HAROLD B. SHILL

Open Stacks and Performance

Although the utility of open stack systems has been widely debated, little empirical research relevant to the controversy is available. Using circula­ tion, availability, and search and library-use statistics, major elements of the direct access debate are tested in a six-year study of a library that has recently undergone the transition from closed to open stacks. Direct shelf access, it was found, contributed to an increase in library use and a decrease in c~rculation . Contrary to expectations, book availability perfor­ mance also improved significantly during the study period.

D IRECf ACCESS is a prevailing following the opening of the stacks. practice in American academic Moreover, book availability studies had today and enjoys considerable support been conducted during peak use periods be­ among faculty members, students, and li­ fore and after the decision. Access to this brarians. Despite their popularity, open body of data made possible a study of li­ stack systems have not been incontroverti­ brary performance, using several dependent bly proven more effective than the closed or variables, under changing conditions of shelf semiclosed alternatives as user access access. mechanisms. Nor is the ongoing debate likely to resolve the issue, since the discus­ THE ISSUE sion is largely impressionistic and specula­ The debate over open stacks is multifa­ tive. ceted, encompassing such diverse issues as Considerable research on patron behavior the effectiveness of browsing, the value of is available. However, the few quantitative close classification, storage policy, the edu­ studies relevant to the controversy focus on cational role of the library, and collection user behavior in the stacks rather than the security. It also reflects various conceptions broader issue of open access per se. Infer­ of the patron/collection interface articulated ences about the validity of open stack sys­ by librarians during the past 150 years. tems have been drawn from measures of pa­ Closed stack libraries were predominant tron shelf failure and browsing effectiveness in Europe until the post-World War II pe­ in these reports. riod, with shelved in accession order The effectiveness of direct access could be or by size. This arrangement conserved established from statistical case studies shelf space and made detailed subject clas­ comparing library performance before and sification unnecessary. after the stacks were opened. However, no The demand for general access to the such empirical impact studies were iden­ shelves originated in the mid-1800s, largely tified in a search of Library Literature. as an outgrowth of the public library move­ The opening of the main library stacks at ment. The concept was consonant with West Virginia University in February 1976 democratic theory and was adopted by provided an opportunity for such a longitu­ many college libraries and some university dinal analysis. Complete circulation, build­ libraries, though most of the large research ing use, and search statistics had been kept collections in this country remained closed during the three-year periods preceding and or semiclosed until well into the twentieth century. 1 Harold B. Shill is chief circulation librarian In recent decades, however, most and assistant professor of library science, West academic library buildings have been de­ Virginia University, Morgantown. signed to support open stack operatio~s, !

