College and Research Libraries Cataloging U.S. Depository Materials: A Reevaluation Alice Harrison Bahr Automation is reshaping the rhetoric and content of an old debate: to catalog or not to catalog federal publications. The availability of quality records online, keyword search capabilities of online catalogs, and other developments argue persuasively against patent acceptance of old premises. This article reviews previous arguments, discusses conditions redefining them, and proposes partial guidelines for reevaluating current U.S. depository cataloging policies and practices. ataloging United States govern- ment depository materials is an old debate. Beginning in the 1930s and continuing into the 40s, it was sparked by an increase in the number of depository materials. From 400 in 1900, for example, the number rose to 4,300 by 1930. 1 Typically, numbers dic- tated organization. Libraries that selected small numbers cataloged them, a practice endorsed for the smaller library by Mary Hartwell, cataloger for the Office of the Superintendent of Documents. 2 Those that received larger numbers maintained separate collections arranged alphabeti- cally by agency, by type of material, or by Superintendent of Documents (SuDocs) classification number, thus organizing materials br, agency and series rather than by subject. The latter system was particu- larly easy, since materials were sent to de- pository libraries with shipping lists sup- plying SuDocs numbers. Other libraries used both approaches. They cataloged some materials and placed others in sepa- rate collections. In time, arguments for the two basic ar- rangements became set. As Waldo points out, their bases "were merely unsup- ported opinions and assumptions.'' 4 Sep- aratists pointed to the shortcomings of the card catalog: its inability to .index serial publications, its paucity of subject head- ings, and its difficulty of use, especially for the user confronting the unwieldy U.S. author drawer. There were shortcomings with cataloging itself. It was not suitable for all materials, particularly pamphlets and posters. It imposed a classification system, either Dewey or Library of Con- gress, that precluded arrangement by agency, useful to some researchers. It in- creased delays between receipt and avail- ability of materials. It was expensive. Quality records were sparse, necessitating original cataloging, and frequent title and agency name changes required multiple record handling. Relying on higher- quality indexes, with better quality index- ing, provided better access to materials and information. Promoting staff familiar- ity with government materials, separate collections improved the quality of refer- ence service. On the other side, advocates of catalog- ing acknowledged its expense, but justi- fied cost on the basis of improved, simpli- fied access. Cataloging spared users the aggravation of consulting several indexes and the annoyance of learning a second classification system. It also provided im- mediate feedback on library holdings. In Alice Harrison Bahr is project librarian for online systems at the libraries at Cedar Crest and Muhlenberg Col- leges, Allentown, Pennsylvania 18104. 587 588 College & Research Libraries short, it was expensive and time- consuming, but resulted in better service. The old arguments are changing. Tech- nology is bypassing them and raising new questions. The availability of Government Printing Office (GPO) cataloging on net- works, record retrieval on some networks by SuDocs number, and online catalogs and their related use studies both beg the assumptions of the past and highlight new needs. Among these are the need to keep abreast of technological develop- ments, to reevaluate present policy in light of those developments, and to for- mulate policies based on both demon- strated user needs and technological de- velopments. TECHNOLOGICAL CONSIDERATIONS Technology is reshaping old arguments by creating access other than through the card catalog, by streamlining cataloging procedures, and by creating a new form of the card catalog. Bypassing traditional card catalog access altogether, for exam- ple, Bruce Morton made use of available software and Carleton College's DEC V AX-11 to produce a SuDocs shelflist ac- cessible not only by SuDocs number but by title, keyword, and Boolean opera- tives. 5 This is an exception, however; most libraries have relied on online vendors, national agencies, and bibliographic utili- ties to create new options. The basis for most new possibilities is cooperation among the Government Printing Office (GPO), Library of Con- gress and Online Computer Library Cen- ter (OCLC). Deciding to speed up produc- tion of its Monthly Catalog, the official catalog of U.S. government congressional and departmental publications, GPO be- gan cataloging on the OCLC shared cata- loging network in 1976. To do so, it had to abandon its local thesaurus and conform to Anglo-American cataloging rules and Library of Congress subject headings. 