Ackerson.p65 468 College & Research Libraries November 2003 468 Identifying the Role of Multidisciplinary Journals in Scientific Research Linda G. Ackerson and Karen Chapman Linda G. Ackerson is Assistant Engineering Librarian and Associate Professor of Library Administration at the University of Illinois at Urbana-Champaign; e-mail: lackerso@uiuc.edu. Karen Chapman is Busi- ness Reference Librarian, Angelo Bruno Business Library, at the University of Alabama; e-mail: kchapman@bruno.cba.ua.edu. Scientists in focused research areas customarily use specialized jour- nals, and yet multidisciplinary journals also are widely cited. Prior stud- ies have investigated the characteristics of multidisciplinary journals, but none have considered the role this type of journal plays in scientific research. Citation data from Nature, Science, and Proceedings of the National Academy of Sciences were used to profile the articles in the journals and the articles that cite them. In particular, when citation oc- curred across disciplines, the reason for the citation was investigated. cientific communication is ex- pressed in a wide array of for- mal methods, but especially as journal articles. Reports of sci- entific research and correspondence among scientists are published primarily in three types of journals. Some journals, such as Journal of the American Chemical Society, Mathematical Reviews, and Physi- cal Review Letters, publish articles devoted to a single discipline. Others adopt a nar- row focus and publish articles of primary interest to scientists working in special- ized or interdisciplinary areas. Medical Engineering & Physics is an example of this type of journal. Finally, multidisciplinary journals, such as Nature and Science, pub- lish articles from all scientific disciplines. Only small, limited studies have fo- cused on multidisciplinary journals, even though they are widely cited in the lit- erature of many disciplines. This paper reports on a comprehensive study to de- termine the function this type of journal plays in supporting scientific research. The specific aim of the study was to de- termine whether multidisciplinary jour- nals facilitate the transfer of information across disciplinary lines. Citation data from three journals were used to investi- gate these questions. Evolution of Multidisciplinary Journals Brian C. Vickery’s excellent volume on the history of scientific communication traces the development of journal literature.1 Prior to the sixteenth century, individu- als who kept records of observations were largely unaware of one another. Geogra- phy was a powerful barrier to sharing information. The formation of universi- ties in densely populated urban areas and the succeeding development of organiza- tions such as the Royal Society of Lon- don permitted scientists to attend infor- mal meetings and exchange information. The minutes of these meetings were for- Identifying the Role of Multidisciplinary Journals in Scientific Research 469 mally documented, but the proceedings were slow to arrive to members of the society and not accessible to nonmem- bers. It was evident that a central source was needed to effectively disseminate information from formal meetings, infor- mal discussions, and reports of experi- ments and discoveries. Philosophical Trans- actions of the Royal Society of London was the first English-language journal estab- lished to maintain communication among scientists in societies. The Transactions, first published in 1665, included reports from all areas of science. During the 1700s, science flourished throughout Europe, accompanied by the proliferation of scientific societies. As the amount of knowledge increased, scientists began dividing themselves into more spe- cialized areas and, as a result, the first spe- cialized journals were established. With a smaller scope of coverage, scientists in the same specialty could share results more effectively without having to sift through all studies published at the same time. In addition, studies could be replicated more easily and discussed to stimulate contin- ued work on research projects. By the late 1800s, scientists found they were unable to keep up with the increasing number of new scientific developments, not only in their own disciplines, but in others as well. The pace at which full reports of original research were published and dis- tributed was slowed due to the larger amount of literature being published in all subject areas. Multidisciplinary journals were first established as news magazines that could provide brief reports of com- pleted research before the full reports were published elsewhere in peer-reviewed jour- nals. By