220 I Open Stacks I 221 and many libraries have liberalized access or when even one class becomes so large that it despite architectural shortcomings. In the cannot be scanned efficiently in a fraction of an 8 latter case, the decision has often been hour. made without full advance appreciation of Ratcliffe contended that a classified sub­ its consequences. 2 ject arrangement is helpful to the browser Many theoretical and practical arguments only in a small or medium-size library. He have been advanced in favor of open access. identified the University of Manchester, In a survey conducted in the 1950s, Hicks with holdings exceeding one million items found widespread expectations that an open in 1969, as a library that has passed. this size stack system would eliminate unnecessary threshold and is gradually restricting patron barriers between readers and the collection, access to its collection. 9 This argument is increase circulation, permit staff reductions, also supported by Hyman's finding that and elicit patron support. 3 In his chapter most librarians consider shelf classification "The Educational Function of the Library," more valuable as a locational device than for Lyle concluded that "open stacks makes subject searches. 10 [sic] possible the intelligent use of library Common to all these contributions, both resources," assuming good library manage­ pro and con, is a virtual absence of empiri­ ment and faculty promotion of systematic li­ cal data to support the authors' contentions. brary research. 4 Celoria, an archaeologist, Factual data relevant to the controversy maintained that researchers can remain ab­ must be sought elsewhere. reast of advances in sister disciplines by mastering the "higher browsing. " 5 Other PREVIOUS RESEARCH RESULTS advocates of direct access insist that it en­ Although the open access question is ad­ hances human dignity and produces a vari­ dressed indirectly in many user behavior ety of other social benefits. studies, Cooper's 1957 article remains the An equally impressive set of arguments only analysis of the impact upon library per­ has been arrayed against the open stack formance of a transition to open stacks. concept. The most familiar of these are cus­ While generally supportive of the Univer­ todial in origin-increases in theft and muti­ sity of Washington's decision, her report lation, reduced book availability due to does identify several dysfunctional conse­ misshelving, greater costs for book replace­ quences of the decision, including increases ments and staff, inadequacies of building in misshelved books, searches, and noise in 6 design, and inefficient use of shelf space. the stacks and initial dissatisfaction among Criticism has also been advanced at a lof­ undergraduates accustomed to book delivery tier theoretical level. Ratcliffe and others, service. Improved patron service and expanding upon the conventional argument greater long-term user satisfaction, a result that students cannot function well in an attributed to the opportunity to browse, are open stack system, maintain that open ac­ cited as benefits of the decision. 11 cess encourages users to bypass the card Several interrelated factors account for catalog and other bibliographic tools. The book availability performance. Buckland re­ consequence of this tendency, they con­ ported that a variable loan policy and tinue, is to impair the library's performance demand-based selection of duplicate copies of its educational role by fostering a decline produced a sharp increase in book in bibliographic research skills and system­ 12 7 availability at the University of Lancaster. atic library use. Saracevic, Shaw, and Kantor found that cir­ A second argument, applicable particu­ culation performance at Case Western Re­ larly to large research libraries, is that the serve University improved from 77 percent utility of browsing in a classified shelf ar­ to 87 percent after the loan period was rangement varies in a roughly inverse man­ shortened. 13 ner with collection size. Drawing upon The effectiveness of library security mea­ search theory, Morse inferred that browsing sures, whether manual or electronic, is an has its own law of diminishing returns: important determinant of a library's ability The trouble comes when the collection becomes to deliver books when requested. Confirm­ too large for all of it to be easily accessible to all, ing a principal finding of the Case Western 222 I College and Research Libraries • May 1980