6 Conformance ushered in GPO's new role as cataloging authority for U.S. govern- ment materials, which became official on January 2, 1981. GPO's use of OCLC pro- duced the major benefit to libraries: avail- ability of a substantial number of timely, high-quality records. The type of records available has ex- November 1986 panded, and reported hit rates are excel- lent. In October 1981 GPO assumed cata- loging responsibility for materials distributed by the National Audiovisual Center. 7 In 1982 the University of Wash- ington reported finding 75 percent of ma- terials on OCLC. Its hit rate for U.S. gov- ernment materials was 95 percent. 8 A 1985 Muhlenberg College study reported a 96 percent hit rate for selected categories of U.S. materials (see table 1). Backlog at GPO is not significant. As soon as GPO catalogs materials, records are available on OCLC. In 1984 a one- month study of "priority one" cataloging items (congressional materials) indicated that 72 percent were input in 15 days and average input time was 24.4 days. 9 The SUNY/Potsdam Library (State University College at Potsdam, New York), which catalogs U.S. materials on OCLC, begins searching one to two months after receipt of documents. 10 Cooperation -between GPO and biblio- graphic utilities has increased the number of quality records available to libraries rel- atively soon after publication. It has also created new means of access to those rec- ords and new cataloging services. In 1980, OCLC surveyed libraries to determine their interest in an even quicker cataloging process, DARP (automatic distribution of cataloging). Targeted initially for regional depositories (those libraries receiving all available depository materials), the ser- vice would have created individual cata- log records for libraries receiving deposi- tory series. Cataloging would have been automatic, completed as soon as GPO cat- aloged an item in a series but due to the cost of the service, DARP never got off the ground. There is a difference between having records available and having quick, con- venient access to them. For cataloging purposes, access by SuDocs number is de- sirable. It allows catalogers to work di- rectly from shipping lists to search for rec- ords. OCLC made SuDocs number access (OCLC's gn: government number key) available in 1980. In 1985 it became possi- ble to search for records by any SuDocs number attached to a title, as of the point of cataloging. SuDocs numbers change when an issuing agency changes, which Cataloging U.S. Depository Materials 589 TABLE 1 OCLC Record Availability P840601-1 to P840831-5#* M840601-1 to P840831-8:j: Total P840904-1 to P841123-4 M840904-1 to M841129-5# Total Combined Total 97 12 109 66 14 80 189 *P = paper shipping lists . Available Series Cataloging (#) 4 4 5 5 9 Hearings 27 10 37 15 9 24 61 Geological Survey Professional Papers(#) 12 6 6 18 Geological Survey Bulletins (#) 6 6 8 8 14 Non-GPO Record (#) 15 (14/DLC)t 12 27 (15/DLC) 10 (7/DLC) 8 (1/DLC) 18 (8/DLC) 45 (23/DLC) NoOCLC Record (#) 4 4 3 3 7 OCLC Record Availability (%) 96% 100% 96% 95% 100% 95% 96% tDLC = th e number of non-GPO records that are Library of Congress records . :j:M = microfiche sh ipping lists. means one title may have several SuDocs numbers. Both searching capabilities en- hance access to records. Another benefit of GPO's use of OCLC is the generation of machine-readable rec- ords on tape. Used to produce Monthly Catalog, the tapes have other uses. They are loaded regularly onto other shared cat- aloging networks like the Research Li- braries Information Network (RUN) and Western Library Network (WLN) to pro- vide users of those systems with access to the same GPO records as OCLC users. Availability is not as immediate as in OCLC, since GPO tapes are loaded monthly, and these networks do not yet provide SuDocs number access. 11 Tapes have another potential use for li- braries using or planning to implement an online catalog. They provide an alterna- tive means of including GPO records in a local online catalog. One possibility is to download records from a shared catalog- ing network by means of an interface. The other is to load GPO tapes into the online catalog. This requires tape manipulation both to load the tapes and to eliminate rec- ords neither received nor cataloged, i.e. maps, serials, etc. Cooperation between national agencies and shared cataloging networks weakens the argument that quality records are un- available, a main pillar in the argument against cataloging U.S. materials. Simi- larly, the existence of the online catalog erodes another premise, that excessive