College and Research Libraries By LAWRENCE S . . THOMPSON 1 Suggestions for Statistical Records, 1 This is the first part of an article which brings a fresh view of the subject of library statistics and which will be continued in a subsequent number of College and Research Libraries. T HAT MOST librarians dislike statistical records is patent. But without figures capable of intelligent interpretation, we are seriously handicapped indeed. William Thomas Kelvin expressed the need ade­ quately and succinctly, " .. ·. when you can measure whatever you are talking about, and express it in numbers, you know some­ thing about it. But when you cannot ' measure it in numbers your knowledge is of an inadequate and unsatisfactory kind." 2 Especially in res~arch libraries, librarians are ~.lmost universally disinclined toward maintaining such records and presenting them in a utilitarian style. 3 Heinrich Simon observed almost half a century ago that a librarian ·shudders before a statistical report form as does a hard-pressed doctoral candidate before his committee. Also, Georg Leyh, of Tiibingen, the outstanding authority on stati,stical records of libraries, 1 I am indebted to Charles H. Brown, Ralph M. Dunbar, Clyde Cantrell, and Eu~ene H. Wilson for many helpful suggestions of practical value in prepar­ ing .this article. However, any errors of fact or fal­ lacies of speculation are attributable solely to the author. 2 Popular Lectures and Addresses. London, Mac­ millan, 1889-94, I, 8o; cited apud Adams, Randolph G. Three Amet·icanists. Philadelphia, University of Penn­ sylvania Press, I939, p. 53· 3 Brown, Charles H. "Statistical Data and Their Use in the College Library." A .L.A. Bulletin 30:225­ 28, April I936; Simon, Heinrich. ("Osmin"). "Bib­ liotheksstatistische Kuriosa." Zentralblatt fur Biblio­ thekswesen 2I :5I 2'-I 5, I904. Leyh, Georg. "Die Grundlagen einer internationalen Bibliotheksstatistik," p. 103. (In First International Congress of Libraries and Bibliography. Rome-Venice. 1929. Atti. Rome, Libreria dellow Stato, I93I·33, IV, IOJ-I9); and Leyh, Georg. "Die Bibliotheksstatistik." Z entralblatt fur Bibliothekswesen 40:439, 1923. has justly looked with susp1c10n on li­ brarians who consider themselves above the allegedly irksome tasks of compiling and reporting statistics.4 The history of the development of sta­ tistical records in libraries might well be the subject of a monograph. There is a fairly good ·introduction to it by Georg Leyh in the second volume of the H andbuch der Bibliothekswissenschaft. However, it must be sufficient here to point out that the matter of uniform statistical reporting has been of great concern to American librarians ever since the 187o's but that no important advances were made before the 1930's. Especially significant is the recent work of a joint committee composed of representa­ tives from the A.L.A., state library agencies, and the U. S. Office of Education, which drew up uniform statistical report forms for both public and academic libraries, the latter entitled Library Statistical Report for Institutions of Higher Education. 5 This form is used for collecting statistical data both by the A.L.A. and by the Library Service Division of the Office of Education. It is a basic document and represents much intelligent thinking on library problems. However, it has certain limitations, possibly 4 Leyh, Georg. "Die Grundlagen einer internation­ alen Bibliotheksstatistik," p. IOJ. On the other hand, Randolph G. Adams is thoroughly justified in his twen­ tieth annual report of the William L. Clements Library (Michigan University. The P resident's Re­ port, 1942-43, p. 257) in stating that "we have always · contended that quantitative measurements tell little about this library, so that we have presented this sum­ mary before giving the meaningless figure that we added I453 volumes to the divi sion of printed books during the course of the year." Even Leyh accepts this viewpoint in his article "Statistik," p. 567, in Milkau, Fritz, and Leyh, Georg, eds. Handbuch d er Bibliothekswissenschaft. Leipzig, Harrassowitz, I 93 I· 40, II, 566-8o. 11 U.S. Office of Education. Form 8-072. COLLEGE AND RESEARCH LIBRARIES 210 because it is still essentially experimental, although there had been sorne experience with an older A.L.A. form. Report Form While it is not proposed to limit this paper exclusively to a criticism of the Li~ brary Statistical Report for Higher I nstitu­ tions_, that document will be subject to careful scrutiny inasmuch as it should be the starting point for any attempt to .im­ prove uniform university library statistical records. Many suggestions contained in this paper will be practical to only a few librarians for use in their individual statisti­ cal reports. If every possible statistical category were included in one question­ naire, it would break down of its own weight and many libraries might refuse to answer it because it would be too cumber­ some. Some items are especially applicable to university libraries, others to college li­ braries, and still others to noninstitutional research libraries, which is all the more reason for having separate