College and Research Libraries 306 College & Research Libraries July 1984 Stack Capacity in Medical and Science Libraries Justine Roberts Academic medical and science libraries provide informational support for patient care and for research leading to clinical and technological advancement and the growth of basic biological .an~ phy~ical knowledge. Despite such high mtenhon, our success or failure often hangs on mun- dane operational detail, whether we are working with automated or manual sys- tems. Accurate estimation of stack capac- ity is one such detail, important both t? successful facilities planning and to proJ- ect planning and design. For sp~~e pla~­ ning, Metcalf noted in 1965 that there IS disagreement in regard to how many vol- umes can be stored in a given amount of space." He rejected most previous work in this area and stated flatly, ''formulas for capacity are dangerous and none . . . are completely satisfactory .' 11 Nonethel~ss, he carefully tabulated linear space requrre- ments for different kinds of "books," adapting a table ''in common use by stack manufacturers [and] used by Wheeler and Githens [in their 1941 work on public li- brary buildings 2 ] . '' 3 Metcalf labeled his specifications clearly: "Volumes per foot of shelf ... when shelves are filled sol- idly." His figures, if not his advice, have been repeated and reflected regularly in scores of planning guides that postdate his work, though they sometimes are pr~­ sented ambiguously with regard to therr use as planning optima vs. their use as ca- . I-to pacity measures . . . Metcalf and most of his later commum- cants differentiate "medical books, " II sci- ence and technology books,'' and ''bound periodicals" from t~e common average. Metcalf's statement IS: . .. When Shelves are Filled Solidly Technical and sci- entific Medical Bound periodicals Volumes per foot of shelf 6 5 5 Volumes single-faced per (7-shelf) section 126 105 105 These figures are repeated by Zachert in advice to pharmacy librarians, 11 and fur- ther developed there to foster space plan- ning calculations: Calculations . . . based on shelves no more than 2/Jds full Technical and sci- e ntific Periodical volumes (* - computed) Volumes per foot of shelf [4] * [3] * Volumes single-faced per (7-shelf) section 84 60 Government advice to medical school li- brary planners reduces Metcalf's 105 vol- umes per single-faced section to 100 vol- umes, closely equivalent at 4.76 volumes per linear foot (VLF) . 12 The third edition of the Medical Library Association's Hand- book states flatly that stack capacity is 4.5 to 5 volumes per foot of shelf, but is unclear as to whether this is a planning optimum or a statement of maximum capacity. 13 Current experience at the University of California at San Francisco (UCSF) had shown that the II standard" VLF figures could be as much as 100 percent too low in providing reliable estimates for mono - graphs and serials stacks. During 1982, samples of several hundred books at the University of North Carolina at Chapel Hill Health Sciences Library, and of sev- eral thousand volumes in the Science Sec- Ju stin e Roberts is head, Library Sys tems Office, Uni vers ity of Ca!if?rnia, ~an Fran cisco 9414~. Th e author was assisted by A lison Bunting, Jacqueline Doyle, Beverlee French, W1m[red K1 stler, Joyce Loeppnch, Rebecca Mar- tin, and Faith Meakin . tion at the University of Guelph Library, yielded shelf capacity width estimates up to twice the standard. 14' 15 The standard fig- ures have been repeatedly cited with little change or verification for at least forty years, and some of the work on which they are based goes back nearly fifty years. 16 It is hard to believe, for instance, that the size of serial binding units has re- mained stable during all this time as, first, individual journal titles have fattened- with more articles per issue-and then split or changed frequency as specializa- tion and editorial management consider- ations intervened. The dramatic increase in medical specialization also must have had some effect on average monograph size. For example, where there used to be only a 21 /2 -inch-thick Gray's Anatomy, there is now likely to be a Gray's plus a se- ries of considerably thinner regional to- pologies. Similarly, the enlarged interdis- ciplinary scope of many academic health sciences library collections is another no- table trend not reflected in available data. Metcalf and others warn, with obvious good reason, against facilities planning that overestimates stack capacity, but they fail to offer adequate or current evidence to support the figures that are offered . At the same time, repetitive and sparsely re- ported local