l c c ..,... CJ ~ ~ ~ .... C/:) 4 ~ 0 :I: z C/:) 0 z ~ ~ \. '· • # .. C> v 0 40) ~ ~ q, ~ ~ ~ 0 Co 0 0 • ... , ''-V' 40$1o•· ....:.t~......,._ ,_... ·~-: ........... ~ .......... """',. ..... ....~ ~~- <:J: "' .. ,.JJt .,: r ,!A r., l.u.: -- Jf' ~IJJ\:t 'l 1 Jfj-) SU P ... LC~t: •: t" PO~YS, LLEWELLYN The Po~y3 brotherst by R.C. Churchill• ton gmnns , Grce nt l oos and Influence of twalve notable u~·at '• ~~~~. Col l:t ... llo,6 }lclel\~t e.dl .t.or: l if\ c~Jef• C:. ~crlb.11e t & F•"l .C& '!•1 c:,l Co v.l! c,l Co .Itt.~ c.:..l . ~.. ,111 .. "~"'60 .• _6 " ·•-. Y.o:ta ct.l C: ~'1.1.• ~ ~cl.l.t•d. ,bl( ~.G~J:•. JiO.pl\1 114•. ,vo t. !> .1 I\ ' l 11<\elt.f. . &11'6·)·)l~7 . . · DelleJt, Tt.o., u ~. h .utdoilt(.i1 a're~t!"•.••: tk• ... aae and. t.."~· ·~n . . r:rP•. Qearge V~s.f\l11ato11 to th~ .Present~ Ill!. Tka•.u ·"·-.. B•ll•ll • 1\ppleta!I,.,C.ellt.lll:.lft .~!1116 •. , 3d~ .Jh ..• .P RES.J DE:NTS~:-U •. $, tJUe E1.7~.l:,.B77 Co.2 &171it1 •. J<3 Kllll .t:t .!ot.eph tla~han . t:ec:.t• abou.t .the Presidents.:~ .CiollpiU.tlOI\ p,t ,b_logrephloe.l ~ncl hl,.,t.qr.lcal .d!ll.t!'• 811 "o .. ph Nathan Kane, H,,w, IU.l•olla ,\,96.0 .•. ~41\ .. P.·· . ~hlll!~d onl~ In Area ~~o. PRilS1D£11T!i--U, S• '-' tte Col C ~l!tl.l!!!!. 111. Re.hr.c.nc!\ ~! .. cove ~40• U,S,--HISTO.RY ~:;s~:~.:sQcJ/.1. j.IJI:: ~ND (;~STQ/1.~ 1..l.t.l!! . ' . !(~J c:, _J, !'\,. ~(·~ ~.·,1, K.:t ~.·3. c::~.1 ICa v_._4 .~.1, Kt ~~~ ~~ .J. 1\,. v.,G c:.l 1\ tl78o:S91\tl 1.94?. fa.J.I.q 811ttertletd 1 Roger ;fh~ :"""'rl.c•ll Pa~t i . •. ~lstOI')I ot 't;he Uri (ted StAt."s r:r~ .... Con~o.rd ,tP .H.I.~o.shlu, .. 1776:-1~4~ .• . B:v: ,R.O!J~r But.t.e.rt.l•~.l~~ SJ.aon •11cl Schuster, 19'17. 47!1 P• t lllu•• tl:~:~;:n~~~g~:~~:P l<:T~RfA~i; Wil~i';fi . :r.t..t.le 1!1~ ... .- . ~'""' ~I • ··~ ·· : • ,1 ' ~:·· ~ ' ~ ; 0 1!-t\) " \D }., \)( t·..;-o . . ~ : l I ~ ... • ! • u· , : ..... 1: I I ._.~ j J .. .- > l:z:j 0 0 ;>~-' ~ I ~ ..... en ~ -it a ~ 0 :I: z en 0 z t!Z 44 Journal of Library Automation Vol. 1/ 1 March, 1968 imprinted on the spine. Fifty sets of the catalog were ready when the building opened in November, 1966. The shelf list, printed in four copies, contained 3,261 pages. Each set required seven binders. In view of the fact that activation of the Library continued through the first year of opening, the collection grew at a much greater rate than is anticipated for subsequent years. Hence when the building opened, there was available besides the annual catalog a first supplement, in ten copies, listing the 4,000 titles cataloged from July through September, 1966. Although it was proposed originally to prepare monthly supplements, factors of cost and staff time led . to acceptance of quarterly supplements. The second supplement, issued in January, 1967, included the 8,000 titles cataloged from July through December, 1966. A third supplement, issued in April, 1967, included the 12,000 titles cataloged from July, 1966, through March, 1967. The April supplement had 1,934 pages in its author & title section, 1,206 pages in the subject section, and 1,752 pages in the shelf list. The major drawback to an off-line, batch-process book catalog is that it is an obsolete document when produced. This was especially true during the first year when the library grew at the rate of 100 volumes per working day. As a partial remedy to this situation a brief, dated catalog card accompanies each book cataloged for the Meyer Library. In- formation included consists of call number, author, and title. This card is placed in an alphabetical file at the reference desk and purged when a new supplement is issued. The Meyer Library s~aff considered the ten copies of the supplement inadequate for use in the building; during the second year twenty copies of each supplement are to be prepared by running the print program twice. Supplements in the second and succeding years will, of course, be considerably shorter than those issued in the first year. THE SECOND ANNUAL CATALOG Preparation of the second annual catalog began in the spring of 1967. This catalog lists the 41,000 titles cataloged as of the end of June, 1967. The first procedure was to emend the 1966 tape by purging the entries to be changed or deleted; corrected entries come in with new data. In July the information for titles cataloged from April through June was transferred to magnetic tape and merged with the data in the April, 1967, supplement. All programs were run through the Author-Title and Subject Update Program and at that point merged with the emended catalog from the preceding year, the Split Program was run, and the pages for a new catalog were created. As in the preceding year, library staff scanned the completed pages and masked noticeable errors. The