College and Research Libraries By WILLIAM L. EMERSON Adequacy of Engineering Resources for Doctoral Research In a University Library SINCE WORLD WAR II librarians have been aware of the problems caused by the vast growth of publication activi- ties, particularly in the sciences. In terms of discussion, from Fremont Rid- er's The Scholar and the Future of the Research Library to the Allerton House Conference on Problems and Prospects of the Research Library~ there has been concern with the organization and ad- ministration of research libraries as a consequence of this rapid growth. In terms of investigation, there has been during the past forty years a number of studies of various uses of library mate- rials by research workers. 1 One special type of study, however, that has been neglected is concerned with the prob- lems of doctoral research. Only Swank's2 and a more recent unpublished disser- tation by Stevens3 have dealt with the relation of library materials to doctoral research. These few studies have given libraries which are involved in support- ing doctoral research programs little factual know ledge of the needs of re- 1 For an informative study of the se surveys see: Rolland E. Stevens, Characteristics of Subject L iter- atures. ACRL Monographs, No. 6. 2 Raynard Coe Swank, "The Organization of Library Materials for Research in English Literature," Library Quarterly, XV (1945), 49 -74. 3 Rolland E. Stevens, "The Use of Library Materia ls in Doctoral Research: A Study of the Effects of Dif- ferences in Research Methods." Urbana, Ill. , 1951. 149 I. Mr. Emerson is head~ Science and Technology Department~ Long Beach Public Library. NOVEMBER 19~7 search workers. As Dunlap has pointed out, "The paucity of information re- garding the needs and desires of- readers has compelled librarians of institutions of higher learning to make decisions based on a number of widely accepted but unsubstantiated assumptions.'-' 4 The only way to overcome a paucity of information is to gather some. This paper is concerned with the library needs of a small group of readers in a particular field in a university. It con- siders some of the characteristics of the literature used by doctoral candidates for their dissertations in engineering at Columbia University. It also shows to what degree the resources of the Colum- bia libraries were able to support these dissertations. The major hypothesis underlying the study was suggested by "the widely ac- cepted but unsubstantiated assumption" that scientific personnel, in general, pri- marily use recent material of serial na- ture in their research. In other words, the study hypothesized that doctoral re- search in some fields of engineering, !is carried on at Columbia University over the past few years, primarily required recent material of a serial nature in the _!,nglish language. By "recenf' is meant material published within a five-year span of the date of the dissertation. Se- rial material was defined as, "a publica- tion issued in successive parts, usually 4 Leslie Dunlap, "Services to Readers," Library Trends, I (1952), 49. 455. at regular intervals, and, as a rule, in- tended to be continued indefinitely."5 This was in opposition to monographic material, which was defined as, "a sys- tematic and complete treatise on a par- ticular subject, usually detailed in treat- ment but not extensive in scope."6 In addition to identifying these two major characteristics of the literature cited, this study includes an analysis of some of the other features. Each of these two rna jor groups will be identified as to general type of publisher, i.e., a commercial or trade publication, a publication of one of the engineering or other learned or professional societies, a governmental agency publication, a university publica- tion, or a publication of an industrial concern. Secondly, the language factor and the time span of these materials are , analyzed. Such an analysis of the liter- ature might enable the Engineering Li- brary to ascertain how its collection could be organized to serve this portion of its users. The basic assumption of the study was that the items listed in the bibliogra- phies of the dissertations in question would reveal the materials used by the researcher. The methodology was to list on a separate punched card each item listed in these bibliographies. These cards were coded for the factors listed above, and a statistical analysis was made. The dissertations studied were taken from those dated over the years 1950-54. This would seem to give an adequate sampling of the type of research on the doctoral level done in these fields at Co- lumbia. The field of chemical engineer- ing was not included, as such material is p~imarily the concern of the Chemistry Library. There were twenty-three dis- sertations listed in the years under study. They fell into the following subfields: 6 ALA Com_mittee on Library Terminol ogy. ALA C!Iossary of L:~brary T erms With a Selection of T erms ·m Related Ftelds. (Chicago ALA 1943) p 124 6 Ibid., p . 