statistical report forms for each of these three types of li­ braries. If we feel it desirable to collect highly specialized information which only a few libraries will be able to supply, we might provide supplemental slip sheets with the report forms, instructing libraries to ignore items not pertinent. An especially suggestive practice is one followed by the New York State Library at Albany. This library publishes in its Annual Report6 not only a table of statisti­ cal records arranged according to the "Re­ vised Form for College and Reference Library Statistics" issued by the A. L.A. · Committee on Library Administration but also other statistical records · of specific interest to itself, e.g._, the number of vol­ umes acquired by salvage from its fire. e Subtitle, A Treasure House of. New York. H~story, 1942, 134-35, ",Appendix." The reports of ~h1~ hbrary in recent years are models of both statlstlcal and narrative excellence. JUNE_, 1945 Limitations of Compilations In any consideration of statistical records, and especially of compilations based on re­ turns from a number of libraries, we must never forget that lack of space and funds often limits any program to publish them. By reason of this, the A.L.A. and the Office of Education have in their files a great deal of information supplemental to that which has been published. A fundamental point in connection with statistical records is the comparison of li­ braries on matters suggested by · the various statistical categories. It is natural to com­ pare libraries, even libraries which are not fairly comparable. The popular mind, whether American, French, English, or German, enjoys nothing more than to indulge in the conceit that his particular national library contains more books than any other library in the world. And some librarians will compare everything, even such figures as the soap and towel expendi­ tures in the New York Public Library and the price paid by the Library of Cop.gress for the V ollbehr collection. Since they feel compelled to do this, we should attempt to make statistical records of libraries as com­ parable as possible. After all, comparisons of an appropriate nature can be useful,· particularly when we are interested in trends and problems of development over a period of years. That it is not impossible to collect uniform data on even the most irrational and variable of human activities is illustrated by the magnificent work of the Federal Bureau of Investigation in organiz­ ing a system for the collection of uniform crime statistics and publishing them regu­ larly in an intelligible manner. A Complete List Needed The answer to the criticism that the list · of institutions on which data )have been annually published (by College and · Re­ search Libraries) "needs revision in order 211 to include a more representative sampling of libraries, better distributed geographi­ cally,"7 is that the criticism is stated inade­ quately, without the possibility of a con­ structive answer from the long viewpoint. Actually, a complete list is needed. Of all types of American libraries on which the A.L.A. has published statistics, only the class including large reference libraries (public libraries serving over two hundred thousand population) is even reasonably · complete. We can accept no omissions if we are to put such records to actual use.8 The Office of Education' College and University Library Statistics_, 1939-409 and previous publications in the series of Sta­ tistics of Public_, Society and School Li­ braries (irregular from I 870 on), are far more valuable than the compilations for college and university libraries for the same year published in the A .L.A. Bulletin_, sim­ ply because the former, although including considerably fewer statistical categories, / attempts to publish figures on all such li­ braries. .An even more urgent reason for making complete such tables as · have been published in College and Research Libraries is that these tables have been the only regu­ larly published figures on many libraries. 7 A.L.A. Committee on Statistics. "Annual Report." A.L.A. Bulletin 34:598, Sept. rs, 1940. See also Burgess, Robert Stone. "The Sources of Library­ Statistics." Unpublished M.A. thesis, Graduate LI­ brary School, University of Chicago, 1942, p. 59· Burgess aptly points out that the present policy of inclusion of college and university libraries in the published tables of statistical records gathered by the A.L.A. is based more on the willingness of librarians to take time off to return the statistical questionnaire than on an evaluation of these institutions as being in any way typical or representative. Furthermore, in any discussion of statistical work we should be careful to distinguish between functions of collecting and preserving and of publishing. 8 Mr. Dunbar writes: "My experience over the las~ fifteen years has been that among the reasons for this situation (i.e., incompleteness) has been: ( r) some private institutions feel that their data are their own · business; (2') some institutions keep their statistical records in a manner which they feel fits their own particular needs and ·c·onsequently do not have their figures in a form to comply with the general blank; (3) some do not wish to invite comparisons with other institutions, as they are either too far in advance or are too poor; (4) many of the religious institution s have the complication of 'contributed services' and methods at variance with those at secular institutions." 