measurement projects con- tinue to be undertaken in the absence of reliable current stack capacity figures for local planning. These various uncertainties and the con- tinuous pressure of planning work prompted the cooperative development of a stack capacity study by librarians at five health sciences libraries and two science li- braries at the University of California (UC) Research Notes 307 (see table 1). Funding support and librari- ans' time were made available through a university-supported research program initiated by the systemwide Librarians As- sociation. The ecumenical effect of the stack capacity problem is indicated by the investigators' working titles : they were branch heads in two cases, from public services at three libraries, and from techni- cal services and systems at the remaining two. The project was initially undertaken as the first step of a larger study, with the in- tent of testing data collection and sam- pling methods at each library. For this rea- son, only small samples, consisting of one hundred serials shelves and one hundred monograph shelves, were measured at each location, and no attempt was made to predetermine serials/monograph weight- ing factors , or to set final statistical re- quirements . The autonomy, varying scopes, widely varying sizes, and very dif- ferent chronologies of the libraries were expected to result in very disparate obser- vational environments and statistics. The former, but not the latter, proved to be true. The pilot study results are therefore reported here with the expectation that they may be useful in other academic sci- ence libraries. METHODOLOGY Participants agreed on a common set of material definitions and exclusions and marked out qualifying tanges on existing or newly drawn stack maps . These were sent to UCSF where each map's stack sec- tions were sequentially numbered. Sec- tions holding over- or undersized books, rare or reference books, and/or unboxed, TABLE 1 UNIVERSITY O F CALIFORNIA- SIZE O F COLLECTIONS libraries Berkeley-Biology Library Davis-Health Sciences Library Irvine-Biomed ical Library Los Angeles-Bio med ical Library San Diego-Biom edical Library San Diego- Scien ce & En gineering Library San Fra ncisco-Health Sciences Library Total Bound Volumes 202,579 157,691 109,612 388,366 144,367 117,240 489,793 Serials Currently Received 3,981 4,819 1,596 7, 068 3,330 2, 062 3,909 Source: University of California library Statistics July 1982. Office of the Assistant Vice President-library Pla ns and Policies, Univer- sity of California System wide Administration [1982] . 11p. 308 College & Research Libraries single issues of serials, were generally ex- cluded. The remaining qualified shelves, distributed through some ten thousand stack sections, constituted the population from which one random sample each of monograph shelves and of serials shelves was drawn for each library. As expected, serials and monographs were shown on the maps to be separately shelved at each of the five health sciences libraries and intermixed at the two science libraries. However, we found that at the University of California at Berkeley Biol- ogy Library (UCB) nearly 27 percent of the stack shelves were in mixed sections, but that this was true for less than 4 percent of University of California at San Diego (UCSD) Science & Engineering Library shelves. In both cases, one half of these mixed sections were arbitrarily assigned to the serials population and one half to monographs . The SAS Statistical Package (Version 79 .5) PLAN procedure was used to select the fourteen random samples of sections and shelves from the seven libraries and to create the work sheets that would be used for data collection. It was also necessary to know the maximum number of shelves that would be found in any section so that a ("randomly" selected) shelf could be identified within each section . Inspection of the maps showed that slightly over one fourth of the monograph sections and 15 · percent of the serials sections were other than the normal7-shelf sections . The SAS procedure could not sort its random selec- tions into an order corresponding to the numerically sequenced stack sections. The computer output therefore was saved into and sorted from separate online files, and the resulting work sheets were printed from these files rather than di- rectly from the SAS program. Data Collection For the two hundred selected sections for each library, data collectors recorded two numbers for each shelf: (1) the num- ber of volumes on the shelf, and (2) the width in centimeters of the shelf space oc- cupied by these volumes. No troublesome data collection problems were reported by the participants, with