record handling contributes to the high cost of cataloging. In manual catalogs, rec- ord changes require pulling, editing, and refiling cards, usually seven to a set. An online catalog eliminates these steps. While this is a benefit regardless of what materials are being cataloged, it is a partic- ularly strong one for depository materials, which undergo frequent main entry and title changes. Those frequent changes are inherent in the nature of both serials and government organizations. While little can be done to eliminate the former, technology can help simplify the latter. Some online catalogs incorporate authority control for personal authors, and a few offer it for corporate authors and series. If the catalog also of- fers online global change, one change to an authority record will automatically al- ter in the same way all affected headings throughout the database. A common argument against cataloging U.S. materials is the paucity and inade- 590 College & Research Libraries quacy of subject headings. Online cata- logs have the capacity to increase the tra- ditional author, title, and subject heading access points. To search the manual cata- log successfully, users must know precise authors and titles, guess the correct sub- ject term, or know how to use LC Subject Headings volumes. Most online catalogs are more forgiving. They provide key- word access to some or all fields as well as SuDocs number access. The latter is in- creasingly useful as more and more sources include these numbers. Public Af- fairs Information Service announced re- cently that 65 percent of indexed U.S. gov- ernment materials include SuDocs numbers. By simplifying record changes, elimi- nating filing, and increasing access points, online catalogs can facilitate a variation of the old either/or proposition to catalog U.S. materials or to house them sepa- rately. These catalogs make it easy to com- bine approaches, namely, to catalog mate- rials, classify them by SuDocs, and yet keep them separate. The same arrange- ment is possible in a manual catalog, but filing alone argues against it. The library at SUNY/Potsdam, which has cataloged on OCLC and classified by SuDocs since the 70s, reduces filing by cataloging selec- tively. Technology, of course, is not a panacea. An automated SuDocs shelflist has the ad- vantage of creating more access points than a traditional card catalog, but it also forces users to consult more than one cata- log to locate information. The availability of records on shared cat- aloging networks has been an important technological advance. However, not every library uses one, and not every net- work offers SuDocs access. For libraries using networks, whether or not SuDocs access is available, there are other possible constraints. GPO is slow to catalog certain materials. 12 Possible downscaling of cata- loging might also present problems. There has been no report about GPO's 1982 dis- cussions on the possibility of downscaling AACR2 cataloging to augmented level II and no longer providing corporate and personal name authority work. 13 Specific GPO practices may be of greater concern. For example, GPO provides no November 1986 collective series entries for titles in mono- graphic series, only individual analytics. This means, for example, that if a library staff decide that a few subject headings will suffice to lead users to the content of U.S. Geological Survey Professional Pa- pers series, they must, if using GPO copy, provide individual analytics for titles in those series. The Depository Library Council, an advisory body comprised of li- brarians, has repeatedly passed resolu- tions requesting series cataloging, but offi- cial GPO response has been that current practice conforms to depository law. This problem is not insurmountable, however; it requires only that a library create a few collective series entries. Technological changes are catalysts. They provide new possibilities and chal- lenges, not perfect answers-microfiche records provide an example. GPO cata- logs only paper copies of publications that are converted to microfiche for depository distribution. A note indicates that distri- bution to some depositories was on micro- fiche, and the word microfiche appears af- ter the item number. However, libraries with substantial fiche collections might not welcome the editing required. Ex- changes between the Depository Library Council and GPO have thus far produced only the following resolution: GPO's pol- icy is in accord with Title 44 of the U.S. Code, and librarians are free to modify rec- ords.14 Actually, a number of libraries do so, creating suitable records in most cases. This is especially true for government hearings. Shipping lists from GPO are not always correct, creating another cause for concern depending on the delay between