88. ' ' ' . . Electrical engineering 9 Mechanical engineering . . 6 Civil engineering . . . . . . . . 7 Mineral engineering . . . . . . 1 23 Since there was only one dissertation in the field of mineral engineering, it was included as part of the sample, but the evidence of the material within the dis- sertation should be taken as only pos- sibly indicative of the field. For the twenty-three dissertations, there was a total of 761 citations. A total of 756 could be sufficiently identified to war- rant inclusion in the study. This gave an arithmetic mean of 38.8 citations per dissertation. The breakdown of the num- ber of citations per dissertation by fields was as follows: Electrical engineering, 43.9; Me~hanical engineering, 38.5; Civil engineering, 14.7; lVIineral engineering, 27.0. . The major breakdown of the citations between serial material and mono- graphic material revealed that 70.9 per cent of the citations was for serials and 29.1 per cent for monographs. The breakdown within the fields showed the following variations: Mono- Serials graphs (Per Cent) (Per Cent) Electrical engineering Mechanical engineering Civil engineering Mineral engineering 78.4 21.6 65.8 34.2 53.3 46.7 66.6 33.3 Since there are no comparable data in this or any other field, it is not known if the variations within the field presented here are typical or atypical. That there was this variation within the engineer- ing field suggests hypotheses which might be tested to answer the question. These will be discussed later. The gen- eral findings in this respect tend to bear out part of the major hypothesis being tested. Serials seem to be the primary source 456 COLLEGE AND RESEARCH LIBRARIES TABLE I PUBLICATION SOURCES OF MONOGRAPHIC MATERIAL (PER CENT) Industrial Commercial Society Governmental University Company Electrical engineering 70.9 Mechanical engineering . ..... 78.5 Civil engineering ...... . .. 66.6 Mineral engineering ..... . 88.8 Mean for group •• 0 • •••••• 73.4 3.3 2.3 2.5 5.1 27.1 0 0 0 2.4 2.7 18.9 13.9 0 11.2 12.6 4.6 0 6.3 0 3.1 TABLE II PUBLICATION SO URCES OF SERIAL MATE RIAL (PER CENT) Industrial Commercial Soci ety Governmental University Company Electrical engineering 27.7 Mechanical engineering 34.1 Civil engineering ... . 0 Mineral engineering 5 .6 Mean for group . . . . . ...... . 26.5 52.4 2.2 48 .1 11.1 78.0 3.9 88.9 5.5 54.3 5.2 2.9 4.6 16.3 0 4.6 14.8 2.1 1.8 0 9.3 TABLE III LANGUAGE DISTRIBUTION (PER CENT) English German French Other Electrical engineering . ... . .. . . . .. 77.8 11 .8 6.2 4.1 Mechanical engineering . . .. .. .... . . 75.2 14.4 7.1 3.2 Civil engineering . . . . . . . . . . Mineral engineering . ........ . Mean for group . . . . . .. . .... . . in supporting doctoral research in cer- tain fields of engineering at Columbia University. This type of material consti- tutes from slightly more than half in the case of civil engineering to more than three-quarters of the material in the case of electrical engineering, with an arithmetic mean of 70.9 per cen..t for the group as a whole. This would sub- stantiate claims made for the impor- tance of this type of material in the en- gineering field. At this point, it is possible to analyze the publication sources for the materials used in these dissertations. Tables I and II identify the type of publisher and show the percentile ratings for each group in each of the engineering fields 4VOVEMBER 1957 84.4 7.5 1.5 6.5 100.0 0 0 0 80.8 11.7 6.1 1.2 studied. The mean is given for the group as a whole. From Tables I and II it can be seen that commercial, or trade, . publications for the monographs, and society publica- tions for the serials are the two rna jor publishing sources for the literature cited in these dissertations. If it is re- membered that serials accounted for 70.9 per cent of the citations, it can be said that serial publications of societies are the single most cited group. Again it is to be noted that considerable vari- ations are found among the fields stud- ied. The next major characteristic ana- lyzed was the language distribution of the citations. Since this factor operated 457 independently of the type of material cited, no breakdown was made between serials and monographs in this group. Table III shows the language distribu- tion of the citations. It should be pointed out that the min- eral engineering sample included only one dissertation. The languages in the " Other" column were Russian, in the case