College and Research Libraries STEVEN LEACH The Growth Rates of Major Ac~demic Libraries: Rider and Purdue Reviewed In The Scholar and the Future of the Research Library Fremont Rider hypothesized that since in the past academic libraries have doubled in size every sixteen years, this rate of growth can be expect- ed to continue through the immediate future; he invoked this hy- pothesis to demonstrate the scope of the problem that such growth will engender. The 1965 Purdue study of library growth in a sense verified Rider's hypathesis and its use as a predictive tool. This study, which measures tlie growth rates of the twenty-five largest ARL li- braries from 1962-63 through 1973- 7 4, discredits the validity of Rider's hypothesis, notes the limitations of the Purdue study growth- rate figures, and tentatively identifies the collection size level past which growth-rate deceleration begins. THE PURPOSE OF TinS PAPER IS THREE- FOLD: ( 1) to review critically Fremont Rider's hypothesis about the exponen- tial growth of research library collec- tions as well as the growth-rate projec- tions of the Purdue study; ( 2) to re- .., port and interpret the results of a study which measured the growth rates of the twenty-five largest Association of Re- search Libraries ( ARL) libraries from 1963 through 197 4; and ( 3) to deter- mine if the resultant growth-rate figures ,. corrobor~te the Rider hypothesis.1 RIDER's HYPOTHESIS ABOUT FUTURE LmRARY GROWTH Fremont Rider began the text of his ~ book The Scholar and the Future of the Research Library by announcing that ., "although it had been known for a l9ng Steven Leach is, information services li- brarian; , TechnQlogical Institute Library, Northwestern .University, Evanston, Illioois . . time that American research libraries were growing at a very rapid rate, it was not realized until a few years ago that they were, on the average, actually doubling in size every sixteen years."2 This opening statement summarized Ri- der's interpretation of the results of his study, which had measured the past rates of growth experienced by the cate- gories of libraries shown in Table 1. The significance of exponential li- TABLE 1 RIDER's STUDY: LIBRARY CATEGoRIEs AND GROWTH RATES Number of Libraries Average Doubling Period. per Category 1831-1938 10 Men's Colleges 22 years 5 Women's Colleges 0 10 Pre-1849 Universities 16 years 10 Post-1849 Universities 9.5 yearst • Unspecifled, but ••a more rapid growth than the men's colleges" (p.5). t 1876-1938 . Source: Rider, The Scholar and the Future of the Re- aearch Library, p.4-8. I 531 532 I College & Research Libraries • November 1976 brary growth to the future efficacy of the university library was not lost upon Fremont Rider: In fact, this may be asserted as almost axiomatic: unless a college or univer- sity is willing to be stagnant, unless it is willing not to maintain its place in the steady flow of educational devel- opment, it has to double its library in size every sixteen years, or there- abouts.a Here, Rider appropriated the truism that there exists a direct correlation be- tween continuous library growth and the educational effectiveness of any uni- versity, combined it with his conclusion about the past rate of library growth, and thereby constructed a seemingly in- contestable argument from which he ex- trapolated his hypothesis about the rate of future library growth. Rider then proceeded to demonstrate that the hypothesis reliably could be used to project library growth. However, his demonstration rested upon a rather tenuous argument consisting of ( 1 ) a reiteration of his preliminary argu- ment; ( 2) a rhetorical sleight of hand; and ( 3) the following rationale: If the observed rate of library growth has pre- vailed for the past 200 years as is the case with Harvard and Yale, then in all probability that same rate of growth will prevail for the next 100 years.4 Moreover, Rider neglected to test the truth of his hypothesis by expanding his original study to include more