Reserve study, Smith and Granade discov­ shelves. Conversely, increased shelf disor­ ered that 14 percent of the titles not located der and search failures in a classified collec­ by patrons in the University of Tennessee's tion are identified. as counterproductive re­ Undergraduate Library were in their proper sults of open stack access. shelf locations. 14 This last finding is particu­ Collection access questions should not, larly germane to the open-access debate, however, be regarded as strictly technical since it demonstrates that undergraduates questions that can be resolved in a political encounter some difficulty with classified vacuum. The University of Toronto Library shelf arrangements even in medium-size li­ confronted the latent power of its student braries. and faculty clienteles in 1972, when an at­ Studies of stack use at the Library of tempt to institute a closed-stack policy in a Congress, the University of Chicago's new university library precipitated student Harper Library, and Johns Hopkins Univer­ demonstrations and critical responses from sity indicate that collection size does not the Canadian library community. 21 Faculty deter browsing in an open-stack research li­ members and students, it should be recog­ brary. Dubester reported that 38 percent of nized, constitute interest groups able to the individuals interviewed in the Library of define broad parameters within which shelf Congress stacks were browsing rather than access decisions can be made. searching for specific titles. 15 Data from Fussier and Simon's Chicago study revealed HYPOTHESES that 56 percent of the history and physics Certain conceptions about the conse­ books removed from their shelf locations quences of open access reappear in both had been identified through browsing. 16 In philosophical discussions of the idea and the a follow-up of the Fussier and Simon study, small body of relevant empirical literature. Bowen found that many more books were In particular, one encounters predictions browsed than checked out, with graduate that circulation will increase, while book students comprising the bulk of the open availability will decline. The anticipated di­ shelf user population. 17 The Johns Hopkins rection of user reaction varies with indi­ study conducted in the early 1960s vidual preferences, with advocates of open confirmed that many items wanted by pa­ access predicting increased user satisfaction trons are initially located through brows­ and opponents forecasting a decline. All of ing.18 these implicit propositions are susceptible to The real utility of browsing is questioned empirical testing. The decision to open the in Greene's study of methods used by stacks at West Virginia University's (WVU) Georgia Tech faculty members to select main library provided an opportunity to re­ books from the library. Greene found that formulate these ideas as testable proposi­ browsing was the most frequently cited tions and to examine their validity under method for locating books that were sub­ real-life conditions. sequently borrowed. However, browsing The main library contains approximately was also the least effective method for iden­ 530,000 of the 930;000 physical volumes in tifying books that ultimately proved use­ the WVU collection. Books in the ful.19 humanities and social sciences constitute the The potential difficulty of user adaptation bulk of the collection, since science mate­ from an open stack system to closed stacks rials are concentrated in several branch li­ is raised by a study conducted at the Uni­ braries. The stack area was semi-closed versity of North Carolina. Clay reported prior to February 1976, with admission re­ that 80 percent of a faculty/graduate student stricted to faculty members, graduate stu­ sample interviewed in the stacks felt unable dents with stack permits, and a few under­ to rely on catalogs and as a graduate honors students, The stacks were substitute for browsing. 20 opened with minimal publicity, on an "ex­ The studies cited above indicate that perimental" basis, to permit close monitor­ users will browse in an open stack library, ing of the decision's impact on library ser­ do locate some books specifically through vices and to avoid abrupt changes in use browsing, and do value direct access to the patterns. Book delivery service was main~ Open Stacks I 223 tained for those patrons preferring not to re­ garding patron access to the stacks can be trieve their own books. formulated. Somewhat controlled conditions for analysis existed during the six-year study RESULTS AND ANALYSIS period, since no branch libraries were The hypothesis that circulation levels opened and no policy changes that might should increase with open access derives have affected study results were im­ from the assumption that users will borrow plemented. Enrollment increased from more books if they are permitted to browse, 15,203 in 1973 to 20,964 in 1978, but rather than having their access limited to allowances for this change have been made specific books identified through biblio­ in the analysis. graphic tools. The availability of comparable data for the Given the substantial growth in enroll­ 1973-78 period permits both the testing of ment during the study period, circulation propositions derived from the literature and would be expected to increase even without the investigation of a possible relationship a change in stack access. Table 1, however, between stack access and library use levels. reveals a decline in nonreserve circulation For purposes of analysis, these propositions throughout the study period. This decline is are presented in the form of three testable sharpest in 1976 and 1977, the first two hypotheses: years after the decision to open the stacks. 1. Book circulation will increase if pa­ The decrease is less abrupt when we con­ trons are given greater access to library trol for building use loans, a type of transac­ stack areas. tion that logically occurs with greater fre­ 2. If patrons are given greater access to quency in a closed stack system, but a de­ areas, book availability rates cline of nearly 8,000 loans is found for 1976 will decline: even when building use data are excluded a. More books will be unavailable when re­ from the analysis. (The building loan cate­ quested. gory includes books returned across the cir­ b. Book losses through theft will increase. culation desk that have not been checked 3. If greater patron access to library stack out. It excludes books teshelved by patrons, . areas is permitted, use of the library build­ books left on tables or carrels in the stacks ing will increase. and reshelved directly by library personnel, The results of this analysis, it is assumed, and reserve books.) will have both theoretical and practical sig­ Clearly hypothesis 1 must be rejected. nificance. On a theoretical plane, the study This unanticipated decline in circulation should contribute to the development of would seem most readily explicable as a user behavior theory by explaining patron consequence of reduced book availability or behavior under changed conditions of col­ declining building use. Both explanations lection access. It should also benefit library can be tested with data derived from the decision makers by providing a reliable em­ study. pirical foundation from which decisions re- The book availability explanation can be