receipt of lists and actual cataloging. Revisions to shipping lists, called "corrections lists," are issued periodically to clear up incor- rect SuDocs numbers, typographical er- rors, and wrong item numbers (the num- bers used to order depository series). GPO tape users face greater problems. For example, tapes include all GPO cata- loging, and unless a library has received all these materials (some of which are non- depository, that is, not offered to deposi- tory libraries), records must be eliminated either in-house or by a vendor. Regardless of who processes the tapes, records are Cataloging U.S. Depository Materials 591 usually extracted by item number. Since it is not uncommon for depository libraries to fail to receive some items ordered, rec- ords may be created for titles not in fact held. The result is like cataloging on the basis of what is ordered instead of what is received. In addition, tape loading on an online system requires manipulation of data. Not every library has, or plans to have, an online catalog. Even for those who do, GPO tapes present additional difficulties: they come without documentation. Con- trol numbers like OCLC, SuDocs, techni- cal reports, and contract numbers are ei- ther omitted or appear in incorrect fields. 15 Correction tapes are not issued. Conse- quently, errors must first be detected and then corrected manually. There are multi- ple records for errata slips and periodical issues, and not all series, subjects, and names conform to LC form. Fortunately, many of the GPO tape er- rors and limitations have been corrected. The Depository Library Council has been requesting since spring 1979 that GPO convert traced names, series, and subjects to LC form; add control numbers such as item, SuDocs, and stock numbers (used for direct purchase of materials) to appro- priate fields; and correct typographical and other errors. 16 The Public Printer an- nounced at the fall 1984 Council meeting that the project would be undertaken and supervised by Judy Myers at the Univer- sity of Houston. 17 Funds were never allo- cated. One year later Brodart announced its GPO file, which includes LC subject and name headings and control number corrections. Correction pages from annual Monthly Catalog volumes are being keyed in. While Brodart is still investigating pro- cedures to handle separate serial records, its GPO file offers dramatic evidence of how quickly technology changes situa- tions. COST CONSIDERATIONS The availability of quality records in var- ious forms, a major obstacle to cataloging, is part of a larger consideration: cost. Even if quality records are available, are they af- fordable? The assumption has been no, despite the absence of comparative cost data for processing, maintaining, and su- pervising separate collections. Fortu- nately, new possibilities have been ex- plored with an eye to reducing expenses. Available cost data highlight the weak- ness of previous assumptions and also un- dercut them. The cost of Carleton College's auto- mated SuDocs shelflist was determined by counting the bytes per record and then calculating the required disk space for a given number of records. At 310 bytes per record, the storage cost for 431,600 records was $8,500, the same price as a 260,000- block disk and drive. 18 Disregarding input time and use and storage costs, all of which were absorbed by the Computer Center, the unit cost for the shelflist was $1.98, slightly higher than OCLC first- time use charges. 19 In 1984 Bowerman and Cady, suspect- ing libraries might include records in their online catalogs if they were available at ''a small fraction of the cost incurred in a tra- ditional cataloging environment, " 20 com- pared costs for various methods of obtain- ing records. The researchers took a sample .from a GPO test tape to develop a cost- effectiveness model on which to base a comparison of four ways of obtaining machine-readable records. Two of the four ways involved tapes: the first manip- ulated in-house, the second by Marcive, a commercial firm that strips records from tapes for libraries. The other two options involved automated cataloging services, either online through OCLC' s network or offline, inputting records on a microcom- puter and sending them to Library Sys- tems Services, Inc. (LSSI) for tape or card production. The results were as follows: a library cat- aloging approximately twelve thousand publications (about 35 percent of available items exclusive of maps, serials, and mi- crofiche) would spend about $1,500, $1,800, $17,000, and $2,600 to obtain rec- ords from GPO, Marcive, OCLC, and LSSI, respectively. 