libraries. In his review of Rider's book, Keyes D. Metcalf called attention to the omis- sion of evidence which tended to dis- credit Rider's hypothesis: If Mr. Rider had checked the reports of the New York Public Library and of the Harvard and Yale college li- braries . . . thirty-two years ago . . ·. he would have found that they have failed to quadruple since that ·time .and that their rate of growth on the per- · centage basis has been steadily" de- creasing. If he had checked the "Gerould statistics" for libraries that had passed the five hundred thousand- volume mark sixteen years ago, he might have modified his figures, be- cause he would have seen that, as li- braries of any type grow larger, they tend to grow less rapidly. 5 j \ Needless to say, the rate of library growth as specified by Rider's hypothesis ~ cannot be sustained indefinitely by any ~ university library. Rider too conceded that "doubling every sixteen years" can- not "continue indefinitely" but qualified his concession by maintaining that in "so far as the foreseeable future is con- ~ cerned" a deceleration in the rate of li- brary growth was only "remotely prob- • able."6 Even with his qualified aclmowl- edgment, Rider did not impose any practical restrictions upon his use of the hypothesis to forecast levels of library growth. r In his review of the factors that in ,. the future either would retard or stim- ulate library growth, e.g., new subject disciplines, the general level of educa- tion, etc., Rider did not discern that the 1 diseconomies of large scale affect large university libraries as well as large cor- porations.7 To elaborate, in time the ~ continuous expansion of a library's col- "" lection at a N-percent rate of growth will give rise to the diseconomies of large scale, whereupon that N-percent rate of growth will become increasing- ly difficult to maintain because of the · ~ size of the collection. Certainly, Met- calf recognized that eventuality when he • took exception to Rider's hypothesis and noted that "as libraries of any type grow larger, they tend to grow less rap- idly."8 ~ Moreover, in his analysis of the growth statistics, Rider did not assess the ~ context in which past library growth had occurred. An extended evaluation of those factors, e.g., economic, mathe- matic, social, etc., which had · affected • past library growth admittedly would .-4 have gone beyond the purview of his study. Nevertheless, the author did not even indicate recognizance of any such relationships. Instead, Rider represented the growth of university libraries as being an isolated phenomenon. Given the incompleteness of his analysis, it is understandable why Rider failed to im- pose any practical restrictions upon his use of the hypothesis to predict library growth. Furthermore, it must be allowed that Rider was not specifically , CQn- cerned about demonstrating the I,, 'ti- macy of his hypothesis, for in· his I~ to MetcaH's review Rider insisted: This seems to me the vitally significa - t point-that, even though their [i.e., libraries'] rate of increase should shrink to doubling-every-twenty-years-or e"- ery thirty, or every forty years-the growth figures would still be so "astro- nomical" that the practical problems posed by them would be staggering ones. 9 Rider was interested in demonstrating the validity of the hypothesis only inso- far as the hypothesis was useful in dem- onstrating to the profession the reality of the library growth problem. One purpose of the book was to alert the li- brary profession to the consequences of continuous exponential library growth. To that end, Rider used the hypothesis to dramatize the seriousness and the scope of that problem and to under- score his warning that the "problem of research library growth is, whether they realize it or not, by all odds the most serious one that librarians and educators face."10 Another purpose of the book was to publicize Rider's proposed solution to the growth problem, i.e., the extensive use of microtext. Although