Photo Reproduction Service printed A Book Catalog at Stanford/ JOHNSON 45 75 ·copies of each page during the first half of August, and the Stanford University Press bound the catalog during the following month. Com- pleted sets of the catalog were delivered on September 20, 1967, a week before classes were to begin for the new academic year. The 1967 catalog is 4,612 pages in length-2,683 pages in the author & title catalog, divided into five volumes of 530 pages each, and 1,929 pages in the subject catalog, divided into four volumes of 480 pages each. As in 1966, there is a title page and an explanatory introduction in each volume. Floor plans of the library are on the end sheets; and imprinted in gold on the spine are the letters included in each volume, there being clean alphabetical breaks between volumes. The second annual shelf list is 5,634 pages long, and each of the four copies requires eleven binders. Some confusion resulted in 1966, when both author & title and subject catalogs were bound in the same color. In 1967 the author & title catalog was bound in tan bookcloth and the subject catalog in light green. MACHINE TIMING As the above figures demonstrate, the 1967 edition of the book catalog is no brief document, Similarly, time required to process the information on the computer was not brief. As stated earlier, the addition of the expanded print chain considerably reduced the speed of the line printer. Instead of printing in excess of 600 lines per minute, the printer speed was reduced to 250 lines per minute. This speed was determined by timings made of the print programs. To print each page in the annual author & title and subject catalogs, eighteen seconds were required. To print each page in the supplements or in the shelf list, ten seconds were required. Thus, for example, to print the 4,612 pages in the 1967 annual catalog, twenty-three hours were required on the computer printer. In processing the supplements and annual catalogs, it has now become necessary to talk of time required for processing in terms of hours and not seconds or minutes. During the preparation of the 1967 annual cata- log timings were made of the various internal programs, whose output was magnetic tape and which were not tied to the mechanical limitations of the line printer. Sample times are shown in Table 4. Table 4. Program Running Times Format and Update Program ( LB020) Shelf List Split Program ( LB030) Author-Title and Subject Update Program (LB050) Author-Title and Subject Split Program ( LB060) 6.5 hours 11.2 hours 3.7 hours 28.5 hours Throughout the year time is required on the computer for the prepara- tion of edit lists. Timing was conducted for this particular program as well For each 100 records entered, four minutes of machine time are required to prepare an edit list. 46 Journal of Library Automation Vol. 1/ 1 March, 1968 · The computer employed for the project is a university facility, and the Library was billed for its use at the rate of $32.00 per hour. The Library receives a monthly statement for various charges from the Ad- ministrative Data Processing Center, and these have served as one basic record to employ in calculating the actual costs of the book catalog. COSTS The determination of actual costs is a difficult undertaking, and a mean- ingful comparison with costs estimated during the planning process is filled with problems, uncertainties, questions of definition, etc. In a sense, it is impossible to make a meaningful comparison. An element measured during planning is not the same as the element actually achieved. For example, during the early planning stages, before systems design actually began, there was no clear plan for the shelf list nor idea of what its role would actually be. The shelf list as finally designed and implemented is a far more sophisticated document than was then visualized. Second, there was no clear thought given to the inputting of separate records for each added volume or copy of a given item in order to achieve an in- ventory control document as well as a classed listing of items in the li- brary. Third, there was no clear determination as to the length of the sort field required and its effect on processing. Fourth, a principal study . conducted to justify the book catalog compared its projected costs with the costs of three dictionary card catalogs in the new library. Although the book catalog was implemented, the three card catalogs were not, and we have no idea as to the accuracy of our calculations of their cost, even though our experience with the preparation of card catalogs is greater. Fifth, cost studies were based upon the preparation of a 40,000- title catalog as the first product. This was an