the 1960s, even news journals could not keep stride with new developments, so letters journals began to appear. Unlike multidisciplinary journals, letters journals were used to announce the results of re- search in one discipline, especially in newly developing interdisciplinary areas. Over time, letters journals were subdivided into very specialized newsletters, such as Arthro- pod-Borne Virus Information Exchange. Most recently, some journals have begun publish- ing accounts of research online first for rapid dissemination, followed by publica- tion in the print journal. Despite the continuous subdivision of research—and therefore of journals— multidisciplinary journals maintain a strong presence in scientific correspon- dence. Impact factors for Nature, Science, and Proceedings of the National Academy of Sci- ences of the United States of America are among the highest in all disciplines. For this reason, these three journals were se- lected as the subjects of this study. Nature Nature began publication in the United Kingdom on November 4, 1869. The earli- est issues were composed of news items but also included letters to the editor, edi- torials, book reviews, and a few reports of original research. Letters to the editor changed over the years. In the early issues, they were primarily comments on earlier articles. Around 1930, letters to the editor began to look like short research notes. By 1970, a well-developed table of contents had been added and what is traditionally considered as letters to the editor appeared in a section called “Correspondence,” while short news items appeared in “News and Views.” Currently, this weekly jour- nal features results of original research; a variety of informal material, such as opin- ion pieces and news stories; contributed material, such as correspondence and com- mentary; literature reviews; and book, soft- ware, and product reviews. Nature is part of a large family of inter- national titles published by Macmillan Journals, Ltd. Seven titles in the family— Nature Genetics, Nature Structural Biology, Nature Medicine, Nature Biotechnology, Na- ture Neuroscience, Nature Cell Biology, and Nature Immunology—publish original re- search in specialized areas. Three journals in this family—Nature Reviews Genetics, Nature Reviews Molecular Cell Biology, and Nature Reviews Neuroscience—provide monthly literature reviews. The general weekly journal Nature is the subject of this study. 470 College & Research Libraries November 2003 Science The first issue of Science was published on February 9, 1883. The stated aim was that Science, published in the United States, would be equivalent to Nature but would focus on publishing immediate reports of scientific research performed by American scientists. Early issues in- cluded some full-length reports of origi- nal research; short research letters; news items; correspondence or letters to the editor; and book reviews. By 1885, brief research letters became more numerous and the journal assumed more structure, grouping types of articles. Science suspended publication after volume 23, no. 581, in March 1894. When it resumed publication on January 4, 1895, the numbering started over with volume 1, no. 1. This new series covered math- ematics, all physical sciences, engineer- ing, life sciences, psychology, and anthro- pology. The editor’s remarks about the new series stated that the purpose of the journal was to facilitate communication among all scientists in the United States on topics of broad interest, rather than in specialized areas. The American Associa- tion for the Advancement of Science has published Science since 1901 as a means of distributing the proceedings of meet- ings and scientific correspondence among the members of the association. By 1930, the contents of the journal also included summaries of scientific events, such as honorary and grant awards, along with reports on apparatus and methods. Dur- ing the 1970s, life science articles began to appear more frequently in this journal. Proceedings of the National Academy of Sciences of the United States of America The first issue of the Proceedings of the Na- tional Academy of Sciences of the United States of America (PNAS) was published on Janu- ary 15, 1915, as the official organ of the National Academy of Sciences. It pub- lished reports of business and scientific meetings and included reports on awards and academy activities. It also was launched to serve as a vehicle for prompt publication of original research conducted by members of the academy. Because the journal attempts to serve all scientists rep- resented in the academy, it favors the pub- lication of papers