in most of the contemporaneous reviews of the book, attention primarily was focused on Rider's solution to the growth prob- lem, ironically it was Rider's use of the hypothesis to predict library growth that Growth Rates I 533 since has become part of the profession- al litany to be cited whenever the growth of university libraries is dis- cussed. This has occurred even though two papers were published in the early 1960s which tentatively concluded that the growth of the twenty-five largest university libraries from 1946 through 1960 had not substantiated ·Rider's hy- pothesis.ll THE PuRDUE STUDY For the previously cited 1965 Purdue study, all available statistics from 1950- 51 through 1963-64 for fifty-eight ARL libraries were examined. The derived growth-rate figures in tum were used to project library growth through 1980. Initially, a composite library was cre- ated out of these statistics; however, be- cause of the distortion caused by the presence of a few extremely large li- braries, other composite libraries were created by assigning the fifty-eight li- braries in the study to four different categories of collection size. The size categories and the corresponding growth figures are presented in Table 2.12 TABLE 2 PmmUE STUDY: LmRARY CATEGORIES AND GROWTH RAT~ Category Name Composite Library Large Library Medium-Large Library Medium-Small Library Small Library Approximate Doubling Period 17 Years 18-20 Years 12-14 Years 12-14 Years 14-16 Years Source: Dunn, Past and Likely Future, 1965, p.21, 37, 40, 43, 46. These figures seem to substantiate the accuracy of Rider's hypothesis to pre- dict library growth. Similarly, in his re- view of the study, Guy Garrison ob- served that "the chief impression the reader gains from an examination of the report is that Fremont Rider was, in general, correct in his ominous pre- dictions."13 Robert L. Talmadge re- marked in his review that "this absorb- 534 I College & Research Libraries • November 1976 ing work might well have been dedicat- ed to the late Fremont Rider," for, ac- cording to Talmadge, the growth rate computed for the fourteen largest li- braries did not constitute "essential un- dermining of Rider's thesis."14 Yet, in contrast to the derivation and utilization of the Rider hypothesis, the Purdue composite growth projections were developed from a more elaborate study of past library growth, were the product of a more sophisticated meth- odology, and were to be used in con- junction with such growth-related pro- jections as volumes added, staff size, and book expenditures. Furthermore, as ccan outgrowth of past studies" which had "proved helpful in developing plans for future growth," the Purdue growth projections specifically were intended to be used in that capacity, whereas no such application of Rider's hypothesis to library planning had been proposed by its author. 15 The authors of the Purdue study con- curred with Rider that exponential li- brary growth cannot continue indefinite- ly: Even though collections, acquisitions, expenditures, etc. which are ten, twen- ty, or fifty times the present levels can- not be seriously proposed for the fore- seeable future and although some of the fitted curves, when extended some years beyond 1980, indicate incon- ceivably high levels, there is little ba- sis for expecting an early deceleration in library growth. In short, the records of growth since 1951, including the most recent years, and the unfaltering growth of even the largest libraries, in- dicate that this growth will not soon decelerate .IS The authors speculated that if decel- eration in the rate of library growth does occur, c'it can hardly be expected before 1980 and perhaps not for many years thereafter.''17 However, in a re- cent issue of the Purdue study, which examined library growth through 1969-- 70 with a follow-up stage beginning in that year, the authors qualified their earlier prediction to read as follows: Collections, acquisitions, expenditures, etc., which are ten, twenty, or fifty