9 U.S. Office of Education. Biennial Survey of Edu­ cation in the United _ States, I938-40, vol. 2, chap. 6. Washington, Government Printing Office, 1943. An Annual Handbook ' Obviously College and Research Libraries cannot publish complete statistical records of college and university libr~ries under present arrangements. Perhaps the only way to do this is to publish an annual hand­ book of statistical records of libraries, in­ cluding all libraries in the United States, separated into their appropriate classes. This handbook might possibly be combined with the American Library Directory in a publication similar, to the 1 ahrbuch der Deutschen Bibliotheken or Paul Schwenke's Adressbuch.10 Probably separate sections should be maintained for (I) university libraries, ( 2) college libraries, ( 3) large public reference libraries, and ( 4) smaller public libraries and large circulating systems of a strictly popular character (e.g._, most of the branches of the Circulation Depart­ ment of the New York Public Library). pue to their very nature, special libraries might perhaps be considered in a separate directory similar to those already issued. 11 In addition to statistical information, the first number of this handbook might include historical and descriptive information simi­ lar to that contained in Schwenke's still invaluable but J.lllrevised A dressbuch. This information might be supplemented irregu­ larly, possibly every five years, but the sta­ tistical records of current interest must be published annually. The period covered should be identical for all libraries, insofar as possible. However, it might be necessary to recognize periods of both the calendar year and the fiscal year, inasmuch as the 10 The lahrbuch has been published fairly regularly since its first volume in 1902 and has undergone few­ perhaps too few-changes in scope and content. Paul Schwenke's Adressbuch der D eutschen Bibliotheken. Leipzig, Harrassowitz, 1893; Zentralblatt fur Biblio­ thekswesen, Beiheft no. 10, includes only figures on holdings, book fund, number of employees, and periods when the libraries listed were open and closed. There is also historical, descriptive, and bibliographical in­ formation on each library included in the Adressbuch. 11 Burgess, op. cit., p, 76-77, examines and describes present sources for statistical records of special li­ braries. 212 COLLEGE AND RESEARCH LIBRARIES ~overning bodies of some institutions re­ quire reports for one period and others for another. But under no circumstances should reports be included for three­ quarters of a year. Inadequate Publication Related to the fault of inadequate cover­ age of institutions is that of inadequate publication of information which is avail­ able, a condition which ·may generally be traced to inadequate funds and shortage of personnel. It is regrettable that the latest compilations in College and Research Libraries failed to include the figures on numbers and salaries of professional as­ sistants.12 The A.L.A. probably has spent Headquarters staff man hours on less im­ portant matters, and the A .L.A. Bulletin and College and Research Libraries have published many items of less value than this one ·page. Such import(llht items as the number of days open per year; the schedule of hours in which the library is open (a) for circulation and (b) for reading and study; and the number of newspapers and periodicals currently received, all of which are provided for on the Library Statistical Report Form for Institutions of Higher Education, find no place in the printed tables. There would be no need for omitting them from printed tables if we had an adequately financed yearbook of statis­ tical records. . Special emphasis should be placed by the editors of this prop.osed yearbook on dis­ couraging "confidential items," particularly those on personnel-salaries, classification, 12 It is important to include these categories when we realize that statistical records of personnel are the only categories in which American statistical records are superior to the Swiss (annually published in Der Schweizer Sammlrr) and the German. On the other hand, even these figures can be improved. Burgess, op. cit., points out on _p. 59-60 that the practice of listing salary ranges by departments rather than giving individual salaries of professional staff members vir­ tuaJly nullifies the value of this information except in the few rar e cases where there are but two or three professional employees in each department. hours, vacations, and holidays. Eugene H. Wilson directs attention to the fact that in the last published reports the salary of the librarian of the University of Illinois is not reported and the . salary of the a~sociate librarian is marked "confidential." How­ ever, the biennial report of the trustees publishes the