that are of broad interest to scientists from many disciplines. At the beginning, PNAS covered math- ematics, all physical sciences, life sciences, psychology, and anthropology in a single volume. In 1965, engineering was added to the list of disciplines covered. Articles on biomedical topics became dominant by the 1970s. In 1980, PNAS broke into two parts, and the scope changed from pub- lishing meeting proceedings to publish- ing reports of original theoretical and ex- perimental research written by members of the academy. Part 1, “Physical Sci- ences,” covered mathematics, astronomy, chemistry, physics, and statistics. Part 2, “Biological Sciences,” covered biochem- istry, biophysics, botany, cell biology, de- velopmental biology, evolution, genetics, microbiology, neurobiology, physiology, population biology, and ecology. In 1985, PNAS reverted to a single publication and published fewer articles by being more selective in accepting manuscripts. Literature Review Earlier studies and informal observations of multidisciplinary journals have sug- gested that they serve one or more of the following functions: • Stimulating new ideas: Carole L. Palmer ’s study of scientists at an interdis- ciplinary research center showed that sci- entists believe it is very important to re- main aware of current research outside their own areas.2 Reading the primary lit- erature in other areas was too time-con- suming. Instead, they regularly scanned a number of general and multidisciplinary journals, such as Scientific American, Na- ture, and Science, to find new ideas. A 1960 study of research chemists found that the most creative scientists read twice as much literature outside their own specialties than those who were less creative.3 By reading widely, scientists sometimes come across an interesting analogy or read an essay that suggests a different perspective on a problem. Identifying the Role of Multidisciplinary Journals in Scientific Research 471 • Disseminating information to a broader audience: Multidisciplinary journals publish papers from a variety of disciplines, but the papers are not necessarily interdisciplinary.4 An article spanning more than one disci- pline may potentially double its readership. In addition, a combination of papers from the sciences, social sciences, and engineer- ing in a single journal is likely to reach a more diverse audience of readers. • Providing an outlet for unusual papers: Some studies involve research that does not fit into neat categories. For example, a newly emerging interdisciplinary area may not be developed enough to support a spe- cialized journal. If the specific topic does not fit into any existing specialized journals, a multidisciplinary journal may be the only appropriate publication outlet. • Providing cohesion within disciplines: S. Sarasvady and P. Pichappan hypothesized that the impact of a multidisciplinary jour- nal such as Nature would be comparable to the impact of a specialized journal within the biomedical literature.5 They selected the sixty-three journals that the Institute for Sci- entific Information assigned to the subdis- cipline of immunology, added Nature to the list, and ranked them according to impact factor. Nature was ranked twelfth on the list, affirming its place as a core journal in the subdiscipline of immunology. Selection of the Sample Multiple journals were selected for the study because the scope of journals may differ. Each may cover specific subject ar- eas or favor one type of article over others.6 In separate studies, Braun and others and Kaneiwa and others compared Nature and Science.7,8 They found that Nature published papers from many countries in Europe, the United States, Australia, Canada, Japan, Switzerland, Sweden, The Netherlands, and Israel. Science, on the other hand, gave preferential treatment to the publications of scientists from U.S. institutions. Articles from Nature, Science, and PNAS that were published in 1997 formed the basis for this study. The year 1997 was selected because it allowed enough time for the three-year citation peak to have occurred. The size of the population was assessed by the number of issues pub- lished in 1997. Nature and Science each published fifty-one issues, and PNAS published twenty-six issues. Within each issue, the number of articles was counted by the article type. A previous study of the citation impact of letters to the editor influenced the decision to include more than just original reports of completed research, demonstrating that different types of articles may serve different pur- poses.9 Four types of articles are pub- lished in multidisciplinary journals, so the number of each