times the present levels cannot be seriously envisioned for the foreseeable future and some of the fitted curves, when extended some years beyond 1980, indicate inconceivably high lev- els, thus there is some basis for expect- ing some deceleration of library t wth. Although there has been con- :uous growth since 1951, including e pf the most recent years, library wt)l-currently is faltering. IS e above conclusion is supported by the results of the study that is reported in the following section of this paper. REPORT OF THE STUDY The twenty-five largest ARL libraries as ranked on the basis of collection size by the 1973-7 4 ARL statistics are listed in Table 3. Within this group of librar- ies nine private and sixteen public in- stitutions are represented. The data for the study were compiled from ARL sta- tistics for 1962-63 through 1973-74. After being corrected for transcription and keypunching errors, the data were used in a PL- 1 computer program writ- ten expressly for the study. No attempt was made to resolve inconsistencies in the data which were caused either by re- porting discrepancies or errors as it was presumed that such inconsistencies would affect the results of the study only minimally. The time period examined was select- ed on the basis of the following ration- ale: ( 1) The twelve years from 1962-63 through 1973-7 4 were perceived to have encompassed a unique phase in the post-World War II development of uni- versity libraries. The year 1962--63 was selected as the beginning point for the study because it marked the first full year of recovery from the minor eco- + j . .... ~ .A Growth Rates I 535 TABLE 3 !tANK BY CoLLECTION SIZE ( Mn.uoNs oF VoLuMES), 1962-63 AND 1973--74 Collection Size Library 1962-63 1. Harvard 7.074 2. Yale 4.693 3. Illinois 3.635 4. Michigan 3.134 5. Columbia 3.088 6. Berkeley 2.829 7. Cornell 2.413 8. Stanford 2.379 9. Chicago 2.271 10. Minnesota 2.221 11. Toronto 1.944 12. UCLA 1.867 13. Pennsylvania 1.836 14. Princeton 1.834 15. Indiana 1.787 16. Duke 1.593 17. Ohio State 1.591 18. Northwestern 1.587 19. Texas 1.578 20. Wisconsin 1.446 21. North Carolina 1.350 23. Washington, Seattle 1.267 24. Johns Hopkins 1.263 25. New York University 1.223 37. Michigan State .960 nomic recession of 196~61. There is little need to inventory in detail those events and economic conditions which affected library growth during these twelve years, e.g., the infusion and sub- sequent withdrawal of federal funds for library development; the accelerated expansion and subsequent retrenchment of academic programs; numerous li- brary construction projects; and, more recently, an economic recession char- acterized by double-digit inflation. Brev- ity necessitates that a fuller justification of this perception be omitted, and for that reason the reader is referred to the McAnally and Downs article "The Changing Role of Directors of U niver- sity Libraries.''19 Their article is cited because a similar perception seems to underlie the authors' review of events in the 1960s, germane to the develop- ment of university libraries, and their examination of those events in relation Collection Size Library 1973-74 1. Harvard 9.028 2. Yale 6.351 3. Illinois 5.328 4. Columbia 4.573 5. Michigan 4.549 6. Berkeley 4.477 7. Toronto 4.212 8. Cornell 4.158 9. Stanford 3.982 10. Indiana 3.665 11. Texas 3.519 12. Chicago 3.485 13. Minnesota 3.479 14. UCLA 3.395 15. Ohio State 2.912 16. Wisconsin 2.785 17. Princeton 2.615 18. Pennsylvania 2.557 19. Duke 2.530 20. New York University 2.390 21. Northwestern 2.349 22. Washington, Seattle 2.102 23. Michigan State 2.083 24. North Carolina 2.044 25. Johns Hopkins 2.006 to the problems confronting university libraries in the 1970s. ( 2) In the Economics of Academic Libraries, Baumol and Marcus noted that "the 1950s and sixties were a period during which colleges and universities [and by extension, their libraries] were comparatively well supplied with finan- cial resources" but ccthe recent financial crisis of higher