College and Research Libraries DAVID J. NORDEN AND GAIL HERNDON LAWRENCE Public Terminal Use in an Online Catalog: Some Preliminary Results The authors have studied the transaction counts from two and one-half years' activity at the public use terminals of the Ohio State University Libraries' prototype online card catalog to determine what search options academic li- brary patrons use the most often and whether this pattern varies from that reported in major catalog use studies. The preliminary findings indicate sig- nificant differences in search strategy that may result from a unique user group that prefers to search the online catalog, more useful searches in the online system, or special search patterns imposed by the computer hardware itself Both the different searches used by patrons and why they choose them should be important factors in the design of future online catalogs. INTRODUCTION Academic and research libraries recognize that for a variety of reasons they must now consider new forms of patron access to biblio- graphic information. Many of these libraries have had extensive experience in automating such internal routines as acquisitions and cataloging; this experience, however, pro- vides little guidance in planning for alterna- tives to the manual catalog that library pa- trons can use. For some guidance library planners may turn to the major catalog use studies. 1 There also exist certain studies that analyze user acceptance and use patterns of David]. Norden is assistant head, Circulation Department, the Ohio State University Libraries, Columbus. Gail Herndon Lawrence is librarian, Reference Department, University of Arkansas at Little Rock Library. The authors would like to ac- knowledge the assistance of a number of persons and agencies in conducting this research: Reba Harvey and Bob Thorson , OSU Libraries' Circula- tion Department; Marty Goldsmith and Russ Hand , OSU Libraries' Mechanized Information Center; William Studer, director of libraries; Larry X. Besant , assistant director of libraries for public services; Susan Miller, coordinator of automated systems; the Libraries' Advisory Committee on Re- search; and the Instructional and Research Com- puter Center for a generous grant of computing time . 308 I commercial online databases. 2 Both of these may offer only minimal or tangential assis- tance, though, in predicting how patrons will respond to online or microform versions of the catalog. Researchers have investigated library pa- tron involvement with microform catalogs at the University of Toronto and the University of Oregon libraries. 3 Because few institutions currently have online public-use biblio- graphic systems, little research has been con- ducted on how patrons respond to and use computer terminals in searching biblio- graphic and holdings information. To help fill this gap , this paper analyzes the patron use of the prototype online catalog at the Ohio State University Libraries. The Ohio State University Libraries has operated its Library Control System (LCS) for nearly ten years. The LCS database contains online holdings and circulation records for all of the 3.5 million cataloged volumes in the libraries' collections (1.5 million titles). The system is used to provide certain types of ref- erence information, expedite order search- ing, handle general circulation routines, and aid in cataloging new material. In January 1975, a number of computer terminals were put in the main library lobby, so that patrons could use LCS directly without specialist or librarian intermediaries. From that time the number and level of use of these public ter- minals has grown steadily. Via these termi- nals patrons may search all cataloged holdings by author, title, author and title, call number, and browse the computerized shelflist. They may also search by Library of Congress sub- ject headings for items cataloged since August 1977. In order to understand how patrons have exploited the capabilities of public LCS ter- minals, the authors have collected data on the use of all public terminals from January 1977 to June 1979. At the beginning of the study seven public terminals had been installed in five library locations; by the end of the study twenty-one terminals were in nine locations .* By the end of 1980, there will be almost one hundred terminals available for use by pa- trons. The overall objective of this study was to determine how patrons utilized public termi- nals and if this use differed in any degree from known patterns of use of the card catalog. The authors hypothesized that the change of mode of access-from search of cards alphabetized in card catalog drawers to keying searches into a computer terminal-would in fact pro- duce a different search pattern. As a result the first specific task of the study was to iden- tify the relative level of use of the available searches. The second specific task was to de- termine whether the pattern of use changed over time . The third task was to determine whether the introduction of new searches af- fects significantly