21 Record extraction by Brodart, unavailable at that time, reduces costs further: $600 for twelve thousand records ($.05 per record), excluding tape and profiling charges. While the study notes the need for data processing personnel for in-house manip- ulation and raises appropriate questions 592 College & Research Libraries about quality control in Marcive and LS.SI processing, it confines costs to the acquisi- tion, not the storage, of records. 22 Regard- less, it dispels the the idea that under any and all circumstances cataloging U.S. ma- terials is prohibitively expensive. USER CONSIDERATIONS Under some circumstances, then, cata- loging of GPO materials is affordable. But is it desirable from the users' perspective? Early studies indicate catalogs are under used and misused. Recent studies reveal little enthusiasm for the enhanced search capabilities of online catalogs. One use study indicates certain faculty locate gov- ernment publications outside of the li- brary. All studies indicate that much more needs to be known about the use of li- braries, government materials, and cata- logs before a decision is made about cata- loging U.S. depository materials. According to previous surveys, only 59 percent of library patrons use the cata- log. 23 Most catalog users look up one entry and stop. 24 They locate correct subject headings only 50 percent of the time. 25 In 1958 ALA surveyed 5,494 catalog users in thirty-nine libraries. Results showed a fail- ure rate of 20 percent for known-item searches and 13 percent for subject searches. 26 The capabilities of the online catalog are not as important to user satisfaction as suspected. The Council on Library Re- sources (CLR) supported a 1982 study of twelve thousand online catalog users in twenty-nine libraries. Eighty-five percent reported finding some or all of what they were searching for, and subject searching was of greater interest than the ALA study concluded. 27 Online catalog capabilities such as keyword access and search qualifi- ers by date, language, and Boolean opera- tives, however, were not perceived uni- versally as benefits. The CLR study found that language and call-number search lim- its had a slightly negative effect on satis- fact~on and that keyword and Boolean op- eratives were unrelated to satisfaction. 28 Some users bypass the catalog alto- gether. In his 1984 study of government publications use, Peter Hernon indicated that "academic social scientists rely upon November 1986 their subject literature and interpersonal sources (e.g., colleagues) for awareness of source material. They do not make exten- sive use of indexing and abstracting ser- vices and bibliographies housed in li- braries. " 29 Frequently, they obtain materials from outside the library. Rather than making definitive state- ments, these studies point out how little is known and how much remains to be seen. After examining two hundred catalog use studies in the most extensive book on the subject, Redesign of Catalogs and Indexes for Improved Subject Access, author Pauline A. Cochrane concludes, "we come away from catalog use studies quite discour- aged about the present state of use of our catalogs, with no sure guidance about im- provements and the impact of changes. " 30 No evidence supports the assumption that cataloging is always desirable. No substantive evidence supports the oppo- site view. Technological changes make that clear. They encourage investigation of present situations, their corresponding possibilities, their costs, and their effec- tiveness in meeting user needs independent of old assumptions. For single answers, they substitute questions. PARTIAL GUIDELINES FOR REEVALUATING CATALOGING POLICIES The old question of whether or not to catalog U.S. depository materials raises no single query. It was made into one by technologies that limited options, ele- vated and fixed cost considerations, and made assumptions about user needs. Newer technologies suggest several ques- tions: they focus on current and future levels of automation, available technolo- gies, and the need to know more about collections and their use. 1. Does the library catalog on a shared cataloging network? If so, what access does the network provide to GPO rec- ords? 2. Does the library have or plan to im- plement an online catalog? Will it permit keyword, SuDocs number searching? Does it include authority control and global change capability? 3. How does the library's clientele learn Cataloging U.S. Depository Materials 593 of government publications (e.g., through journal articles, colleagues, indexes, GPO sales brochures, news broadcasts, etc.)? 4. How does the library's clientele look for and obtain government publications? Do patrons look in the library? 5. What are the major strengths of the government publications collection (e.g., congressional materials, geological mate- rials)? 6. Is there a clientele for those collection strengths? 7. Is cataloging the best means of ad- dressing the needs of that clientele? 