type of article in each is- sue was counted in order to generate a stratified sample, which would reflect the distribution of each type of article. The first type of article is correspon- dence/opinion, which includes scientific correspondence among scientists, commen- taries, essays, traditional letters to the edi- tor, and public forums on scientific policy. The second type is completed research, which includes full-length articles of origi- nal research and brief research-quality notes that are published ahead of the full papers. The third type is reviews, which are vari- ously titled as insights, reviews, progress reports, and perspectives. The fourth type of article is research in progress, composed primarily of brief summaries of ongoing research. Colloquium papers are included in this type, as the authors routinely sub- mit their papers for comments prior to sub- mitting them for peer review. Only signed articles were included in the count; named column editors were considered to be au- thors. Issue introductions, corrections, and obituaries were omitted, along with book and software reviews. The counts for each type of article in each issue, summarized in table 1, were submitted to the Illinois Statistics Office, a statistical consulting center operated by the Department of Statistics at the Uni- versity of Illinois at Urbana-Champaign. The office used a sophisticated random number generator to create a stratified random sample of articles for each jour- nal that reflected the numbers and types of articles that appeared in the journal. 472 College & Research Libraries November 2003 The sample comprised sixty correspon- dence/opinion articles (twenty each from Science, Nature, and PNAS); sixty com- pleted research articles (twenty each from Science, Nature, and PNAS); twenty-three review articles (five from Nature and eigh- teen from PNAS ); and forty research-in- progress articles (twenty each from Sci- ence and PNAS). Not all journals pub- lished the same types nor the same num- bers of articles, which was reflected in the stratified sample; the total number of sample articles was 183. Having identified a stratified random sample of articles that represented the ar- ticles published in 1997 in the selected multidisciplinary journals, the next step was to consider the articles that cited them. Each article in the sample was searched in Web of Science to identify the articles from the year 2000 that cited them. It was found that the 183 articles were cited a total of 7,551 times. Because this number of articles was too large to work with, a sample was created. Using the random number gen- erator in the Excel software program, 10 percent of the citations of each cited article was randomly selected and 741 citing ar- ticles were identified. Data Collection The following data elements were gath- ered from each sample article: basic bib- liographic details (title, author, and page numbers); the number of times the article was cited in 2000; and the subdiscipline and discipline of the article. The follow- ing data elements were recorded about each of the selected citing articles: basic bibliographic details (title, author, and page numbers), and the subdiscipline and discipline of the article. Each article was assigned a unique identification code to tie together the sample articles and citing articles so that they could easily be iden- tified for further analysis. The assignment of disciplinary labels is crucial to validate the cross-citation behav- ior among subject areas. One can safely as- sume that articles from the Journal of the American Chemical Society are about chem- istry and articles from Tectonics concern earth sciences. The Institute for Scientific Information’s Journal Citation Reports is an appropriate method of verifying this. However, multidisciplinary journals pub- lish articles from all subject areas, so it is not possible to ascertain the true nature of an article only from the title of the journal. Using Journal Citation Reports in this in- stance would imply that all articles in multidisciplinary journals are multidisci- plinary. Therefore, information about the articles in multidisciplinary journals must be collected at the article level, rather than at the journal level, in order to measure the characteristics of individual articles.10 A three-step method was employed to identify the disciplines of both sample and citing articles. Each article was searched in the Current Contents data- base. By using keywords assigned to the article and reading the abstract, the subdiscipline(s) of an article was deter- mined. When an article could not be lo- cated in Current Contents, it was searched