TABLE 1 No RESERVE CIRCULATION BEFORE AND AFTER DECISION TO OPEN LIBRARY STACKS

Circulation Total excluding Non reserve Building Building Year Circulation Loans Loans 1973 194,899 45,143 149,756 1974 192,520 46,068 146,452 1975 185,593 43,355 142,238 1976* 167,167 32,832 134,335 1977 149,647 16,176 133,471 1978 146,949 14,277 132,672 *First year of open stacks . 224 ( College and Research Libraries • May 1980 tested with search statistics and data ducted during the 1973--78 period are sum­ gathered from book delivery studies con­ marized in table 3. ducted at several intervals in the 1973-78 Since patrons who have searched unsuc­ period. Search statistics, shown in table 2, cessfully for a book in the stacks may also obviously do not explain the decline in cir­ present call slips at the circulation desk, one culation, since a decline in the number of would expect a deteriorating delivery rate searches was reported in the second and after the stacks were opened even if shelf third years of open stacks. The high number order remained constant. Instead, table 3 of searches conducted in 1976 appears to be reveals a mild improvement in the percent­ the culmination of a long-term trend that age of books delivered. This finding indi­ was reversed after the stacks were opened. cates even more conclusively than the search statistics in table 2 that the library's TABLE 2 book delivery performance was not impaired SEARCHES CONDUCTED, 1973-78 by the adoption of open access. Although the question of building use has Number of Year Searches not been raised previously in the literature, 1973 3201 it was assumed that direct access would. 1974 3775 make the library more attractive to users, 1975 4285 thereby contributing to increased use of the 1976* 4471 1977 4039 facility. 1978 3573 Table 4 confirms that building use did in­ *First year of open stacks. crease dramatically after the opening of the stacks, and particularly in the second year of Since hours of shelf remained rel­ the new system. This upsurge in building atively constant throughout the study pe­ use cannot be attributed to the increase in riod, the reduction in searches in 1977 and enrollment during the study period, given a 1978 can probably be attributed to students' simultaneous decrease in circulation figures . willingness to "settle" for an alternative This interpretation of the data is book on a particular topic when their first strengthened by the finding of significantly choice was not on the shelf. Thus, while the increased library use in 1976, the first year decrease in searches does not in itself of the open stack system and a year in demonstrate an improvement in book which enrollment actually declined. availability, it suggests that the collocation of related books on the shelf helped satisfy DISCUSSION AND CONCLUSION the needs of many browsing patrons. These findings conflict with the conven­ A more meaningful measure of book tional assumption that open stack systems availability in an open stack system can be contribute to increased circulation and a de­ obtained by comparing book delivery rates cline in book availability. Instead, a decline longitudinally. Since the main library did in circulation and a mild increase in book continue to provide book delivery service at availability were found in the three-year pe­ the circulation desk even after opening the riod following the initiation of direct access stacks, delivery success data are available at West Virginia University. A sharp in­ for several peak use periods. The results of crease in building use was also observed. three statistically comparable studies con- Taken together, these findings both chal-

TABLE 3 DELIVERY S UCCESS RATES FOR BOOKS REQUESTED AT CiRCULATION DESK

Date of Percent Delivered Percent Not Delivered Percent Not Study When Requested but Accounted For• Accounted For Nov. 1974 65.8 19.2 15.0 April1977 69.2 19.6 11 .2 April1978 70.9 16.9 12.2 *Includes books in circulation, reserve books, materials at bindery, branch library titles, patron errors in recording call numbers , etc. Open Stacks I 225

TABLE 4 Specifically, it appears that after the stacks BUILDING USE BEFORE AND AITER have been opened users increasingly make ADOPTION OF AN OPEN STACK SYSTEM relevance decisions at the shelf, rather than Persons borrowing a large number of books with the Using Year Library hope of finding something useful. If this conclusion is correct, improvements in book 1973 431,285 1974 425,708 availability are partially attributable to 1975 439,607 greater patron selectivity, a tendency that is 1976* 457,373 encouraged by direct access to the collec­ 1977 493,068 tion. If this is the case, and the increase in 1978 500,178 building use suggests that it is, declines in • First year of open stacks. circulation after the stacks have been lenge prevailing thought about open stacks opened are probably indicative of improved and suggest some important conclusions library service. about patron behavior in an open access sys­ The concurrent findings of increased tem. building use and declining circulation also The improvement in book delivery per­ suggest that user adaptation to an open formance indicates that unrestricted patron stack system is a generational phenomenon. access does not invariably produce increases Library use rises and circulation decreases, in misshelved books and theft if proper pre­ it appears, until a plateau is reached when cautions are being taken. Close supervision the majority of the student body has been of shelf readers and conscientious checks at exposed only to a direct-access system. the security desk clearly kept the adverse Initial dissatisfaction with the opening of effects of direct shelf access to a minimum. their exclusive domain may also be ex­ Increases in shelf reading and stack pressed initially by faculty members and patrolling are essential in libraries undergo­ graduate students, but this discontent can ing the transition to open stacks, since many likewise be expected to subside over time. users will misshelve books inadvertently and Several important issues in the shelf ac­ some do it deliberately, thereby denying cess debate are not addressed in this access to other potential users. Shifts of per­ analysis. The study does not, for example, sonnel from the circulation desk to stack provide any clear evidence whether open duty may also be required. Since a tight stack systems facilitate reader access to student assistant budget did not permit paticular books, save money, or promote these adjustments at WVU, meticulous the educational function of the library. supervision of shelf readers and conscien­ However, this study does indicate that, with tious security checks were used to minimize proper managerial anticipation of shifting the adverse effects of direct shelf access. use patterns, the stacks can be opened in a The findings of increased building use medium-size academic facility with sig­ and decreased circulation suggest that an nificant benefits for individual patrons and important change in user search behavior without adversely affecting overall library occurs once the stacks have been opened. performance.