one exception: the July 1984 draft workbook instructions had assumed a uniform width for filled standard three- foot shelves. Midway through the project we discovered that the filing capacity of standard three-foot stack shelves in the seven libraries ranged at least from 87 to 90cm (34.25-35.4 inches), with four li- braries reporting 88.5 and two reporting 90cm filing widths . Returned work sheets were sent for key- ing to a commercial data entry agency that keyed and verified the fourteen hundred observations from clean, legible copy at a cost of approximately six cents per obser- vation. Times reported for the data collec- tion ranged from 8 to 15 hours (2.4-4.5 minutes per measured shelf). This in- cluded the time required to locate and identify selected sections and to select al- ternate sections when the sample selec- tion was disqualified or invalid for any reason. RESULTS Overall Results The stack populations in this study con- sisted of approximately 68,708 shelves in seven libraries, distributed as shown in ta- ble 2. The number of books and their occu- pied shelf width were counted and mea- sured respectively for a random sample of one hundred serials stacks shelves and one hundred monograph stacks shelves at each library. All measures represent sol- idly packed shelves. For all seven li- braries, the combined serials stacks mea- surements (n = 700) averaged 7.6 (VLF), with a range of 1.6 volumes between the lowest average, 7 VLF at UCSD-B, and highest average, 8.6 VLF at UCLA. Both science libraries were within this range, averaging 7.3 VLF each (see table 3). The combined stack monograph mea- surements (n = 700) averaged 11.4 \lLF, with a range of .6 volume between the lowest health sciences library average, 10.9 at UCI, and highest health sciences figure 11.5 at UCSF. The UCSD Science & Engineering Library was outside this range, at 11.9 VLF; UCB was within it, at 11.5 VLF. The 95 percent confidence interval for the combined 7-library serials and mono- graph means was ± .2, that is, if similar Research Notes 309 TABLE2 UNIVERSITY OF CALIFORNIA STUDY POPULATION Libraries Berkeley-Biology Library Davis-Health SCiences Library Irvine-Biomedical Library Los Angeles-Biomedical Library San Diego-Biomedical Library San Diego-Science & Ensineering Library San Francisco-Health SCiences Library Total All Libraries Percent of Total Serials Shelves 4,358 8,727 3,062 12,407 6,196 3,542 9,037 47,329 (68.9) 1,159 2,868 1,023 5,631 2,965 1,249 4,280 19,175 (27.9) Mixed Shelves 2,016 188 2,204 (3.2) Total Shelves 7,533 11,595 4,085 18,038 9,161 4,979 13,317 68,708 (100.0) TABLE 3 SAMPLE SHELF MEASUREMENTS: VOLUME WIDTHS-ALL LIBRARIES Mean Volumes Format ~Foot of Shelf ( en Shelves Are Standard (n) Filled Solidly) Deviation Serials (700) 7.6 2.4 Monoraphs (700 11.4 2.9 *Using the " unbiased " average (see text). samples were repeated, the averages found ninety-five out of one hundred times would be expected to be between 7.4 and 7.8 VLF for serials, and between 11.2 and 11.6 VLF for monographs. All figures shown in tables 3 and 4 are computed on the basis of solidly filled shelves, and therefore do NOT represent working stack capacity or planning opti- mums, discussed in the next section. It is also important to note that the sampling method used for these tests results in an equal but therefore nonproportional rep- resentation of serials and monograph shelves in the total sample. Thus mono- graphs are overrepresented in the overall mixed format results shown in table 4: the average 9.5 VLF shown for the total sam- ple necessarily understates the average volume thickness at the participating li- braries. A weighted mixed format average can be computed for the combined sam- ples by using the relative serials/mono- 95 Percent Range of Volumes Average Confidence ~er Foot of Inches per Limits ( illed) Shelf Volume* ±.2 2.8-19.6 1.60 ±.2 3.1-31.9 1.06 graph shelf proportion (from table 2) as a weighting element.* This yields a com- puted average of 8.7 VLF for the entire sample. Serials volumes can be seen on the average to be one and one half times the width of monograph volumes, an im- portant planning consideration for science . and health sciences libraries, where serials are dominant. The data show small differ- ences, at least for these samples, between average widths in the health sciences li- braries and science libraries. Health sci- ences serials volumes are found to be slightly thinner than science serials vol- umes, and the opposite true for mono- graphs: science monograph volumes are slightly thinner than their health sciences library counterparts. Slightly different figures resulted from two different methods of calculating an "average" VLF. The first method, used for most of the results reported here, takes the results of each VLF calculation for each *There are roughly 2.5 serials shelves for each monograph shelf in the serials/monograph stack pop- ulation. If this weight is used to increase the contribution of serials measurements to the computed mean, then the number of serials observations increases to 1750 and total observations increase to 2450. The weighting equation is taken from R. P. Runyon and A. Haber, Fundamentals of Behavioral Statistics, 4th ed. (Reading, Mass.: Addison-Wesley, 1980), p.78. 