the proportion of the vari- ous searches. PROCEDURE The authors chose to examine patron use of the public terminals for the thirty-month pe- riod extending from January 1977 through June 1979 for three reasons. First, the most complete data on patron use of public termi- nals were available for this period. Second, although additional terminals were installed during the period under study, library pa- trons had had access to public terminals in *At the beginning of the study, public terminals were in the main, agriculture, education, engineer- ing, and undergraduate libraries; by the last month there were public terminals in the commerce li- brary, the history, English, and foreign language graduate reading rooms of the main library, and the west campus' learning resources center as well . Nine public terminals were located in a bank near the circulation desk. Public Terminal Use I 309 some locations for several years prior to the study, a sufficiently long period of exposure to allow the effects of the novelty of the system to have diminished. Third, certain significant changes and enhancements were made to LCS during this time period that permitted the study of the effects , if any, of the en- hancements on the use of the prototype on- line catalog. The patrons whose use of LCS is the sub- ject of this paper were self-selected because , outside of specific class assignments, no one is forced to use LCS in place of the card catalog. Those patrons who do use LCS , however , quickly learn how to do the basic searches. Indeed, many come to prefer LCS to the card catalog because of the information LCS pro- vides on the current circulation status and holdings. Informal surveys have also shown that there are patrons who do not like to use LCS and prefer to use the card catalog. 4 Since public terminals were available in the under- graduate libraries , the main library , some of the graduate reading rooms, and the larger department libraries, a large part of the pa- tron community at Ohio State University was exposed to public-use library terminals and had an opportunity to use them . All of the terminals studied were cathode- ray-tube (CRT) devices with the exception of one that was a thermal printer. At a number of points in the study, terminals of one man- ufacturer were replaced by those of others for technical and economic reasons. In any case , all terminals in place during the study had similar keyboards and operational features . Certain terminals, although designated as public terminals , were excluded from the study because their locations or because other factors caused them to receive substantially more use by staff than by patrons . The authors will not present an extensive description of the operation of LCS since op- erational descriptions of LCS have appeared elsewhere. 5 However, the system has evolved to the point where only one such de- scription captures the current state of the sys- tem. 6 Among the features of LCS is the abil- ity to monitor the amount and types of activ- ity on the system. Although these monitoring and report-writing capabilities have not been refined into a full-scale management informa- tion system, one of the reports provided the data for this study. 310 I College & Research Libraries • july 1981 This report, the monthly transaction re- port, summarizes the total number of each type of transaction performed at each termi- nal during each month. For locations with more than one terminal, whether staff or pub- lic, the system provides a summary of the total activity for that location as well. The transactions that are counted for each termi- nal include all the search commands as well as commands to "tum" the pages of the display. Thus, a transaction is any command that the user enters. Thus, not all commands or trans- actions represent a search. Because the authors were concerned with the patterns of searching LCS, they collected data only on the seven commands that dem- onstrated a choice of search (table 1). The au- thors chose to examine both the four searches that can be made in the manual catalog-title (TLS), author (AUS and AUT), and subject (SIS)-and three others that have no coun- terparts in the card catalog. One of the latter, the combined author-title search (ATS), has no direct equivalent in a manual catalog. It exemplifies the new forms of access that a computerized bibliographic system can provide and was, therefore, in- cluded in the study. The shelf position search (SPS) displays the fifteen items on either side of the call number typed into the terminal. It was included because it resembles an impor- tant manual file search , and it provided a crude form of subject access prior to the in- troduction of LCS subject heading searching (SIS). On LCS a patron may also search a specific call number (DSC) to determine the location and availability of that item. Because of the versatility and utility made possible by this linkage of bibliographic with circulation information, this search was included in the study. . In addition to the information on