complete salary statistics ·and accounts for every dollar spent by the university. From the very beginning of the serious study of statistical records of libraries, the matter of quantitative measurement of book stock has been a point of general disagree­ ment. The University of Gottingen Li­ brary was distinctly embarrassed · on one occasion because of definitions inadequate for an accurate count of its holdings. In 1854 its size was estimated at 500,000 vol­ umes, but in I8go, when an exact count was made on the basis of carefully defined terms, it could show only 442,3 7 I volumes, a shrinkage on paper of I 2 per cent in a period of constant expansion. Definition of Terms The problem is basically that of defini­ -tion of terms. No two lipraries seem to agree as to precisely what a volume is and what a pamphlet is, 13 whether books should be counted by titles or by bound units, whether duplic-ates are included or not, and 13 The Library Statistical Report for Institutions of Higher Education states that "a volume is any printed, typewritten, mimeographed, or processed work, bound or unbound, which has been cataloged and fully pre­ pared for use." Some libraries distinguish pamphlets from books according to the number of pages they con­ tain (less than one hundred pages in Germany, Switzerland, and the Newberry Library, less than forty-eight in the Bibliotheque N ationale constitute a pamphlet). Others add a certain number of pamphlets together to make one volume [fifteen equal one volume for Schwenke ten equal one volume for modern German librarians ana1 James Duff Brown in Manual of Library · Economy -(London, Grafton and Company, 1937; fifth ed. by W. C. Berwick Sayers), p. 6g]. Tho~as Franklin Currier in "Harvard Rules for Countmg Volumes and Pamphlets" (Library Journal 42:242, 1918) quotes from a pertinent letter from Herbert Putnam: "The disadvantages of the arbitrary distinc­ tion between volumes and famphlets based merely on the number of pages are o course apparent to us, so much so that . . . we discard the distinction entirely in our annual reports." JUNE, 1945 213 so on. The innumerable· points on which there is disagreement reveal clearly the fact that no two libraries may be fairly com­ pared according to present standards of quantitative 'measurement. The only im­ mediate aid in this matter would be our handbook of statistical records in which each library's rules for counting its holdings could be stated concisely, perhaps once every five years. The problem of counting total holdings cannot be solved here. A special committee in the Library of Congress has been trying to work out a basic procedure for an · ac­ curate recount of that library for two years without coming to any definite conclusions. While we may never be able to compare libraries accurately on the basis of volume count, it is conceivable that other standards of quantitative measurement of actual hold­ ings may be used as bases. One standard of this sort is measurement of book stock in linear meters or feet. The notion of measurement in linear meters was first proposed by Christian Berghoeffer, then librarian of the former Rothschild Library in Frankfurt am Main, in his art.icle on "Messen und Zahlen bei Festellung des Biicherbestandes" in I893.14 The editors of the Zentralblatt fur Biblio­ thekswesen expressed themselves in afoot­ note to Berghoeffer's article as being very skeptical of the validity of this procedure. Nevertheless, the idea has gradually won headway in England, Norway, Italy, and Germany.15 14 Zentralbiatt fur Bibliothekswesen IO:J.26·JI, 1893. 15 Madan, Falconer. "Statistical Survey of the Bodleian Library, with Notes on Book Standards, by Bodley's Librarian." Bodleian Quarterly Record I :.254-6.2, 1916. Current numbers of the Annual Re­ port of the Library Syndicate, Cambridge University Library, give linear footage of volumes and pamphlets added to the "Upper Library." Italy. Ministero dell' lstruzione Pubblica. Le Biblioteche Governative Italiane nel MDCCCXCVIII; Notizie Storiche, Biblio­ graphic he e Statistiche. Rome, Societa Editrice Dante Alighieri, 1900. This Italian compilation not only gives linear meters of shelf space actually occupied by books but also notes the number of linear meters of shelf space available for future expansion. "Regier for Bibliotekstatistik." Bog og Bibliotek 6:.210-IJ, 1939. Linear Measurements Berghoeffer's original plea for measure­ ment in linec:tr meters was partially based on the rapidity of his method. The Inter­ national Federation of Library Associations Sub-Commission on Statistics of - Libraries reported that six hundr~d thousand volumes were measured in four days by one profes­ sional and one clerical em.ployee at the U ni­ versity of Tiibingen but warned that it will be considerably more difficult for a classified library in which the shelves are not filled solidly.16 However, the really significant · aspect of the scheme of measuring a collec­ tion of books in linear feet is in providing a check against such discrepancies ' in count­ ing either volumes or titles as may be caused by peculiar administrative procedures of individual libraries. As to