education is not re- flected in the trends reported" in their analysis of recent library growth, based upon data from the Purdue study.2o However, the effects of this recent finan- cial crisis should be reflected in the trends reported in an analysis of library growth from 1962-63 through 1973--7 4. As was noted in the introduction, one purpose of this study was to determine if the rates of growth sustained by the twenty-five largest ARL libraries from 1962-63 through 1973--74 corroborate the rate of library growth specified by 536 I College & Research Libraries • November 1976 the Rider hypothesis. To this end, tables which report the growth-related mea- surements obtained for these libraries were prepared. Before proceeding, two of the tables require some explanation. Table 4: For each library the years re- quired for the 1962-63 collection to double in size were projected on the ba- sis of the annual mean rate of growth for 1962-63 through 1973-74. For cer- tain libraries the wide range in yearly percentage changes in collection size can be attributed either to adjustments post- ed because of previous reporting errors, to the initial inclusion or exclusion of the holdings of a satellite library, or simply to a significant increase or de- crease in the volume of acquisitions for one year. Table 5: For each size level a volume range was defined, e.g., for the 2 mil- lion-volume level the range was 1.9 to 2.39 million volumes; for the 2.5 mil- lion-volume level, 2.4 to 2.89 million volumes; and so forth. The "pre-" mean rate of growth represents the annual mean rate of growth posted from 1962- 63 through the year when the collection reached the specified size level, provided that there were at least two years of growth to be measured. Similarly, the "post-" mean represents the mean rate of growth registered from the year when the collection attained the speci- fied size level through 1973-7 4, again provided that there were at least two years of growth to be measured. To il- lustrate, Texas in 1966-67 reported hold- ings of 1.945 million volumes; there- fore thP. "pre-" mean is the annual mean rate of growth from 1962-63 through 1966-67 (four years) while the "post-" mean is the annual mean rate of growth from 1966-67 through 1973-7 4 (seven years ) . In order for a library to double in TABLE 4 RANK BY ANNUAL MEAN RATE OF GROWTH AND DouBLING PERIOD wrm RANGE oF YEARLY PERCENTAGE CHANGES IN COLLECTION SIZE (1962-63 TO 1973-74) Library Mean Rate of Growth Doubling Period 1. Texas 2. Toronto 3. Michigan St. 4. Indiana 5. N.Y.U. 6. Wisconsin 7. Ohio St. 8. UCLA 9. Cornell 10. Stanford 11. Washington 12. Johns Hopkins 13. Duke 14. Berkeley 15. Minnesota 16. North Carolina 17. Chicago 18. Northwestern 19. Columbia 20. Illinois 21. Princeton 22. Michigan 23. Pennsylvania 24. Yale 25. Harvard 7.67% 7.39 7.31 7.01 6.51 6.34 5.65 5.60 5.08 4.83 4.71 4.36 4.30 4.26 4.17 4.17 3.97 3.73 3.67 3.55 3.49 3.46 3.07 2.79 2.25 Average mean rate of growth for all libraries: 4.77% Median mean rate of growth: 4.30% 10 yrs. 10 10 11 12 12 13 13 15 15 16 17 17 17 17 18 18 20 20 20 21 21 23 26 32 Range (±) 4.34/21.80% -4.70/13.67 2.42/9.52 -7.48/22.56 -6.99/22.45 -8.05/22.16 4.35/7.75 3.39/9.49 2.86/6.79 1.57/11.92 3.27/5.87 -2.60/9.25 3.52/4.84 2.14/5.78 3.18/5.69 -9.64/24.15 2.75/5.19 -6.18/14.74 .96/11.79 -1.00/5.61 -9.25/16.84 .24/5.77 -1.07/4.32 .21/3.59 1.65/3.04 TABLE 5 MEAN RATES OF CoLLECriON GROWTH (1962-63 TO 1973-74) CoNTRASTED TO MEAN RATES oF GROWTH AT VARious LEVELS OF CoLLECTION SIZE 2.0 Million Vol. 2.5 Million Vol. 3.0 Million Vol. 3.5 Million Vol. Mean Growth Level Level Level Level Rate Pre-2.0 Post-2.0 Pre-2.5 Post-2.5 Pre-3.0 Post-3.0 Pre-3.5 Post-3.5 Library ( 12 yrs.) Mean Mean Mean Mean Mean Mean Mean Mean 1. Texas 7.67% 5.37% 8.98% 5.53% 13.37% 2. Toronto 7.39 7.71 7.21 8.41% 6.55% 9.52% 3.67% 3. Michigan St. 0 7.31 4. Indiana 7.01 3.12 8.47 5.87 8.37 7.34 6.14 5. N.Y.U. 6.51 10.88 2.87 6. Wisconsin 6.34 7.27 5.57 6.91 4.83 7. Ohio St. 5.65 5.73 5.61 6.02 4.66 8. UCLA 5.60 7.51 5.41 7.26 4.66 6.59 3.87 9. Cornell 5.08 6.15 4.45 6.11 3.84 10. Stanford 4.83 5.52 4.43 5.49 3.67 11. Washington 4.71 4.84 4.14 12. Johns Hopkins 4.36 6.68 2.93 13. Duke 4.30 4.07 4.49 14. Berkeley 4.26 ... 