in Science Citation Index and the key- words and abstracts were used to deter- mine the subdiscipline(s). In instances where information about a paper could not be located using one of these two in- dexes, the entire article was read and TABLE 1 Number of 1997 Articles, by Type Journal Correspondence/ Completed Reviews Research in Total Opinion Research Progress Nature 740 862 13 0 1,615 PNAS 71 2,515 18 67 2,671 Science 957 1,086 0 350 2,393 Total 1,768 4,463 31 417 6,679 Identifying the Role of Multidisciplinary Journals in Scientific Research 473 subdisciplinary labels were assigned based on subject content and author af- filiation. When an article truly covered more than one subdiscipline, both were recorded rather than trying to fit the ar- ticle into one subdiscipline or the other. Disciplinary labels were assigned based on the subdisciplines, as listed in table 2. Life sciences literature was heavily rep- resented in the three multidisciplinary journals. Using Francis Narin’s example, the life sciences articles were separated into two separate disciplines.11 The areas that make up the biomedical sciences fo- cus on human health, typified as research on the treatment of diseases, medical tech- niques, and medical specialties, and are of most interest to medical practitioners. Subdisciplines, such as botany, genetics, and zoology, were assigned to the biologi- cal sciences because they are of more in- terest to traditional biologists. An additional aspect of this study was to determine the reason(s) why the sample articles were cited. The motive for citing an article can be determined by the sec- tion of the paper in which the citation oc- curs. An analysis of scientific and engineer- ing papers from the past two hundred years shows that scientists and engineers organize the information they report in a topical structure and that the same type of TABLE 2 Key to Disciplinary Assignment Discipline Composed of Following Subdisciplines Astronomy Astronomy Biological sciences Agriculture, anatomy, aquatic science, biology, biotechnology, botany, cytology, developmental biology, ecology, environ- ment, entomology, evolutionary biology, experimental biology, genetics, microbiology, molecular biology, nutrition, physiol- ogy, zoology Biomedical sciences Endocrinology, immunology, medicine, oncology, pathology, pharmacology, toxicology, psychiatry, public health Chemistry Chemistry, materials chemistry, organic chemistry, physical chemistry Earth sciences Atmospheric science, earth sciences, geology, oceanography, petrology Education Education Engineering Computer science, engineering, electrical engineering, materials science, mechanical engineering Mathematics Mathematics Multidisciplinary (sample combinations) Astronomy/physics, biology/chemistry, biology /geology, biology/mathematics, biology/ physics, chemistry/computer science, chemistry/engineering, chemistry/geology, chemistry/ physics, earth sciences/political science, geology/physics, materials science/physics, physics/engineering, sociology/ anthropology Philosophy Ethics Physics Applied physics, optics, physics Political science Political science, science policy Psychology Psychology 474 College & Research Libraries November 2003 information is almost always reported in the same section.12 The pattern of citation activity among disciplines (as opposed to within disciplines) was of special interest in assessing the purpose of multidisciplinary journals. In ninety-six cases, the subjects of a sample article and its citing article differed. These ninety-six articles were read to determine the section in which the sample article was cited. Sample articles were cited in seven sec- tions in the papers of other authors. The introduction provides justification for the research, formulation of the hypothesis, and the author ’s assumptions about the study. The information from a sample ar- ticle cited in the introduction generally was used to support the author ’s line of rea- soning. The literature review highlights prior relevant studies, thereby implying a common area of research across disci- plines. The theory section discusses the principles and consequences of a particu- lar premise and provides philosophical evidence to support it. The probable rea- son a sample article was cited in the theory section is because the citing author drew upon the theoretical principles of a differ- ent discipline. The experimental details specify materials, methods, and proce- dures used in performing the experiment. When the sample article was cited in this section, it was most likely because tech- niques and computer programs used in the original study were adopted or modified