the types of searches entered by patrons, data on the number of invalid commands they entered were also gathered. Commands entered in- correctly or resulting from improper opera- tion of the terminal by the patron are rejected by LCS and counted as invalid. Excluded from the study were housekeeping commands that do not indicate a choice of search. Cer- tain other commands that are used in- frequently or can be employed profitably only by library staff also were excluded. The monthly transaction reports give the frequency counts for the various commands at a given terminal in a given month . The fre- TABLE 1 Type of Search Author-Title Exact Author Truncated Author Call Number Subject Search Computer Shelflist Title SEARCHES , COMMANDS , AND ASSOCIATED SEARCH KEYS Command ATS AUS AUT DSC SIS SPS TLS Search Key Formation First four letters of author's last name and first five letters of first significant word in the title.* Exact spelling and punctuation of author 's name as it appears in the author field of the LCS master circulation record. First six letters of author's last name and first three letters of author's first name in the case of a personal author . First six letters of first word in name and first three letters of second word in name in the case of a corporate author. Exact call number as it appears on catalog card , spine of book, or LCS record . Anf phrase whether authorized LC heading or not ; search wil display actual headings used at OS U that fall before and after the input search key . Any string of characters whether an actual call number or not; search will display the fifteen actual call numbers on either side of input string. First four letters of first significant word in the title and the first five letters of the second significant word in the title. *A stop list is a list of words that occur with high frequency and thus would form search keys of low precision and high recall ; all words on the stop list are not significant and are not used in formation of search keys . The LCS stop list for English language words is used for both the author and title fields of English language records ; the stop list for foreign language words operates only on the title field of foreign -language records. quencies for those commands identified as within the scope of this study (ATS, AUS , AUT, DSC, TLS, SIS, and SPS) and the number of invalid (INV) responses were transcribed from reports and keypunched. The authors then employed the Statistical Analysis System (SAS) to transform the raw frequencies of these commands into percent- ages of the total number of searches for each terminal for each month. The difference in total transactions among the terminals may vary by as much as a factor of seven due to differences in patron traffic. Conversion of counts to percentages of total transactions of- fered a method for comparing relative propor- tions of use from terminal to terminal and month to month. Because LCS provided two distinct types of author search but did not report the total number of author searches in the monthly transaction reports, SAS was used to total the author searches for each terminal in each month and compute the per- centage of author searches to total searches as well. Two other categories were computed as percentages of the total number of transac- tions (search plus housekeeping) at each pub- lic terminal. The first was the percentage that the total of the search commands (ATS, AUS , AUT, DSC, SIS, SPS , and TLS) represented of the total transactions at the terminals studied. Alsq, the percentage of invalid com- mands of the total number of transactions was computed to provide a measure of one type of patron failure in using both the search and housekeeping commands. During the thirty months of the study, new commands and searches were added to LCS. Period I ran from January 1977 through July 1977, the time just before the introduction of the AUT search . Period II bridged the time from August 1977, when the AUT command became available, through May 1978. Finally, Period III ran from June 1978, when the SIS search became operational, to June 1979. Then S'AS was used to compute the mean percentages for each available search for each period. Initially the means for each search for each month were computed to determine if there had been a change in search patterns over time. However, patterns varied from month to month according to the vagaries of the academic year, which masked any sig- nificant long-term changes from period tope- Public Terminal Use I 311 riod. Computing mean percentages for each search by period smoothed these monthly changes and made the differences from pe- riod to period more apparent. RESULTS Use data for the public terminals observed during the study indicate these terminals have been well received by OSU library pa- trons. It is worth pointing out again that pa- tron use of LCS is totally voluntary because the card catalog is still being maintained. Nonetheless, during the thirty months of the study, public terminals recorded 3,687,124 