of mechanical engineering, and Italian, in the cases of electrical and civil engineering. These languages occurred in single dissertations within these fields rather than being distributed through- out the group. French and German, on the other hand, occurred throughout the sampling in the three fields. In all cases, it can be seen that German is the second largest language cited although it falls far below English. There is finally to be considered the time span of the material used in these dissertations. A check analysis of the material revealed that both monographic and serial materials were falling into a similar pattern; so both types were an- alyzed together. Table IV shows the findings. The results presented in Table IV test the other major section of the basic hypothesis. From the evidence presented, it can be seen that this part of the hy- pothesis is not as true as that concerning the relative importance of serial and monographic material. Less than half of the cited material fell into the 0-5 year span, which was taken to be recent by arbitrary definition. A more correct re- statement in this connection would be that although the largest single percen- tile group will probably be found in material of recent date, there will be a wide range of material considerably old- er. Since ahnost 15 per cent of the mate- rial fell into the category, "Over 25," a separate run was made of this group. The findings revealed that in all of the fields represented some of the material used was between 25-50 years old. In me- chanica} and civil engineering fields, some material used was between 51-100 years old, and a scattering of material was over a hundred years old. However, the materials for "Over 50" were con- centrated in one dissertation in each of these fields rather than being distributed throughout the field. These particular dissertations dealt with the historical development of the particular problem under study. In general, then, the characteristics of the research literature cited in these dis- sertations revealed that English language serial material, particularly the publica- tions of the various engineering societies, forms the major group numerically. A detailed ·profile of the literature cited · shows a wide diversity in time as well as in form. German and French are the major foreign languages. It is now possible to turn to the cita- tions and analyze thetn in terms· of actual titles used. The 534 serial cita- tions covered 181 serial titles. The arith- metic mean for the ungrouped data gives an average of 3.38 ~itations per title. This, however; is rather misleading, be- cause when the scores are grouped into a frequency distribution, the median becomes .88 citations per title. This large differential is owing to the fact that of the 181 serial titles, 102, or 56.3 per cent, were cited only once; 59 titles, or 32.5 per cent, were cited between 2 and 5 times and 'only 20 of the titles, or 11.1 per cent, were cited more than five times. The im- plications of this tend to bear out the fact that researchers on a doctoral level must, in general, have access to a wide r variety of resources. There will be a core of basic material, but the pattern re- vealed by this analysis is that there must be a large fringe of material less often used but which contributes to doctoral research. Along with this pattern, there was a pronounced tendency for any given dissertation to show a high citation count for a particular serial title. At least four of the serial titles which had 458 COLLEGE AND RESEAR CH LIBRARIES TABLE IV TIME-SPAN OF MATERIAL CITED (PER CENT) 0-5 6-10 Years Years Electrical engineering 58.4 14.9 Mechanical engineering . .... 27.3 15.2 Civil engineering . .... .. . . . . . . 35.9 12.8 Mineral engineering 55.5 14.8 Mean for group . . . .. .. . . . ... 45.5 14.7 a frequen 1 cy citation of ten or more were all from single dissertations. This again is possibly a normal pattern, but no comparable findings are available to test such a hypothesis. A similar pattern was found when the monograph citations were analyzed. The 222 monograph citations were produced by 192 titles. Of these titles, 181, or 94.2 per cent, were cited only once. Eight titles were cited twice, two titles were cited three times, and only one title was cited four times. No title among the monographs was cited more than four times. On the basis of this sampling, it might be possible to state that so far as monographic literature is concerned, for the studies in these fields of engi- neering, there was no heavily used core. The title which was cited four times was Kent's Mechanical Engineers' Hand- book., a standard reference source. One other aspect of the general pat- tern of literature cited might