methods of measureme~t in linear feet (or meters) , Georg Leyh, a confirmed advocate of meas­ urement, gives very simple instructions: "Length is easured without regard to size or form. Books charged out at taking of the inventory are included in proportion of 3 3 vols. = I meter's length. " 17 New accessions could be readily measured simply by filling solidly a new bookshelf of standard size before sending the books to the stack and counting the number of times that the shelf was fill~d. As already indicated, measurement has its cnttcs. J. Muszkowski, director of the Biblioteka i Muzeum Ordynacji Krasins­ kich in Warsaw, argued that measurement was unfair because English publishers bind 16 [Reportl. Publications of the International Fed­ eration of ibrary Associations, IV, 193.2-Actes du Comite ' International des Bibliotheques, sme Session, Berne, Juin 9-10, 193.2, p . .21. However, there has been at . least one recent example of a large American library which has been forced to surrender the "Dogma der systematischen Aufstellung" in favor of shelving by accession number because of the great difficulty of frequent shiftings. Such a retreat from classified shelving, if it becomes general, would help open the way for estimating extent of holdings by measurement in linear feet. 17 "Statistique lnternationale des Bibliotheques." Publications of the International Federation of L1.brary AssociationS, V, 1934-Actes du Comite International des Bibliotheques, 6me Session, Chicago, October 14, Avignon, November 13-14, 1933, p. 139. COLLEGE AND RESEARCH LIBRARIES214 I more books and use thicker paper than con­ tinental publishers.18 Paul Gering offers more objections, which are not listed here because he seems to unde~stand measuring as an auxiliary, not as an absolute device for ascertaining the size of a library.19 On the other hand, the idea was carried to an absurd extreme when L~yh proposed to measure not only current acquisitions and total holdings but also manuscripts and incunabula. 20 Measurement is primarily for those parts of the collection which are not otherwise adequately described. Incunabula, manuscripts, maps, prints, dissertations, photographs, and mus1c should all be counted separately. The An­ nual Report of the Library Syndicate~ Cambridge University Library, notes whether maps are received by q>pyright (ordnance, other official, miscellaneous, and atlases), purchase, or presents and ex­ changes (English and foreign). It might also be advisable to have a separate rubric for Short Title Catalogue books, particu- • larly in reporting annual accessions, as a guide for future revisions of the list of American locations for books printed in .England prior to I640. If at all possible, it would be desirable at least to give an aJ?proximate indication . of the proportion of the total number of volumes which are government documents. This practice, which is followed by the American Council on Education's American Universities and Colleges~ is especially useful as a qualitative standard for evaluating small college li­ 18 International Committee of Libraries, Sub-Commis­ sion on Library Statistics. [Report]. PublicaJions of the International Federation of Library Associations, V, 1934-Actes du Comite International des Biblio­ theques, 6me Session, Chicago, October 14, Avignon, November 13-14, 1933, p. 56-57. We need more discus­ sion of how size, format, kinds of bindin~, paper thick­ ness, etc., are going ·to be accounted for m a , system of linear measurement. 19 "Die Bibliotheksstatistik in Deutschland." Allge­ meines Statistisches Archiv 17:99-100, 1927. This article d ese rves careful study as the basic work on German library statistics. It will be found that many of the German practices will shed light on problems of American libraries. 20 "Statistique International des Bibliotheques," loc. cit., p. 144. 146. braries. As for keeping actual statistics of document acquisition, the practice of re­ cording all separate pieces of · U.S., foreign, state, and municipal documents may .be recommended (e.g.~ as at Stanford U niver­ sity Libraries). A fair estimate of the number of duplicates available for exchange might be of some value, especially if classi­ fied as to subject field. The Library Statistical Report for I nsti­ tutions of Higher Education misses the mark completely when "photographs, pictures, prints" are lumped together as though they represented a single type of material. This grouping might conceivably be legitimate in a . small public library, where the form is less important than the fact that the items are similar kinds of visual aids, but for a reference library to place its fifteenth-cen­ tury woodcuts in the same category with the postal cards donated by a library-minded citizen is hardly appropriate. But perhaps the most important single improvement of methods of counting total holdings would be to draw up uniform methods for counting current acquisitions. The day will come when present holdings are but a fraction of total holdings; and if accessions are counted in a uniform manner over a period of years, there may be some future possibility of both quantitative and qualitative comparison of libraries. At present the Library Statistical Report for Institutions of Higher Learning · provides for current acquisitions solely from a fiscal viewpoint, analyzing them only by their source (purchase, gift, exchange, binding of material not included in the first three items, and restoration of volumes previously lost or withdrawn) .21 21 It would be most desirable for university libraries to ·analyze their exchanges in some detail in their annual reports, noting esJ.