310 College & Research Libraries July 1984 TABLE4 SAMPLE SHELF MEASUREMENTS: VOLUME WIDTHS- COMBINED MONOGRAPHS AND SERIALS*-BY TYPE OF LIBRARY ~£5~~J~:~ ~~~~~~t Range of Volumes Libraries Are Filled Solidi). Standard ~er Foot of (n) (Weighted Mean * Deviation ( illed) Shelf All libraries 9.5 (1400) (8.7) 3.3 2.8-31.9 5 health sci. libraries (1000) 9.4 3.2 2.8-30.5 2 science libraries (400) 9.5 3.3 2.8-31.9 *Monograph shelves are overrepresented in these calculations, so that the weighted mean is a more accurate representation . See text . shelf in the sample and obtains the av~r­ age of these calculations for each (serials or monograph) sample group. This set of means is shown in tables 3 and 4 in the column "Mean Volumes per Foot of Shelf." The second method finds an "un- biased" average, which is computed by adding up all of the book counts for all of the (serials or monograph) shelves, add- ing up the total shelf widths occupied by these volumes, and then dividing the two totals to find an ''average'' VLF for the ( se- rials or monograph) sample group. The unbiased computations are more appealingly "true" than the shelf-by- shelf computations, but do not seem more useful. There is always some shelf capac- ity loss in any real shelf loading situation regardless of the true width of the books: two 3-foot shelves are filled by either thirty-six 2-inch books or by thirty-four 2.1-inch books although the unbiased VLFs differ. The differences between the two sets of averages is small for this group of samples. The average absolute differ- ence for the fourteen sets is .08 VLF, with the unbiased computation showing a thicker book than the per-shelf statistic in twelve of fourteen cases. The unbiased averages were adjusted to correct for the different filled-shelf capaci- ties that were discovered after the test data collection had been completed. Inspection showed the adjusted means to be slightly different for two libraries, but not to an ex- tent warranting reentry of the data. Differences among the Libraries It's usually reported that average VLF measures differ not only according to for- mat, but according to the subject content and binding policies of individual li- braries. Data were analyzed to determine whether any significant differences ex- isted between the VLF averages of the seven UC libraries. This was done using the SAS GLM procedure, which carries out Analysis of Variance tests.t In gen- eral, these tests inspect a group of differ- ent samples to compare the variation found within each of the samples with the variation between the samples. If the vari- ation between the seven VLF averages is much different than the chance variation of the VLF measurements within each li- brary, then there's reason to believe that there may be a significant difference be- tween the libraries with respect to the sta- tistic under inspection. No significant difference was found be- tween the libraries with respect to their monograph VLF averages. For serials, the averages partition into three different but overlapping groups. The VLF averages for serials at five of the libraries, including the t A helpful guide was found in R. J. Freund and R. C. Littell, SAS for Linear Models; a Guide to the ANOVA and GLM Procedures (Cary, N.C.: SAS Institute, 1981). two science libraries, are not significantly different from each other. Only UCLA's and UCSF ' s averages differ from the oth- ers to any meaningful degree, and only UCLA's average stands by itself. That is, its serials VLF difference from the other li- braries is larger than you might expect to occur by chance alone. It is difficult to ac- count for this by subject content, since the broader-than-medical scope of this bio- medical library should bring its average "down" to the science libraries average, rather than "up" to a thinner volume . UCLA reports two conditions that may contribute to this finding. The library's au- tomated serials system, in use for about ten years, provides very flexible bindery support and makes it simple to bind in midvolume. Thus the library's policy