unrealistic assumption to make, because the library was to open with only 25,000 titles in its col- lection. Given such reservations and conditions, an effort has been made to summarize estimated costs and so attempt an understanding of how they compare to actual costs. Even the determination of actual costs is diffi- cult. It must be borne in mind that the complete operation was per- formed "in house." Cost statements thus omit considerations of such nec- essary factors as overhead and considerable administrative supervision. For example, during the second year the Library was not charged for program maintenance, a signi.Jicant contribution from the Administrative Data Processing Center. Initial planning was based upon preparation of a 40,000-title ( 60,000- volume) catalog, ·and it is possible to present cost approximations in two sections, the first recording · costs required to prepare 50 copies of the 25,000-title catalog issued in 1966; and, second, the additional costs re- quired to input the next 16,000 titles, issue three supplements, and pre- A Book Catalog at Stanford/JOHNSON 47 pare 75 copies of the 41,000-title ( 60,000-vohime) catalog issued in 1967. They are shown in Table 5. . . Table 5. Cost Approximations July 65-Aug. 66 Sept. 66-Aug. 67 Input (@ $.40 per title) $10,000 $ 6,400 Programming Computer charges Edit lists Test catalogs Supplements Annual catalog Reproduction Binding . 5,945 3,000 4,000 2,500 4,570 1966 ( 350 vols.); 1967 ( 675 vols. ) 805 Binders for shelf list and supplements 84 Totals $30,904 1,660 4,460 4,950 5,270 1,690 300 $24,730 . If we eliminate the costs directly related to the production of the 25,000-title catalog, we may be able to isolate the cost of the 41,000-title catalog issued in 1967. This calculation is subject to a certain amount of error, because some processing done in preparation of the 1966 catalog was used again in 1967. This may be compensated for, however, by the time required for the utility programs to emend and delete items from the 1966 tape. Test catalogs and their cost were not considered in early planning, and so their $4,000 cost is eliminated as well. In Table 6 below the actual costs for the 41,000-title catalog so de- rived are compared with the estimates prepared in the fall of 1964 and the estimates offered in April, 1965, at the conclusion of the preliminary systems design. Various adjustments have been made so that these figures are as comparable as possible. For example, the systems estimate did not include a cost for inputting, and this has been added. The actual figures have been adjusted to include costs only for the printing and binding of fifty sets of the catalog instead of the seventy-five which actually were prepared. Although the December, 1964, estimate included under com- puter charges a factor for supplements, these are not included in the systems estimate or actual charges. The format of the supplement par- ticularly became so sophisticated in design and implementation, both in format and number of copies, that this discrepancy is minimal. These figures necessarily cannot be precise, but they give some magnitude of the work undertaken. Although the cost figures indicate that the actual cost was more than fifty per cent greater than actually estimated in 1964, it does remain close to the estimate prepared in the systems design. The chief reasons for the discrepancy may be summarized as the underestimation of the amount 48 Journal of Library Automation Vol. 1/ 1 March, 1968 of machine time needed for the various programs, the underestimation of the programming job involved, underestimating the charges for edit lists; and, most important, the design of a system that was very much more sophisticated than that originally foreseen in 1964. Table 6. Comparison of Estimated and Actual Costs Input of 40,000 titles Computer charges Reproduction (50 copies) Binding Programming Totals Dec. 64 April 65 Estimate Estimate $11,060 $16,647 1,750 8,595 4,324 4,500 . 3,750 2,385 3,000 6,000 $23,884 $38,127 Actual Costs $16,400 9,610 5,115 1,600 5,945 $38,670 Even though costs were greater than expected, one estimate did hold up, namely the time required to complete the job. Delivery and installa- tion of equipment, programming, program testing, inputting of data, re- production, and binding-all were on schedule with only minor slippage that did not affect the completion date of the overall job. THE FUTURE The publication of the second annual book catalog coincided with the completion of the library's activation project. Continued work on the addition of materials to the new library has been assigned to existing divisions within the University Library. Inputting of cataloging informa- tion for