>ecially the numbers of do­ mestic and foreign institutions with which an exchange relationship is maintained for each .publication or for all publications of each scientific society or institute which deposits its publications with the library for exchange purposes. The Annual Report of the Li­ brary Syndicate, Cambridge University Library, ana- JUNE~ 1945 215 A needed improvement in statistical methods of American libraries is to classify current accessions. Everyone concerned with the library would find that a record of this sort would be invaluable. For the scholar it would be a partial guide to the library's holdings in his field. For the lay governing body it would be the most effec­ tive . way of understanding concretely the library's growth and contents . . Above all, for the librarian it would be a guide to book selection, apportionment of funds, evalua­ tion of trends, accurat~ comparison with other libraries, general policy, and innumer­ able other possibilities in setting up stand­ ards of service. The fact that the Iowa State College Library has a budget of about $8o,ooo for books, periodicals, and binding, as compared with book budgets of $150,000 for some other institutions of university rank, does not mean much unless it can be shown that Iowa State's acquisitions are largely confined to the pure and applied _ sciences. When holdings of the one hun­ dred most frequently cited chemical periodi­ cals were checked in the Union List of Serials_, Iowa State ranked among the first four in compl~teness of its sets, far ahead of many institut,ions with book budgets two and three times as large. Classification of acquisitions by subj~ct fields would show roughly the same thing over a period of years, although not quite so accurately as this detailed and painstaking study. Few American libraries entertain such advanced conceptions of statistical records as the Library of Congress, which tells us every year the exact number of volumes classified and shelflisted by the Subject Cataloguing Division in each major classifi­ cation. The slight additional work of keep­ ing classified records of titles passing through the hands of classifiers and subject catalogers is surely commensurate with the lyzes presents and exchanges by countrx of origin. See also Iowa State College Library, Report, 1933-35. 1935. p, 8. value of the information made available thereby. An alternative method is to count directly from the shdflist. Most libraries usually let new shelf-list cards, or those to which additions are made, accumulate for a period of one to four or more weeks. When these cards are in their preliminary classified arrangement preparatory to filing, it would be very simple to divide them into classes and count the number of new accession num­ bers in each class. Libraries which use the original order card as the shelf-list card could also ascertain from this source ]ust how much money is spent for acquisitions in each class. And it cannot be emphasized too strongly that classified acquisitions by volumes or titles mean little unless there is ~lso information as to how much money is spent on each class. Apportionment of book funds by departments of instruction should always be noted separately in the library's annual report, although there may ,be no way to put this information into statistical tables covering a number of libraries. For almost half a century the lahrbuch der Deutschen Bibliotheken ha~ been clas­ sifying accession statisdcs. Acquisitions of" books are divided into eleven broad, but well defined, categories, viz._, general works; theology; law and political science; eco­ nomics; medicine ; natural sciences and mathematics; technology; horticulture, agriculture, and forestry; history; philoso­ phy and pedagogy; and art. While un­ doubtedly there ~re occasional varying classifications of books by different libraries, the subject classes are broad enough to per­ mit over-all uniformity. We might use the ten main classes of the Dewey decimal sys­ tem in this country. All librarians are familiar with them, and it is a very simple matter to assign a book to one of the main Dewey classes at the time it is being classi­ fied under the Library of Congress system COLLEGE AND RESEARCH LIBRARIES 216 or any of those used by such institutions as Princeton or Harvard. Another important breakdown of new acquisitions almost completely ignored in America is the division of current acquisi­ tions into new and secondhand books. The statistical records of Swiss libraries, pub­ lished annually since I929 by the Ver.einig­ ung Schweizerischer Bibliothekare in Der Schweizer Sammler~ include tables of per­ centages for the