transactions, or almost exactly 20 percent of the 18,365,054 transactions registered by all terminal~ (public and others) on the whole sys- tem. On the average over the thirty months, there were about 128,000 transactions per- formed at all the public terminals per month , the actual figure increasing as more public terminals were added (table 2). In January 1977 the number of transactions performed at the public terminals was 63,569. In May 1979, the last month classes were in session in the study, the figure was 202,840. This repre- sents a 219 percent increase in use over the period of study. More than 1,845,000 searches were per- formed during the thirty months, an average of about 61,500 per month or about 738,000 per year at all the public terminals. These figures include just those transactions that represent a choice of search (ATS , AUS, AUT, DSC, SIS, SPS , and TLS) and do not include invalid responses or housekeeping commands . In the main library alone, public terminals recorded an average of 84,862 transactions and 42,093 searches per month during the study . On the average approximately 1,018,000 and 505,000 searches were done at the main-library terminals per year. Lipetz estimated manual catalog searches at Yale to be on the order of 320,000 per year in 1969. 7 Projecting the figures of R. R. Palmer, pa- trons of the general library at the University of Michigan consulted or searched its catalog approximately 310,000 times during the 1967--68 academic year. 8 It can be seen that the number of consultations or searches of the LCS prototype online catalog exceeds the use of manual catalogs in two similar research li- braries and that the level of use of the public TABLE 2 w ~ N UMBER OF TRAN SACTION S AND SEARCHE S, jAN UARY 1977 TO j UNE 1979 tv Tot!ll Transactions Total St>a rt: ht> s Numbt> r All All Publit: Main Lib. All All Public ~lain Lih. of Publit: n Date Terminals Terminals Terminal Termin als Terminals Tl'rminal Tt> rminal s £. January 1977 485, 774 63,596 38,942 191 ,652 33,496 21 ,607 7 ~ ()'Q February 1977 629,821 96 ,055 61 ,513 263 ,494 52 ,956 36,266 7 ~ March 1977 599 ,381 79,083 46, 135 246,847 46 ,059 28 ,513 8 G- April1977 572,796 84 ,774 48 ,933 254,600 48 ,370 29 ,536 8 :;:x, May 1977 636, 101 97 ,851 60,424 257,910 55 ,463 35 ,837 8 ~ c., ~ June 1977 476,055 51,021 29,817 193,397 28 ,948 18,517 8 ;:::, July 1977 432,523 49,980 29,626 182, 167 28 ,694 17,086 8 ., C') August 1977 457 ,038 45 ,804 26,122 201 ,203 25 ,434 14,941 11 ::3"" September 1977 438,713 68 ,352 53,300 192,291 38 ,421 30,218 14 t""" 6=-October 1977 688 ,214 148,387 112,893 298,773 79,250 61 ,413 14 ., November 1977 711 ,364 155,645 113,884 296,676 80,855 59 ,803 15 ;:::, ., December 1977 438 ,235 59,033 42 ,987 183,810 32,330 23,412 15 ~- January 1978 557,733 111 , 197 82 ,654 241 ,904 56,089 42, 718 15 c., . February 1978 703,550 154,072 115,702 310,961 77 ,863 58 , 178 15 ? March 1978 696,262 132,611 103, 134 297 ,896 69,073 53,496 15 ~ April1978 692,616 146,872 108,330 304 ,990 73,.594 54 , 168 15 ...... May 1978 752,257 164,340 122,678 318 , 146 80,012 59 ,944 17 "' Oo June 1978 500, 766 82, 122 63 ,230 201 ,863 40,467 30 ,688 17 ...... July 1978 528 ,651 112,227 86, 720 225 ,077 55 ,289 42 ,770 17 August 1978 531 ,236 102,749 80,241 226,986 50,867 39 ,428 18 September 1978 418 ,639 112,227 79,583 175,473 54 ,420 37 ,272 19 October 1978 768 ,053 211 ,585 150,960 316 ,477 98 ,883 69,345 20 November 1978 778 ,401 217,762 151 ,502 310 ,080 101 ,350 69 ,629 19 December 1978 479, 793 84,772 51 ,712 181 ,343 42,271 24,644 19 January 1979 731,735 183,606 119, 193 292 ,421 85 , 138 53 ,806 19 February 1979 763 ,878 202,053 137,962 300,278 91 ,974 61 ,959 21 March 1979 750,019 167,643 107,943 294 ,890 76,010 47,529 21 April1979 737 ,314 188,205 114,672 292 ,777 84 ,499 50,956 21 May 1979 832 ,533 202,840 132,581 315 ,453 82 ,849 55 ,472 21 June 1979 571 ,603 110,006 72 ,503 220,479 51 ,690 33,646 21 Totals 18,361 ,054 3,686,470 2,545 ,876 7,590,314 1,822,614 1,262,797 Public Termi11al Use I 313 TABLE 3 MEAl'\ PERCEI'\T OF EACH SEARCH BY PERIOD AND MEA!Ii PERCENT OF SEARCHES Al'\D IN\'ALID Co~~IANDS OF TOTAL TRANSACTIONS Search Period I ATS 27.8 AUS 19.5 AUT NIA Total Author 19.5 (AUS+AUT) DSC 18.7 SIS NIA SPS 2.8 TLS 31.2 Mean Percent 55.3 of Searches of Total Transactions Mean Percent 13. 1 of Invalid Transactions of Total Transactions terminals was limited during the period of study by the availability of terminals.* When one examines the percentage of use of each search and how it changed over time (table 3), one finds the most striking change in the use of the AUS search. A immber of ex- planations might be offered for its steep drop in use from 20 percent in Period I to 5 per- cent in Period Ill. To use this search com- mand the patron must enter the author of an item exactly as it appears in the author field of the LCS master circulation record. A trans- position of characters or slight misspelling in the search key means that the desired results will not be obtained. The nine-character search key of the AUT frees the user from having to spell the au- thor's name exactly. Research findings show that users often approach the catalog with in- complete or incorrect information. 