be men- tioned. There was a decided tendency for each of the engineering fields to be somewhat autonomous as far as the ma- terials used were concerned. Only one serial ti tie was common to all four fields. Of the 181 serial titles, 155, or 85.6 per cent, were cited in one field alone. Twenty titles were cited in two fields and five titles were cited in three fields. The same pattern was even more pro- nounced for monographic literature. Only Kent was cited in more than one field. There is probably a correlation between this factor and the pattern dis- cussed previously for each dissertation to show a high citation count from any NOVEMBER 1957 11-15 16-20 21-25 Over 25 Years Years Years Years 9.4 7.8 6.3 3.2 13.9 5.6 8.2 29.8 10.6 8.7 5.8 26.2 7.4 0 14.8 7.4 10.6 7.0 7.1 14.7 one or two of the serial ti ties. There is the further possibility that self-citation in the serials concerned might be a caus- al factor in this regard. The foregoing analysis of the charac- teristics of the literature cited in these engineering dissertations tends to bear out the hypothesis relating to the rela- tive importance of serial material for this field of applied science. However, there was a considerable spread in time for this material although the largest single percentile · group was in the 0-5 year time span. English is the primary language, followed by German and French. There was present a smaller ci- tation group from other foreign lan- guages. It is difficult to say whether this was the result of a lack of facility in these languages or the absence of rele- vant materials. Finally, there was the tendency of each of these fields of engi- neering to develop to a large extent a literature of its own with little or no overlapping with the other fields. The analysis of these findings gives rise to further hypotheses which might be tested in terms of other, or larger, universes. It might be worth while to find out if the tendency of these sub- groups within the engineering field is common to engineering doctoral work in general or is simply a local character- istic. It might also be interesting to see if this san1e pattern is present in other disciplines, and, if so, to what degree. Lastly, it would be extremely useful to establish workable parameters for the establishment of what constitutes a "core" collection needed to support doc- 459 toral research in any given field. If such parameters could be established, it might be possible to work out mathe- matical models for such collections. More will be said on library support' of doctoral research at a later point. It is now possible to turn to the sec- ond major aspect of this study, which deals with Columbia library resources and the degree to which they were able to provide support for the literature needed in this group of engineering dis- sertations. Because of the proportionate importance of serial materials, they have been analyzed a little more thoroughly in this respect than the monographic material. Since these dissertations were of very recent years, only a spot check was made to see if the rna terials were actually available when the work was being done. Such a check showed that the material in the libraries used on these dissertations was available at the time of study. Of the 192 monographic titles, the Engineering Library held 64.2 per cent, including all of the titles cited more than once. Other libraries on cam- pus, mainly the Physics and Chemistry libraries, held an additional 21.8 per cent of these ti ties. Of the 192 mono- graphic titles cited, 14 per cent were not available at Columbia. Thirteen of these lacking titles were foreign publications, including two foreign dissertations. No check was made to see if this mono- graphic material was available elsewhere in the New York area. Of the 181 serial titles listed, the En- gineering Library held 93, or 51.3 per cent. Other libraries on campus, mostly the various science libraries, held an ad- ditional 49, or 27.1 per cent. Of the 181 serial titles cited, 39, or 21.5 per cent, were not available at Columbia. In view of the fact that serials represent a more important source for doctoral study than monographs, the fact that Columbia showed up less well in its support of the serials than of the monographs might be worth a few remarks. At this point, only one or two possibly causal factors might be mentioned. The book budget for the Engineering Library is weighted roughly about 67 per cent for serials and about 33 per cent for monographs. This is fairly close to, but falls at little below, the corresponding weights for serials and monographs as used by the dissertations studied which, it will be recalled, were 70.9 per cent for serials and 29.1 per cent . for monographs. It