the Meyer Library and preparation of supplements to the book catalog and of new annual catalogs are now functions of the Catalog Division. Growth of the library will henceforth proceed at a slower rate, with from 5,000 to 8,000 titles being added each year. The first by-product of the system has appeared-a listing of serial publications in the library for use in ordering and claiming operations. Even as the first annual catalog was being prepared, the feeling was expressed that the equipment employed (the IBM 1401 Computer) was not adequate in an economic sense to undertake this mission for increas- ingly larger masses of material. This · feeling became clearer with the preparation of the second edition. Looking to the future, we see at pres- ent several paths we may follow. First, studies are under way on the conversion of the book catalog operation to larger equipment, in this case an IBM System/360, probably linking it to the overall library program of automation. Not only might this change entail use of more powedul equipment for the ·off-line proc- essing necessary to prepare a book catalog, but there may be possibilities as well of instituting on-line inquiry. There could thus be eliminated the problem of supplements and time-lags. A Book Catalog at Stanford/JOHNSON 49 Second, preliminary inquiries have also been made on the use of the existing tapes in computerized typesetting equipment. The hoped-for re- sult would be the achievement of graphic arts quality on the book catalog page and less bulk to the completed catalog through the greater legibil- ity and greater density thus realized. CONCLUSION Such success as this project has achieved may be attributed to a num- ber of factors: The entire operation was performed "in house;" we were able to draw upon the skills of many staff members on the Stanford campus-in the Library, the Administrative Data Processing Center, the Photo Reproduc- tion Service, the News and Publications Office, and the Stanford Univer- sity Press. IBM representatives, and particularly the systems engineer assigned to the project, gave considerable impetus and guidance to the under- taking. Equipment was delivered on schedule and functioned well. There was a particularly harmonious and understanding working rela- tionship achieved among the many participants in the project, and ad- ministrative support from Library and University officials was constant. There never was any problem in gaining access to the computer, and the staff responsible for its operation gave devoted service in the prepara- tion of the catalog. - Through a happy combination of circumstances, sufficient lead time was available for the project to be completed on schedule. When it became obvious that we should exceed the cost estimates originally prepared, Library funds were available to continue the work. It is clear from student reactions that the book catalog is a useful tool in the new library, and it is hoped that the experience here recounted will prove valuable to the profession at large. REFERENCES 1. McCune, Lois C.; Salmon, Stephen R.: "Bibliography of Library Au- tomation," ALA Bulletin, 61 (June 1967), 674-94. 2. Weber, David C.: "Book Catalog Trends in 1966," Library Trends, 16 (July 1967), 149-64. 3. Freitag, Wolfgang M.: "Planning for Student Interaction with the Library," California Librarian, 26 (April 1965), 89-96. 4. Hayes, Robert M.; Shoffner, Ralph M.: The Economics of Book Catalog Production, a Study Prepared for Stanford University Li- braries and the Council on Library Resources (Sherman Oaks, Calif.: Advanced Information Systems Division, 1964). 5. Hayes, Robert M.; Shoffner, Ralph M.; Weber, David C.: "The Eco- nomics of Book Catalog Production," Library Resources and Techni- cal Services, 10 (Winter 1966), 63-65, 90. 50 Journal of Library Automation Vol. 1/ 1 March, 1968 6. Parker, Ralph H.: ·"Book Catalogs," Library Resources and Technical Services, 8 (Fall 1964), 348. 7. Simonton, Wesley: "The Computerized Catalog: Possible, Feasible, Desirable?" Library Resources and Technical Services, 8 (Fall1964 ), 403-405. 8. Avram, Henriette D .; Guiles, Kay D.; Meade, Guthrie T.: "Fields of Information on Library of Congress Catalog Cards: Analysis of a Random Sample, 1950-1964," The Library Quarterly, 37 (Aprill967), 190-91. 9. "Rule 134," Anglo-American Cataloging Rules. North American Text (Chicago: American Library Association, 1967), pp. 196-97. 10. "Rule 3:6,'' Rules for Descriptive Cataloging in the Library of Con- gress (Washington, D. C.: U. S. Government Printing Office, 1949), p. 14. ll. Hines, Theodore C.; Harris, Jessica L.: Computer Filing of Index, Bibliographic, and Catalog Entries (Newark: Bro-Dart Foundation, 1966)' p. 18.