9 Fur- thermore, the AUS so reduced the response time of the computer that only one A US search was permitted at any time anywhere in the system. Most likely as a consequence of its ease of use, the AUT author search appears to have replaced the A US author search. Dur- ing the last three months of the study the *The number of consultations does not include the informational phone calls to the libraries' tele- phone center, staffed by twenty-eight half-time po- sitions for 106 hours per week, answering more than 200,000 information and circulation calls per year. Period II Period Ill 23.2 20.6 8.9 5. 1 11.9 14.1 20.8 19.2 18.8 18.4 NIA 4.9 ' 2.7 2.5 34.7 34.4 52.3 47.3 12.6 12.2 AUS stabilized at 2.5 mean percent of tl total searches. The sole remaining advanta~ of the AUS rested in its power to discrimina1 among corporate and certain personal autho1 whose names formed AUT search keys th< produced great numbers of matches. The mean percent of total author searchin (AUT plus AUS) for each of the three period did not vary by more than 1.5 percent. Thi low level of variation from period to periot would indicate that while patrons did no alter their overall amount of author searching they did demonstrate a decided preferenct for an easier, more forgiving author searcl when such became available. Patrons used call-number searches (DSC more frequently than had been expected. In all periods of study, the DSC represented al- most one out of five searches at the public terminals. This level of use might indicate that patrons find information on the location and circulation status of library material as important as information as to whether the libraries own the item or not. The authors suspect that it also indicates that a substantial number of patrons are sophisticated enough to combine searching of the card catalog or some other source providing call-number in- formation with searching of LCS. The mini- mal variation from period to period would suggest that call-number searching repre- sents a basic requirement that patrons would have for an online bibliographic system. 314 I College & Research Libraries • july 1981 The author-title search (ATS) dropped 7.2 percent in its share of total search choices from Period I to Period III. The greatest drop occurred between Period I and Period II when the AUT was introduced. Offering a more convenient author search may have caused patrons to alter their search patterns. Also, during Period II , serials-holdings in- formation became available on LCS ; how- ever, it is not clear to the authors what effect this information might have had on the use of the ATS search. The subject search (SIS), although the newest search, apparently has met an impor- tant need of LCS users despite the present lack of online cross-references and authority control. These capabilities are presently being programmed and should be available online in 1982. As a result of these limita- tions, library administrators urged library lo- cations not to promote actively the SIS during the time covered by the study. However, use of this search increased without any formal promotion or instruction in its use. The first month that the SIS search was available it represented only 0.45 percent of the searches system-wide; in June 1979 it had climbed to 9.3 mean percent of searches system-wide. For the whole of Period III the SIS search achieved a mean percent of 4 . 9 percent of total searches . The health sciences library staff decided that, despite its limitations , the SIS offered enough utility to justify training patrons in its use. In addition, one public terminal in the main library had instructional material on the SIS posted near it as an experiment. Sub- sequent to instructing patrons in the use of the SIS, one of the health sciences library terminals showed a mean of 12.8 percent of the choices during Period III. At the other public terminal in that library during the same time, subject searching on LCS was a mean of 9. 5 percent of the searches. Subject searching at the main library terminal was 8. 6 percent of the total searches at that terminal after posting instructions. Prior to that, SIS searching at that terminal accounted for only about 1 percent of the total searches. Online shelflist searching (SPS) was added to the system in the mid-1970s. The SPS search allows patrons to enter a call number to retrieve abbreviated records for the fifteen items preceding and the fifteen following the call number that was entered. Because the patron does not need to enter the call number of an actual item in the collection, he can browse the whole collection by using the SPS as a crude subject search . The libraries have never actively promoted the possibilities of this search, which may account for the low level of use of this search at public terminals. During all periods of the study the mean per- cent of the SPS remained fairly constant and quite low. In contrast to the findings of some manual catalog studies, the title search (TLS) demon- strated the highest overall percentage of searches during all three periods of the study. 