relative amounts spent on new books, secondhand books, continuations, and periodicals. The 1 ahrbuch der Deutschen Bibliotheken lists exact expendi­ tures for new books, continuations, periodi­ cals, secondhand books (in two separate sections, (a) from the fifteenth through the seventeenth centuries and (b) from ,the eighteenth through the twentieth centuries), and manuscripts. The Annual Report of the Library Syndicate~ Cambridge Univer­ sity Library, classifies purchases as to whether they are new (subdividing by country of origin), secondhand (no sub­ division), or come by "special grant." Information showing the relative amounts spent on secondhand books and current trade books would seem to be in­ valuable for the librarian. It is difficult to understand how any workabl~ purchasing policy can be set up without it. Without such figures a librarian is in no position to organize most advantageously large co­ operative buying projects. He cannot tell trustees and legislators precisely who is getting his trade or logically present his requests for appropriations. With them, relations between librarians and dealers • might be placed o; a sounder footing and divested of some pf the mutual suspicion which is almost traditional. Particularly in American libraries, which st.arted grow­ ing only in the latter half of the nineteenth century and have never had adequate oppor­ tunity to acquire systematically books pub­ lished prior to I 8oo, a record of the relative JUNE~ 1945 amounts spent for nova and antiquaria would be a significant guide to the extent of the librarian's efforts to satisfy the needs of scholars. But however useful this information on the relative ·amounts spent for new and secondhand books may be, it must be recog­ nized that it would necessitate much addi­ tional bookkeeping. Before including this item in any statistical report form such as the Library Statistical Report for I nstitu­ tions of Higher Education~ it would be advisable to conduct a special study in order to ascertain how frequently the data need be collected, from what sources they are available, what additional work their com­ pilation might entail, and how far their compilation might be limited to sampling. Another item which might be considered in recording the disposition of the book fund is the relative amounts spent for domestic and foreign books. The Cambridge U ni­ versity Library classifies its purchases of new books according to whether they are published in the United Kingdom, British possessions, the United States, Germany, France, or elsewhere. A breakdown of this sort was proposed by Leyh in his outline for a system of international library sta­ tistics. 22 These figures might have some significance in revealing the policies of cer­ tain municipal public library systems, which have to serve large numbers of patrons whose native language is not English, or for studies similar to Waples and Lass­ well's National Libraries and Foreign Scholarship. Binding is rarely given adequate statistical treatment. The Library Statistical Report for Institutions of Higher Education covers binding only by including a blank for ex­ penditures for "binding and rebinding." Virtually all compilations have a columri for bindery . expenditures, but none of them 22 "Statistique International des Bibliotheques," loc. cit., p. 144·54· (and few annual reports of individual li­ braries) gives the . actual number of books bound, rebound, repaired, and placed in a quasi-bound state. A rubric to provide for these categories would be significant in that it would show clearly, for example, how a scientific library, in which periodical litera­ ture is the most important type of publica­ tion received, has far heavier binding loads than a library whose acquisitions are largely books and other types of monogra.phic pub­ lications. For all a legislator or trustee ,knows, a library which reports five thousand dollars in bindery expenditures may be acquiring fifty especially elegant collectors' bindings rather than spreading its resources ,. judiciously for more practical work. De­ tailed bindery figures are all the more important because there is no direct and constant relation between binding expendi­ tures and expenditures for book purchases. 23 A significant figure for correlation with bindery statistics might be enumeration of the bound and .unbound books received by the library, a practice followed by the Cam­ bridge University Library in recording volumes received by it under the Copyright Act. A few libraries distinguish themselves for the exactness of their bindery statistics. Especially noteworthy are the analyses of bindery figures given in the Annual Report of the Librarian of Congress~ the University of Lund's Arsberattelse~ the lahresbericht of 23 Leyh, Georg. "Die Grundlagen einer Inter­ nationalen Bibliotheksstatistik," /oc. cit., p. 114. · the Deutsche Buche rei (a ~odel of library statistical records in other respects as well) , and the Bodleian's Report of the Curators (otherwise very weak in statistical records). Each of these libraries goes into careful detail as to the number of volumes or pamphlets put into the vari