of providing small bound serials units has been fully supported for some time. UCLA's thinner-than-average serials vol- umes probably also reflect a 1960s decision to forgo any multiple-year binding unit, regardless of journal frequency or thick- ness. Time Differences between Old and New Material We believed there could be a difference in thickness between older and newer books in the stack collections and wished to inspect the samples accordingly. We found that serials at UCSF, UCLA and UCI (Irvine) are shelved in part according to publication date. The date break-off point is different in each case, and ex- cludes History Collection serials at both Research Notes 311 UCSF and UCLA, but a large portion of the three samples could be divided into "old" and "new" segments. The aver- ages of the pooled old and new samples from the three libraries are as follows: Date Gro up " old" serials " new" serials n 184 114 Mean VLF 7 .789 7.912 Avg. W idth /Val. 1 .54 1.52 The expected t-test procedure to inspect the difference between these means proved to be unavailable, and the calcula- tions necessary for a manual test could not be carried out during the project period. However, the width difference is in the ex- pected direction, and further testing of a larger sample may provide useful results. DISCUSSION How Many Books on a Shelf? The monograph findings in this test confirm those of some other recently re- ported studies and suggest that com- monly referenced older medical/ scientific books-per-linear-foot measures are not a reliable base for estimating the number of monographs to be found on full stack shelves. Table 5 summarizes some of these results, 17-20 together with those re- ported here. Table 5 also indicates that a reliable base for serials, or for combined serials and monograph formats, remains elusive. The 1982 study of serials in the Science Section at the University of Guelph21 arrived at a sample serials vol- ume width statistic close to Metcalf's gen- eral purpose (solidly filled shelf) periodi- TABLE 5 BOOK WIDTH ESTIMATES* MEDICAL AND SCIENCE LIBRARY COLLECTIONS Width in Inch es Source Monographs Serials Metcalf, 1965 SUNY (3 health sciences libraries), 1977 Univ. of North Carolina Health Science Library, 1980 Guelph Science Section, 1982 Lieberfeld (various surveys) Univ. of California (5 health sciences libraries), 1983 Univ. of California (2 science libraries), 1983 2-2.4 1.1 1.13 1.05-1 .11 1.0-1.06 (2.4)t 2.18 1.42- 1.73 1.64- 1.66 Combined 1.4- 2.0 1.33 1.27 1.26 *Calcula ted from "volumes per li near foo t" measures give n by Metcalf, SUNY, and in this report; sh ow n as d irectly re ported fro m Gue)ph and the Univ . of North Carolina studies . +Metcalf's " Bound Pe riodical" statistic, at 5 VLF (2.4-i nch wi dth ), is the same as h is medical book measu re, but is no t limi ted by s ubject. · 312 College & Research Libraries cals figure, but a good 26 percent higher than the highest average discovered for serials in this study's samples of UC medi- cal/science libraries. SUNY's related combined-format projections, which err on the si<;ie of overestimating width, are immediately between the Guelph and UC figures. 22 The statistics from Guelph, the University of North Carolina, and UC are all based on sampling rather than on total populations. However, even "worst case" volume widths calculated from the 99 percent confidence intervals for the UC samples remain under those reported ear- lier (Serials, at UCSD-B: 6.4 VLF, 1.9 July 1984 inches per volume; Monographs, at UCI: 10.3 VLF, 1.2 inches per volume). How Many Shelves Do We Need? Statistics for solidly filled shelf capacity must be further developed for use in facil- ity planning. The final "optimal" plan- ning statistic is reliable only to the degree that the library's growth rate and the num- ber of years to be planned for are both accu- rately predicted. Metcalf's commonly cited 86 percent maximum working capac- ity,23 and a standard shelf filing width of 343 /4 inches (88.3 em) are used in table 6 to show how different growth rates affect the TABLE 6 EXAMPLE OF STACK CAPACITY ESTIMATES HYPOTHETICAL STARTING STACK COLLECTION OF 50,000 VOLUMES Compounded Vol/LFfor Vol/LF for Annual Vol /LF for Shelves Filled to Shelves with 10 Yr . No . of Years Growth Collection Shelves Filled Working Capacity Growth Capacity Provided by Metcalf Growth Solidly Da~ 1 Da~ 1 on Day 1 " Add 50%" Calculation Rate* (No . of She ves) (No . of helves) (No . of Shelves) (No. of Shelves) 3 percent (e.g ., UCSF, 9.5 8.2 6.1 13.7 years 1972-82) (1817) (2106) (2830) (3159) 4.5 percent 9.5 8.2 5.3 9.2 years (e/i., UCI iomed, 1973-82) (1817) (2106) (3270) (3159) *This is the