10 Since serials-holdings information was available on LCS after Period I, the au- thors anticipated a steady increase in title searching as patrons came to rely on LCS for this information. Title searching did increase modestly from Period I to Period II, but it then dropped slightly from Period .II to Pe- riod III. The availability of the SIS in Period III might have diverted some searches from the TLS, since some patrons had been ob- served using subject headings as search keys for the title search. Searches form just a part of the total trans- actions performed at public terminals. The authors examined whether the overall pro- portion of searches to total transactions changed during the course of the study. Dur- ing Period I searches were a mean 55.3 per- cent of the total transactions, and in Periods II and III, respectively, 52.3 mean percent and 47.3 mean percent. With the exception of the call-number search. (DSC) and the subject search (SIS), all the searches required one additional transaction in order t~ display a record . The subject search required two additional transactions in order to display a record, and the call number search required no additional transactions. The authors have noted that when subject searching became available and was promoted, it achieved a signific~nt share of the total searches . Be- cause subject search requires two additional transactions to display a record, more subject searches will increase the total number of transactions at a faster rate. This may account for the lower overall percentage of searches in Period III. Also, if the availability of the AUT author search caused patrons to be more suc- cessful in their author searching, then they might have performed more transactions to display additional records from their suc- cessful author searches. The authors were also interested in how many invalid commands patrons were enter- ing. A relatively high level of invalid com- mands might indicate, among other things, that the average person might have difficulty in using terminals to access bibliographic in- formation or that the libraries' training mate- rials were not effective. While efforts to train users increased during the study, more ter- minals were installed during that time, which meant more untrained users, presumably more error-prone , would be exposed to the system. A tension between these two factors might have caused the percentage of invalid commands to remain fairly constant, with the mean percent of invalid transactions decreas- ing from Period I to Period III by only . 9 percent. CONCLUSIONS Three findings of this study may have im- plications for the design of future online bib- liographic systems . First, a significant number of academic library patrons will ac- cept and use an online alternative to the card catalog. The number of transactions per- formed at the public LCS terminals rose con- sistently over the two and one-half years of the study. Second, search patterns were fairly consistent despite increased use of the system and an increase in the number of search op- tions . Even the number of invalid commands remained constant. Third, the amount of on- line title searching differs from that reported in a number of studies of the card catalog. About one out of every three searches on LCS was a title search, whereas about one LCS search in five was an author search. The au- thor search , which the findings of major catalog use studies have shown to be the most favored search, 11 was the third most fre- quently chosen search by LCS users. Despite the fact that during the thirty months of the study LCS offered first a search that required the author's exact name and then a more eas- ily used search, the level of author searching remained fairly constant. The implications of these findings are more fully appreciated when viewed in the light of another finding of the catalog-use studies. Interviews with patrons using the card catalog Public Terminal Use I 315 showed that 60 percent of them came to the catalog with better title information than au- thor information. In spite of having better title information, these patrons searched more by author in the card catalog. 