REFERENCES

1. Mathilde V. Rovelstad, "Open Shelves/ 4. Guy R. Lyle, The Administration of the Col­ Closed Shelves in Research Libraries," Col­ lege Library (4th ed.; New York: Wilson, lege & Research Libraries 37:457~7 (Sept. 1974), p.ll7. 1976). 5. Francis Celoria, "The Archaeology of Seren­ 2. Harold D. Gordon, "Open Stacks: A Second dip," Library Journal 94:1846-48 (May 1, Look," Library Journal 94:1844-45 (May 1, 1969). 1969). 6. Hicks, "Open or Closed Stacks," p.311; Gor­ 3. Warren B. Hicks, "Open or Closed Stacks," don, "Open Stacks," p.1845. College & Research Libraries 15:309-12 (July 7. F. W. Ratcliffe, "Problems of Open Access in 1954). Large Academic Libraries," Libri 18:95-111 226 I -College and Research Libraries • May 1980

(1968); Hicks, " Open or Closed Stacks," 16. Herman H. Fussier and Julian L. Simon, p.311; Gordon, "Open Stacks," p.1845. Patterns in the Use of Books in Large Re­ 8. Philip M. Morse, "Search Theory and Brows­ search Libraries (Chicago: Univ. of Chicago ing," Library Quarterly 40:399 (Oct. 1970). Pr., 1969), p.ll2. This information is derived 9. Ratcliffe, "Problems of Open Access," p.l09. from an analysis of Fussier and Simon's table 10. Richard Joseph Hyman, Access to Library 36 in Robert J. Greene, "The Effectiveness Collections (Metuchen, N .J. : Scarecrow, of Browsing," College & Research Libraries 1972), p.372. 38:313-16 (July 1977). 11. Dorothy B. Cooper, " Open Stacks Open 17. Alice Bowen, "Non-Recorded Use of Books New Doors," Library journal 82:507--8 (Feb. and Browsing in the Stacks of a Research Li­ 15, 1957) . brary," M.A. thesis, University of Chicago, 12. Michael K. Buckland, Book Availability and 1961. the Library Us er (New York: Pergamon, 18. Johns Hopkins University, Research Library, 1975), p.94-97, 130. Progress Report on an Operations Research 13 . T. Saracevic, W. M. Shaw, Jr. , and P. B. and Systems Engineering Study of a Univer­ Kantor, "Causes and Dynamics of User Frus­ sity Library (Baltimore: Johns Hopkins tration in an Academic Library," College & Univ. , 1963). Research Libraries 38:7-18 (Jan. 1977). 19. Greene, "Effectiveness of Browsing," p.316. 14. Rita Hoyt Smith and Warner Granade, "User 20. Sam Clay, " Open Stack Study," Library and Library Failures in an Undergraduate journal 94:3378:--80 (Oct. 1, 1969). Library," College & Research Libraries 21. "Toronto Students Win Access to Library 39:467-73 (Nov. 1978). Stacks," Library journal 97:1655-56 (May 1, 15. Henry J. Dubester, "Stack Use of a Research 1972). Library," ALA Bulletin 55:891-93 (Nov. 1961).