8. Is manpower available to define the means and related costs of cataloging? 9. Will cataloging be selective? If so, what categories or types of materials will be excluded? No single library can address all these questions, but, for some, GPO shipping lists may provide one option for testing as- sumptions, exploring possibilities, and determining costs. What can shipping lists do to answer these questions? For RUN and WLN us- ers, they can test the speed with which records can be searched without SuDocs access. For all shared cataloging network users, they can help determine network costs for cataloging U.S. materials. (See table 2.) They can help identify the type of entries available for specific depository se- ries (analytics or series), their timeliness, and their sources. The 040 field contains the cataloging agency. If GPO appears first, it was the first cataloging agent. If it appears elsewhere, GPO modified an ex- isting record. If the Library of Congress has modified a GPO record, DGPO/GLC will be found in this field. Finally, for both network and nonnetwork users, shipping lists can assist in identifying collection strengths, providing crucial direction for determining what will be cataloged, and forming policy that reflects those deci- sions. Realizing that the card catalog may not be the single most appropriate vehicle for locating government materials, that not all materials (e.g., posters, pamphlets) merit cataloging, and that cataloging all materi- als increases cataloging costs, a number of libraries have either established or recom- mended policies for cataloging select cate- gories of U.S. materials. Peter Graham recommendeds that large research li- braries concentrate cataloging efforts on collections that are not indexed. 32 At the University of Houston, priority is given to "publications of agencies that the average user does not know are government agen- cies. " 33 Whether policies are set or explored, careful arranging and coding of shipping lists can verify the feasibility and costs of policies. Shipping lists must be represen- tative of type and extent of receipts. Monthly statistics can establish the latter. Then lists can be separated into microfiche and paper piles to examine each format separately and into chronological piles to test the timeliness of available records. Coding should reflect specific interests. For instance, items received on deposit can be checked, those considered for cata- loging circled, and candidates for series cataloging marked with an S. Special types of materials that could dramatically increase cost if cataloged may be studied separately (e.g., hearings might be pre- ceded by an H), and cataloging agent can be indicated as GPO, LC, or Other. Such an examination begins to provide a picture of what a library can afford to cata- log on the basis of type of catalog records TABLE 2 OCLC Cataloging Costs Estmated Extended Extended Number of Extended OCLC Card OCLC FfU Cataloged UnitOCLC OCLCFfU UnitOCLC Costs and Card Titles FfU Cost Cost Card Cost (dollars-7 cards Costs (a nnu al) (dollars) (do llars) (do ll ars) per title) (dollars) 400 1.47 588.00 .0495 138 .60 726.60 Notes: An online catalog wo uld eliminate the need for cards and reduce cos ts to $588 . No reliable es timate of staff costs exists . With minor record changes, most titles cou ld be processed in five minutes or less . At $10 an hour, 300 hours of professional cataloging time would cost $3,000 . 594 College & Research Libraries rather than on number of titles; whether records are available; whether they are timely; whether sufficient numbers of rec- ords are available for microfiche; and, con- sequently, whether or not fiche should be included in cataloging programs. For most libraries, whether or not they are able to determine these considerations by search- ing items on a shared cataloging network, shipping lists may suggest criteria for in- cluding or excluding certain types of mate- rials from cataloging. In short, they can be used to qualify general assumptions about the feasibility and affordability of catalog- ing U.S. materials in a specific environ- ment. SUMMARY Arguments against cataloging U.S. ma- terials have assumed that cataloging was difficult in the absence of available rec- ords, time-consuming because of numer- ous title and agency name changes, and expensive as a result. Other negative rea- sons have been that cataloging limits shelf arrangements to Dewey or LC and causes delays between receipt and availability of November 1986 materials to patrons. On the positive side, despite inadequate and difficult entries, cataloging the materials provides the most convenient, direct, and useful means of access. In an automated cataloging environ- ment, all of these assumptions can be dis- proved. Shared-cataloging