4.23 4.29 15. Minnesota 4.17 3.75 4.33 4.12 4.26 16. North Carolina 0 4.17 17. Chicago 3.97 2.92 4.20 3.94 4.02 18. Northwestern 3.73 3.89 3.54 19. Columbia 3.67 20. Illinois 3.55 ... 21. Princeton 3.49 4.92 3.17 3.35 4.11 22. Michigan 3.46 4.33 3.26 23. Pennsylvania 3.07 2.21 3.39 3.09 2.99 24. Yalet 2.79 25. Harvardt 2.25 ° Collection did not reach 2 million-volume level until 1972-73. t Collection reached 4 million-volume level prior to 1962- 63 4.0 Million Vol. Level Pre-4.0 Post-4.0 Mean Mean 0 4.52% 2.61% -; 3.97 3.40 c ~ 4.20 2.56 ;;::s-- !Jj ~ ~ ('\) c, - ~ -:t 538 I College & Research Libraries • November 1976 size every sixteen years, it must maintain an annual mean rate of growth of ap- proximately 4.5 percent. As shown in Table 4, sixteen of the twenty-five li- braries posted annual mean rates of growth of from 4.17 to 7.67 percent. On the one hand, the wide range of the growth rates imposes some restrictions _ upon the use of the Purdue growth-rate figures to forecast collection growth for individual libraries. To elaborate, if the mean rate of growth maintained in re- cent years by an individual library does not correspond closely to the appropri- ate Purdue figure, then for that library the utility of the composite figure as a predictive tool is diminished-unless, of course, the composite figure is cited as the rate of growth that should be main- tained in the future. The composite figure is not a standard which denotes the rate of collection growth that ought to be sustained but instead is a measure of growth comput- ed from the growth rates experienced by X-number of libraries for N-number of years. The authors of the Purdue study also noted this limitation when they acknowledged that their analyses of library growth "may be regarded as suggestive of events within individual libraries, although the course of growth for the individual library is rarely as regular as that of a group."21 On the other hand. the growth rates listed in Table 4 seemingly substantiate the ac- curacy of the Rider hypothesis in pre- dicting library growth; but a closer ex- amination of these and other growth measurements presented in the tables re- veals the limitations of the hypothesis' usefulness. By juxtaposing Tables 3 and 4, it be- comes apparent that the ranking by an- nual mean rate of growth is an impre- cise inversion of the ranking by collec- tion size for 1962--63. A matrix (Table 6) showing the distribution of libraries by rank of collection size and mean rate of growth illustrates this point. TABLE 6 DISTRIBUTION OF LmRARIES: CoLLECTION S1ZE AND GROWTH RAms Rank by Annual Mean Rate of Rank by Collection Growth (1962-63 through 1973-74) Size (1962-63) 1-5 6-10 11-15 16-20 21-25 1-5 6-10 11-15 16-20 ' 21-25 2 1 2 2 1 2 2 1 2 2 1 1 1 3 2 Although Toronto is an obvious ex- ception to the above generalization, its anomalous position in the ranking by mean rate of growth can be attributed to a decision made by the Ontario gov- ernment early in the 1960s to upgrade the quality of graduate education at the university; consequently the library sus- tained a high rate of growth from 1963 through 1970.22 Of course, other exceptions can be discerned, e.g., Princeton and Pennsyl- vania, and undoubtedly their anomalous positions in the ranking by growth rate also could be explained. Nonetheless, these exceptions do not invalidate the observation that the ranking by mean rate of growth is an imprecise inversion of the ranking by collection size-an observation also supported by another matrix (Table 7), which shows the dis- tribution of libraries by mean rate of TABLE 7 DISTRmunoN OF LmRARIEs: GROWTH RATEs AND CoLLECTION S1ZE Mean Rate of Growth (1962-63 to 1973-74) <4 percent ~4 percent Range of Collection Sizes ( 1962-63, Millions of Volumes)