average compounded annual growth rate calculated from the volume holdings figure given in the University of California " Size of Libraries " reports for 1972, 1973, and 1982. A common expression for this growth rate is found in the familiar compound interest rate formula : FV = PV* (1 + r) ", where FV = future amount, PV = present amount, i = interest (i .e ., growth) rate, and n = number of periods for which interest (growth) is being calculated . Thus, UCSF' s holdings in June 1972 of 370,717 volumes grew to 489,793 volumes in June 1982 at an average annual growth rate (compounded) of roughl~ 3 percent, as follows : 489793 = (370717) * (1 + i)10; 489793/370717 = (1 + i)10; 1.32 = (1 + i) 0; 10.j1.32 = 1 + i; 1.028 = 1 + i; i = .028 Campus Berkeley Los Angeles San Francisco Davis Riverside Santa Barbara San Diego Irvine Santa Cruz TABLE 7 PREDICTED VERSUS REPORTED VOLUME HOLDINGS FOR UC LIBRARY COLLECTIONS 1963-1970 No . of Years Projected 7 7 7 7 7 7 7 5 5 Average Annual Compound Growth Rate, • Predicted Holdings .87 6.40 4.75 14.62 13.65 16.00 19.94 26.40 24.57 Average Annual Compound Growth Rate, t Reported Holdings 4.48 6.58 6.42 16.54 18.76 18.01 30.26 31.25 31.12 *Calculated from " Projected Volume Holdings ... Fall Semester" of 1963 (or 1965) and 1970, given in University of California, Budget for Capital Outlay, 1964-65 and Five-Year Major Capita/Improvement Program , 1964-69 (Berkeley: University of California, 1963) . +Calculated from "Bound Volumes" holdings for 1963 or 1965 and 1970, given in Table I of " Size of the Libraries of the University of California on 30 June . . . " published as attachments to CU News 1963, 1965, and 1970. Figures for 1970 are shown as corrected in 1971 for Berkeley and Los Angeles . VLF measures applicable to a hypothetical collection. Table 7 shows the results of comparing UC' s 1964 growth forecasts with the after- the-fact growth rates given in the univer- sity's "Size of Library" reports. The pro- jected growths were frequently far off the mark, despite the short projection pe- riods, and underscore the fact that the problem of arriving at useful shelf capacity planning figures is only peripherally re- lated to the problem of estimating how many books fit on a shelf. An added prediction problem was found to result from the use of area space standards that assume a fixed number of shelves in a stack section. More than one quarter of the monograph and more than 10 percent of the serials stack sections in Research Notes 313 this study had fewer than seven shelves, a loss of 1,495 shelves and nearly 4,500 lin- ear feet assumed to be available to this group of libraries. Planners obviously can- not predict future constraints of site and design, but appropriate early modification of opening capacity estimates appears necessary to an effective ongoing plan- ning process. Means for reliable growth-rate predic- tion were outside the scope of this study, but the width estimate remains a neces- sary base for planning calculations and was found in this preliminary study to vary greatly according to format. Thus the relative format emphasis of collections, as well as their subject content, appears to merit continued discrimination and atten- tion. REFERENCES 1. Keyes D. Metcalf, Planning Academic and Research Library Buildings (New York: McGraw-Hill, 1965), p.153. 2. Joseph L. Wheeler and Alfred Mortori Githens, The American Public Library Building: Its Planning and Design with Special Reference to Its Administration and Service (Chicago: American Library Assn., 1941), p.415 . 3. Metcalf, Planning, p.153. 4. American Library Association, Library Administration Division, Ad Hoc Committee on the Physi- cal Facilities of Libraries, Section on Library Organization and Management, Measurement and Comparison of Physical Facilities for Libraries (Chicago: ALA, 1970) . 5. P . N . Kaula, "Capacity of the Stacks in a Technical Library," Herald of Library Science 7:131-34 (Apr . 1968). 6. Frazer G. Poole, "The Selection and Evaluation of Library Bookstacks," Library Trends 13:419 (Apr. 1965), p.419 . 7. Gayle E. Evans and others, Library Environmental Design: Physical Facilities and Equipment (Los An- geles : University of California Institute of Library Research, 1971), p.131. 8. Lucille J. Strauss, Irene M . Shreve, and Alberta L. Brown, Scientific and Technical Libraries: Their Organization and Administration, 2d ed. (New York: Becker and Hayes, 1972), p.70. 9. Estey Corporation, Library Shelving (Red Bank, N.J .: The Corporation, 1978). 10. William}. Hubbard, Stack Management (Chicago : ALA, 1981), p.41-42 . 11. Martha JaneK. Zachert, Standards and Planning Guide for Pharmacy Service (Bethesda, Md.: Ameri- can Association of Colleges of Pharmacy, 1975), p .15. 