by the citing author. The results section reports the outcomes of the experiment. Often sample articles cited in this section were used to support a point in an opin- ion article. In the discussion section, the cit- ing author explains how the results of his or her study compare with the existing knowledge about a topic, including the study reported in the sample article. The conclusion summarizes the study and gives recommendations for future research. Usu- ally, sample articles were cited in this sec- tion as examples of possible applications. Using this structure as a guide, the number of occurrences in each section was counted. Results and Discussion Table 3 shows the basic framework on which this study was based, giving the number and proportion of sample articles and citing articles in each discipline. Life sciences articles made up a significant pro- portion of coverage in multidisciplinary journals, especially in PNAS. This obser- vation agrees with the study by Glanzel, Schubert, and Czerwon, who performed a subject analysis of Nature, Science, and TABLE 3 Composition of Sample and Citing Articles Discipline Sample Articles Citing Articles Number Percentage Number Percentage Astronomy 8 4% 2 <1% Biological sciences 52 28% 272 37% Biomedical sciences 29 16% 209 28% Chemistry 2 1% 17 2% Earth sciences 15 9% 51 7% Education 4 2% 1 <1% Engineering 5 3% 10 1% Mathematics 8 4% 4 <1% Multidisciplinary 34 19% 132 18% Philosophy 6 3% 1 <1% Physics 7 4% 30 4% Political science 9 5% 2 <1% Psychology 4 2% 10 1% Total 183 100% 741 100% Identifying the Role of Multidisciplinary Journals in Scientific Research 475 PNAS.13 The life sciences articles published in these journals, especially those articles concerning biomedicine, drew the most citations. One sample biomedicine article had been cited 681 times within three years after publication. The following three frequency tables give an in-depth look at the composition of the sample articles. Some disciplines were grouped in tables 4 through 6 when the number of articles in the discipline was too few to work with. The group la- beled “Other sciences” includes as- tronomy, chemistry, engineering, math- ematics, and physics. The group labeled “Social sciences” includes education, phi- losophy, political science, and psychology. Although philosophy is not traditionally considered a social science, the seven ar- ticles in this study that were classified as philosophy concerned ethical issues that are of interest to many social scientists and were included in this group. Descriptive information about the in- dividual journals also is shown in tables 4 through 6. For example, Nature does not publish research-in-progress articles but, instead, favors completed research and correspondence/opinion-type articles. PNAS publishes many reviews, whereas Science publishes none and Nature pub- lishes only a few. Although all three jour- nals published some social sciences ar- ticles, Science offered greater coverage, particularly in the form of correspon- dence/opinion-type articles. Table 7 provides a broad look at cross- citation among the diverse disciplines represented in the sample articles. Cross- citation was based on frequency counts, so it was possible to tell if the sample ar- ticles were cited by the same discipline or by other disciplines, but it was not possible to assess citation among specific disciplines. There was great variation in cross-citation activity among journals. For TABLE 4 Nature: Number of Articles per Discipline, by Type Discipline Correspondence/ Completed Reviews Research in Total Opinion Research Progress Biological sciences 5 4 3 0 12 Biomedical sciences 1 2 0 0 3 Earth sciences 3 3 1 0 7 Multidisciplinary 1 5 0 0 6 Other sciences 5 3 1 0 9 Social sciences 5 3 0 0 8 Total 20 20 5 0 45 TABLE 5 PNAS: Number of Articles per Discipline, by Type Discipline Correspondence/ Completed Reviews Research in Total Opinion Research Progress Biological sciences 8 8 7 6 29 Biomedical sciences 3 11 4 0 18 Earth sciences 0 0 0 7 7 Multidisciplinary 9 0 4 1 14 Other sciences 0 0 3 6 9 Social sciences 0 1 0 0 1 Total 20 20 18 20 78 476 College & Research Libraries November 2003 TABLE 6 Science: Number of Articles per Discipline, by Type Discipline Correspondence/ Completed Reviews Research in Total Opinion Research Progress Biological sciences 2 4 0 5 11 Biomedical sciences 2 5 0 1 8 Earth sciences 0 1 0 0 1 Multidisciplinary 3 7 0 4 14 Other sciences 0 3 0 9 12 Social sciences 13 0 0 1 14 Total 20 20 0 20 60 example, sample biomedical articles in Nature were used extensively by authors in other disciplines, whereas Science showed an opposite proportion. The same is