12 Since it appears that title searching in the card catalog is more difficult, a major advantage of online bibliographic systems may be that they will make title searching viable and in the process will more closely align modes of access to bib- liographic information with the ways patrons actually search for it. This high level of title searching was unan- ticipated; in designing LCS the libraries of- fered a more precise option for known-item searching , the author-title search (ATS), which it was presumed patrons would prefer. The results of this study reveal, however, that this search currently accounts for only about 20 percent of the searches at the public LCS terminals and that its use has declined stead- ily throughout the three phases of the study. This search seemed to be the one most af- fected by the introduction of new, less precise commands, which do not require the user to bring as much information to his search. Perhaps we are seeing here the same phe- nomenon reported in studies of online sys- tems by Briggs and Kobelski. Writing about users of online databases, Briggs reports: There are two indications that users are more se- verely discouraged by too few references than they are by too many. All of the users reporting too many answers still described the search as of some use. But 59 percent of the users reporting too few or no answers found their results oflittle or no use. Nearly all users reporting too many answers indi- cated that revisions were in order, but about one- half of the users with too few or no hits felt they did not have time to determine needed revisions or it was not worth the effort, or it was too late to be of help to them. 13 Kobelski encountered the same responses from users and cites three possible expla- nations for this reaction: (1) a larger number of citations approximates a printed index that a searcher can browse to feel reasonably confident he has retrieved all relevant cita- tions, (2) the high cost of online computer time in comparison to the very low cost of offiine prints, and (3) student willingness to accept and use citations on a related subject along with those of their original topic. 14 Future research in this area will have to 316 I College & Research Libraries • july 1981 address several major questions. First, are the relatively stable search patterns found in this study independent of the design of this particular system and its community of users, or are they unique to this particular system and its users? Do the users of LCS form a special subset of library patrons employing unique patterns of searching? In other words, does a certain kind of patron with a certain kind of need for information choose to use the card catalog while another with different needs chooses to use the online system? Fi- nally, do the stable search patterns result from unique design of the LCS hardware and software, or does the physical difference be- tween the card catalog drawer and the com- puter terminal produce different patterns of searching? REFERENCES 1. F. W. Lancaster, "Studies of Catalog Use," in his Measurement and Evaluation of Library Services (Washington, D.C.: · Information Re- sources Press, 1977), p.19-72. 2 . R. Bruce Briggs, "The User Interface for Bib- liographic Search Services," in The Use of Computers in Literature Searching and Re- lated Reference Activities in Libraries , Papers presented at the 1975 Clinic on Library Appli- cations of Data Processing, April 27-30, 1975 (Urbana, Ill.: University of Illinois Graduate School of Library Science , 1976) , p.56-77 ; James A. Cogswell, "Online Search Services: Implications for Libraries and Library Users," College & Research Libraries 39:275--80 (July 1978); Pamela Kobelski and Jean Trumbore, "Student Use of Online Bibliographic Ser- vices,' ' journal of Academic Librarianship 4:14-18 (March 1978). 3. Valentina de Bruin, "Sometimes Dirty Things Are Seen on the Screen; A Mini-evaluation of the COM Microcatalog at the University ofTo- ronto Library," journal of Academic Librar- ianship 3:256-66 (November 1977); James R. Dwyer, "Public Response to an Academic Li- brary Microcatalog," journal of Academic Li- brarianship 5:132-41 Guly 1979). 4. Phyllis Davis, Saragail Runyon Lynch, and Victoria Welborn Spemoga, "Card Catalog and LCS Users: A Pilot Study" (unpublished paper, 1979). 5. Hugh G. Atkinson, "Circulation System of the Ohio State University," in On-line Library and Network Systems (Frankfurt am Main: Klos- termann, 1977), p.94-103; A. Robert Thorson , "Tomorrow's Library Today, " Theory into Practice 12:191-95 (June 1973). 6. Susan L. Miller, "The Evolution of an Online Catalog," in New Horizons for Academic Li- braries, Papers presented at the First National Conference of the Association of College and Research Libraries, Boston, Massachusetts, November 8-11, 1978 (New York: K. G. Saur, 1979), p.193-204. 7. Ben-Ami Lipetz, User Requirements in Iden- tifying Desired Works in a Large Library (New Haven, Conn.: Yale University Library, 1970), p.33. 8. James Krikelas, "Catalog Use Studies and Their Implications," in Melvin J. Voigt, ed., Advances in Librarianship, V.3 (New York: Seminar Press, 1972), p .211. 9. Sidney L. Jackson , Catalog Use Study (Chicago: American Library Assn. , 1958), p.25. 10. R. Tagliacozzo and M. Kochen , "Information- seeking Behavior of Catalog Users," Informa- tion Storage and Retrieval: Theory and Prac- tice 6:363-81 (December 1970) . 11 . The important findings from the catalog use studies are summarized on p.69-72 of Lancas- ter's Measurement and Evaluation of Library Services. 12. Krikelas, "Catalog Use Studies," p.213. 13. Briggs , "User Interface," p. 70. 14. Kobelski, "Student Use," p.17.