network users have online access to records; online cata- logs minimize the record-editing process and facilitate the simultaneous, separate SuDocs shelf arrangement and cataloging of materials, thus eliminating the usual delays associated with cataloging: the availability of tape records, vendors who can manipulate tapes, and selective cata- loging reduce traditional expenses. Fi- nally, studies have shown that the card catalog is not as pivotal to all users' needs as was once thought. The new cataloging environment man- dates a new look at an old debate, encour- aging exploration of alternatives and dis- missal of some earlier assumptions. Most important, the online alternatives high- light the central importance of knowing more about the need for, use of, and meth- ods of obtaining government information. REFERENCES AND NOTES 1. Michael Waldo, "An Historical Look at the Debate Over How to Organize Federal Government Documents in Depository Libraries," Government Publications Review 4:319-29 (1977). 2. Ibid., p .320. 3. U.S. Government Printing Office, Public Documents Department, An Explanation of the Superinten- dent of Documents Classification System (Washington, D.C.: Govt. Print. Off., 1963). 4. Waldo, p.322. 5. Bruce Morton, "Implementing an Automated Shelflist for a Selective Depository Collection," Government Publications Review 9:323-44 (1982). In an earlier article, "An Items Record Manage- ment System,'' Government Publications Review 8:185-96 (1981), Morton describes automation of an item file. In'' A Microcomputer in a Micro-Sized Library,'' Small Computers in Libraries5:5-8 (1985), Frederick A. Marcotte describes yet another automated access system, a micro-based, check- in/retrieval system used prior to installation of a BUS integrated online library system at Clermont College (Batavia, Ohio). 6. Loche McLean, "GPO Cataloging and Monthly Catalog Production," Public Documents Highlights 41:1-2 (Aug. 1980) . 7. "Agreement Set on Audiovisual Cataloging," LC Information Bulletin 40:389-90 (Nov. 6, 1981). 8. Sharon Wallbridge, "Government Documents, OCLC, and Research Libraries," Research Libraries in OCLC: A Quarterly 8:3-5 (Oct. 1982). 9. "Cataloging Timeliness Study," Administrative Notes 5:1 (Nov. 1984). 10. Selma V. Foster and Nancy C. Lufburrow Eldblom, "Documents to the People in One Easy Stem-An Update," Documents to the People 8:119-20 (May 1980). 11. While RUN and WLN networks provide subject access, they do not provide SuDocs access to GOP records. Subject access to OCLC GPO records is available for the last few years of GPO rec- Cataloging U.S. Depository Materials 595 ords through BRS. According to the February 1986 OCLC Newsletter, subject access should be available to libraries in early 1987. 12. Cynthia Bower, "OCLC Records for Federal Depository Documents: A Preliminary Investiga- tion," Government Information Quarterly 1, no.4: 379-400 (1984). 13. "Documents Cataloging Manual Committee," LC Information Bulletin 41:285 (Sept. 10, 1982). 14. "GPO Cataloging for Microfiched Documents: Fact Sheet," Administrative Notes 6:20 (May 1985). 15. Janet Swanbeck, ''Government Printing Office Cataloging Tapes,'' paper presented at the annual meeting of the American Library Association, Dallas, June 25, 1984. 16. "Responses to Resolutions," Administrative Notes 5:9 (Oct. 1984). 17. Ibid., p.10. 18. Morton, p.326. 19. Ibid. 20. Rose ann Bowerman and Susan A. Cady, "Government Publications in an Online Catalog," Infor- mation Technology and Libraries 3:331-42 (Dec. 1984). 21. Ibid., p.340. 22. A 354MB disk drive holds approximately 82,000 fully indexed records of 4,000 characters each and costs roughly $20,000. For this configuration, per-record storage costs would be $.24. 23. Joseph R. Matthews, Public Access to Online Catalogs: A Planning Guide for Managers (New York: Online, 1982), p.6. 24. Ibid., p.29. 25. Ibid. 26. F. W. Lancaster, "Studies of Catalog Use," in The Measurement and Evaluation of Library Services (Washington, D.C.: Information Resources Pr., 1977), p.19-68. 27. Robert N. Broadus, "Online Catalogs and Their Uses," College & Research Libraries 44:459-66 (Nov. 1983). 28. Ibid., p.465. 29. Peter Hernon and Charles R. McClure, Public Access to Government Information: Issues, Trends and Strategies (Norwood, N.J.: Ablex, 1984). 30. Pauline A. Cochrane, Redesign of Catalogs and Indexes for Improved Subject Access (Phoenix, Ariz.: Oryx, 1985). 31. Government publications staff discarded a departmental card catalog after a one-month study in- dicated that 30 percent of questions could be answered on the basis of prior knowledge and SO percent through indexes. 32. PeterS. Graham, "Government Documents and Cataloging in Research Libraries," Government Publications Review 10:117-25 (1983). 33. Ibid., p.124.