true of citation of biological sciences in Science and PNAS. Earth sciences and so- cial sciences literature exhibited the great- est extremes in citation in Science. A. J. Meadows speculated that scien- tists who scanned multidisciplinary jour- nals were unlikely to read the articles from other subject areas.14 He said they could go directly to specific subject areas of interest because the articles in these journals are grouped by subject. However, an examination of the subject arrange- ment in these journals did not find this to be true. Nature and Science group articles in each issue by type (e.g., research re- port or editorial), so subjects are pub- lished randomly throughout the issue. PNAS contains a table of contents that groups articles by type, but within each type, the articles are arranged by subject. The assertion by Sarasvady and Pichappan that multidisciplinary journals could help scientists working in different subdisciplines to retain their disciplinary identity also was considered.15 By compar- ing the subdisciplines of sample articles with those of their citing articles, there was some evidence that this phenomenon does occur, mostly in the biomedical sciences. However, the instances of this phenom- enon in other disciplines were too few to suggest that this was generally true. Tables 8 and 9 provide an overview of the reasons why the sample articles were cited by articles from other disciplines. These data are derived from the ninety- six cases where the citing article’s disci- pline differed from that of the sample ar- ticle. Sample articles were cited more fre- quently in the introduction, literature re- TABLE 7 Cross-citation Behavior of Citing Articles, by Discipline and Journal of Cited Articles Discipline Nature PNAS Science Same Different Same Different Same Different Discipline Discipline Discipline Discipline Discipline Discipline Biological sciences 59% 41% 69% 31% 30% 70% Biomedical sciences 21% 79% 55% 45% 70% 30% Earth sciences 86% 14% 30% 70% 100% 0% Multidisciplinary 22% 78% 45% 55% 36% 64% Other sciences 49% 51% 44% 56% 59% 41% Social sciences 26% 74% 25% 75% 0% 100% Identifying the Role of Multidisciplinary Journals in Scientific Research 477 view, and discussion sections of other ar- ticles and less frequently in the results and theory sections. This was true in the most common pairings (e.g., chemistry article citing a physics article) and the least com- mon pairings (e.g., mathematics article citing a biological sciences article). Aida Mendez and Isabel Gomez’s 1989 study of citation classics offers some com- parative data.16 Their study revealed that basic research articles (articles that describe a new theory or model, or present new ex- perimental data) received about 50 percent of the total citations. This type of informa- tion is most likely to be found in the intro- duction section of a citing paper. Method- ological articles (articles that describe methods or techniques) received about 25 percent of the total citations. This type of information is most likely to be cited in the experimental details section. Reviews (ex- tensive literature surveys of cur- rent knowledge on a topic) also received about 25 percent of the total. This type of information is most likely to be found in the literature review of a citing pa- per. The results of the present study agree only partly with those of Mendez and Gomez. Although the sample articles were most likely to be cited in the introduction and literature review sections of citing papers, inferring a transfer of ideas across disciplines, there was little evidence that methods and materials were shared among disciplines. Conclusions The objective of this study was to deter- mine the role of multidisciplinary journals in scientific research and especially to as- sess the extent to which multidisciplinary journals support the transfer of informa- tion across disciplinary lines. The charac- teristics of intradisciplinary citation were not measured in this study and may be dif- ferent from citation among disciplines. The three journals chosen for study proved to be diverse, in both the scientific disciplines they include and the types of articles they publish. All three publish articles charac- terized as multidisciplinary, but most of the articles concern one discipline. The citation characteristics also revealed varying results among the disciplines cov- TABLE 8 Among Sample Articles Cited in a Different Discipline, Number Cited per Section, by Source of Sample Article Section Cited Nature PNAS Science Introduction 23 9 7 Literature review 15 3 3 Discussion 15 5 2 Experimental details 1 4 1 Conclusion 5 2 1 Results 3 1 0 Theory 4 