seems that the pri- mary factor to be considered in this re- gard is that the Engineering Library, out of its book funds, must support various groups of users. It would be useful to try to find the comparative use of library materials by these various groups, the undergraduates, the graduates, the fac- ulty, and others. Once this information was procured, some method of determin- ing a value judgment among these groups might be worked out and avail- able book funds prorated accordingly. To return to the analysis of serial material, it is worthwhile noting that Columbia was able to supply all of the serials which were common to more than one dissertation. Some additional as- pects of this serial material should be mentioned. Of the 181 serial titles cited, 57.6 per cent were indexed in the Engi- neering IndexJ and 15.7 per cent were indexed in Industrial Arts Index. Of these titles, the Industrial Arts Index did not list any title which was not list- ed in the Engineering Index. Of the 39 titles which Columbia did not have, only ?• or 17.6 per cent, were in the Engineer- zng Ind ex . This fact might tend to sup- port an hypothesis that the serial mate- rial which was not available at Columbia was generally of less importance to the field of engineering itself. The titles lacking were too diverse to draw any single significant conclusion relating to their absence from Colum- bia's resources. The major factor they had in common was that 24 of them, or (Continued on page 504) 460 COLLEGE AND RESEARCH LIBRARIES Adequacy of Engineering Resources (Continued from page 460) 61.5 per cent, were foreign publications. Also it might be pointed out that 11, or 28.2 per cent of these missing titles were listed in the Union List of Serials as be- ing at the Engineering Societies Library. However, it should be added that a com- plete comparative run of the serial titles was not made, and it is possible that the Engineering Societies Library mght show an equal or greater percentage of titles unavailable. The total number of titles, both serial and monographic, which was used on these dissertations, was 373. Of this total, 66, or 17.6 per cent, were unavailable at Columbia. There are no comparable data to show if this is particularly high, low, or a possible median figure. The only findings which are at all relevant are those compiled by Stevens. He stud- ied one hundred dissertations done in five fields at three universities. The fields were American History, Classical Lan- guages and Literature, Education, Bot- any, and Psychology. He sorted the dis- sertations according to the basic method- ology employed, i.e., historical, textual, and experimental. The following table is taken from his dissertation.7 Type of Dissertation Historical ....... . Textual ......... . Experimental 7 Op. cit., leaf 30. Per Cent Not in Libraries 41.36 21.59 11.52 The engineering dissertations studied do not fall neatly into any one of these cat- egories. It is possible that the majority of them might be classified as experi- mental in nature, but at least two would overlap into the historical. There are too many variables existing between this study and Stevens's much more compre- hensive one to make the figures he gives more than merely analogous. In view of his findings, though, it is probable that an assumption might be warranted that no university library collection can- or even should-supply 100 per cent of the materials used in doctoral research. Here again is a topic for further study. What degree of support is given at Co- lumbia in other disciplines in resean;h on the doctoral level? The figures gleaned in this study show- ing Columbia could not supply 14 per cent of the monographic titles and 21.5 per cent of the serial titles cited in these twenty-three recent dissertations. If enough such "bits" of information can be secured, Columbia would have some quantitative criteria for the evaluation of its collection, at least insofar as its ability to support doctoral research is concerned. If, on the other hand_. other libraries which must render support to doctoral research in the fields of engi- neering covered in this study could gath- er similar data, a set of standards for resources in these fields could be set up as a measuring device. Membership Vote Cancels Council on Move 504 The mail vote of ALA membership determined that ALA Head- quarters will not remove to Washington, D. C., as the Council voted in June at the ALA Conference in Kansas City. The vote was 5,749 to set aside the Council action, 2,199 to sustain the action. Under the ALA Constitution, at least one quarter of the membership had to participate in the vote. ALA's total membership is about twenty thousand. COLLEGE AND RESEARCH LIBRARIES