12. U.S. Public Health Service, Medical Education Facilities: Planning Considerations and Architectural Guide, PHS Publication no .1180-A-1 (Washington : Public Health Service, 1964). 13. Alderson Fry, "Library Planning, Furniture, and Equipment," in Handbook of Medical Library Prac- tice, 3d ed., eds. Gertrude Annan and Jacqueline Felter. (Chicago: Medical Library Association, 1970), p .305 . 14. Linda L. Brogan and Carolyn E. Lipscomb, "Moving the Collections of an Academic Health Sci- ences Library," Bulletin of the Medical Library Association 70:377 (Oct. 1982) . 15 . Ralph M. Daehn, Space for Growth; the Measurement and Projection of Shelf Space at the University of Guelph Library (Guelph : University of Guelph Library, 1982), p.ll. 16. Wheeler, American Public Library Building, p.414-15. 17. Brogan, "Moving the Collections," p .377. 18. Daehn, Space, p.ll. 19. Karen Wittenborg and John F. Camp, Shelf Space Projection Survey (n.p.: SUNY Council of Head 314 College & Research Libraries July 1984 Librarians, 1977), p. 6, 12. ''We calculated the [library's] average number of volumes per linear foot ... by dividing its collection size by the number of its filled or partially filled shelves." (These projections err in the direction of overestimating book width). 20. Lawrence Lieberfeld, "The Curious Case of the Library Building," College & Research Libraries 44:279 (July 1983). 21. Daehn, Space, p .ll . 22. Wittenborg, Shelf Space Survey, p.12. 23. Metcalf, Planning, p.155. Midwest Library Service You won't find more specialized service .. . anywhere College and university librarians: If there are gaps in services being rendered by your current library jobber, then Midwest Library Service may be what you're look- ing for . Midwest has specialized in service to college and university libraries for 24 years-so long that pro- viding books from even the most obscure publisher is standard practice. For prompt, efficient service on all orders, contact Midwest Library Service. Midwest Library Service 11443 St. Charles Rock Road Bridgeton, MO 63044 Call toll-free (800) 325-8833 Missouri librarians call toll-free (800) 392-5024 Canadian librarians call·collect (314) 739-3100 J ,. Why search the Federal Register page-by-page? · Check for regulations with one look-up Find proposed rules . and notices in seconds Monitor the latest federal announcements Call for a free issue Find out what this innovative-yet economical-new service can do for you. Call our toll-free Now there's a faster, more reliable way to keep posted on federal announcements, rule changes, and dead- lines ... the CIS Federal Register Index (FRI). Save time With FRI you can hone in on exactly Introducing the new weekly CIS FEDERAL REGISTER INDEX number for a free sample issue. Or fill · in and mail the coupon below. There is no cost or obliga- tion whatsoever. 800-638-8380 or 301-654-1550 the information you need-in seconds. Just look under a common-sense subject heading that describes your interests. There you'll find a descriptive notation on each relevant new item in the Register. Find deadlines at a glance FRI's "calendar" section gives you an easy-to- scan rundown of all the upcoming comment deadlines, reply deadlines, hearing dates, and effective dates. It's a conv.enient checklist to assure you don't overlook important oppor- tunities-or dangers. Get currency plus convenience FRI is published weekly, so you'll have time to act before deadlines expire. Plus you get regular cumulations, so you have fewer look-ups when searching for items that appeared in past issues of the Register. ~---------------- 11 n Please send me free information on ~ the time-saving new CIS Federal I Register Index. I I Name/Title I J Organization I 1 Address I I City/State/Zip I Area Code/Phone no. Ill• Congressional Information Service, Inc. • 4520 East-West Highway, Suite 800-C Bethesda, MD 20814 The great · div1de. Eon's new Academic Division rewards you with better service. You now get the individual attention of someone familiar with your needs along with access to the vast resources commanded by Faxon. More than 100 years of dedicated service to libraries have put Faxon at the forefront of the information frontier. We were the first agency to completely automate subscription management. First to develop an international online network. First to establish online links with major publishers. We'll give you an edge on tomorrow. A frontier we conquer by not resting on our laurels. 800 225-6055 or 617 329 -3350 &:.on ON THE FRONTIER OF INFORMATION MANAGEMENT The Faxon Company, Inc. 15 Southwest Park Westwood, MA 02090