0 0 TABLE 9 Among Sample Articles Cited in a Different Discipline, Number Cited by Section, by Type of Sample Article Section Cited Correspondence/ Completed Reviews Research in Opinion Research Progress Introduction 2 19 10 8 Literature review 1 15 2 3 Discussion 2 14 5 1 Experimental details 0 3 1 2 Conclusion 1 2 3 2 Results 2 1 1 0 Theory 1 2 1 0 478 College & Research Libraries November 2003 ered by the three journals. In some cases, articles from one discipline were cited al- most exclusively by the same or a closely related discipline. The majority of cross-dis- ciplinary sharing of information occurred in the life sciences, between biologists and biomedical researchers. In other cases, ar- ticles from one discipline were cited almost exclusively by other disciplines. None of the social science articles in the sample were cited by other social scientists. For example, a sample article on political science was cited by a biologist to illustrate a potential problem between social needs and the ge- netic engineering of crops. This study also investigated why articles from one discipline are used by other disci- plines. Sample articles were referenced most often in the introduction and litera- ture review sections of citing articles to sup- port the author’s line of reasoning or to highlight common research areas. This study supports the notion that articles in multidisciplinary journals are exposed to a broad audience, thus facilitating the trans- fer of ideas across disciplines. However, there is little evidence to demonstrate the transfer of methodology across disciplines. Variation among multidisciplinary journals is an important consideration for future research in this area because the results of this study demonstrate that the characteristics of all multidisciplinary journals cannot be generalized from the study of only one. In addition, the results of this study suggest that multidisciplin- ary journals play more than one role in scientific research. Further studies that focus on intradisciplinary citation in multidisciplinary journals are needed. Studies that gather comparative data from single-discipline journals, especially look- ing at the sections in which same-disci- pline articles are cited, also will further refine our understanding of the function of multidisciplinary journals. Notes 1. Brian C. Vickery, Scientific Communications in History (Lanham, Md.: Scarecrow Pr., 2000). 2. Carole L. Palmer, “Information Work at the Boundaries of Science,” Library Trends 45, no.2 (fall 1996): 165–69. 3. A. J. Meadows, “Literature Usage and the Passage of Time,” in Communication in Science (London: Butterworths, 1974), 126–51. 4. Diana M. Hicks and J. Sylvan Katz, “Where Is Science Going?” Science, Technology & Hu- man Values 21, no. 4 (fall 1996): 379–406. 5. S. Sarasvady and P. Pichappan, “Classification of Multidisciplinary Journals: The Study of the Journal ‘Nature’,” in Proceedings of the 8th International Conference on Scientometrics and Infometrics, Sydney, Australia (Dordrecht, Netherlands: Klawer, 2001), 639–46. 6. Meadows, “Literature Usage and the Passage of Time.” 7. T. Braun, W. Glanzel, and A. Schubert, “National Publication Patterns and Citation Impact in the Multidisciplinary Journals ‘Nature’ and ‘Science’,” Scientometrics 17, no. 1–2 (1989): 11–14. 8. K. Kaneiwa et al., “A Comparison between the Journals ‘Nature’ and ‘Science’,” Scientometrics 13, no. 3–4 (1988): 125–33. 9. Bluma C. Peritz, “The Citation Impact of Letters to the Editor: The Case of ‘Lancet’,” Scientometrics 20, no. 1 (1991): 121–29. 10. W. Glanzel, A. Schubert, and H.-J. Czerwon, “An Item-by-Item Subject Classification of Papers Published in Multidisciplinary and General Journals Using Reference Analysis,” Scientometrics 44, no. 3 (1999): 427–39. 11. Francis Narin, Evaluative Bibliometrics: The Use of Publication and Citation Analysis in the Evaluation of Scientific Activity (Springfield, Va.: National Technical Information Service, PB 252 339, 1976), 195–96. 12. Joseph E. Harmon, “A Structure of Scientific and Engineering Papers,” IEEE Transactions on Professional Communications 32, no. 3 (Sept. 1989): 132–38. 13. Glanzel, Schubert, and Czerwon, “An Item-by-Item Subject Classification of Papers Pub- lished in Multidisciplinary and General Journals Using Reference Analysis.” 14. A.J. Meadows, “Diffusion of Information across the Sciences,” Interdisciplinary Science Re- views 1, no. 3 (1976): 259–67. 15. Sarasvady and Pichappan, “Classification of Multidisciplinary Journals,” 427–39. 16. Aida Mendez and Isabel Gomez, “A Comparison of Citation Classics in Three Fields of Science,” Scientometrics 15, no. 5–6 (1989): 621–31.