The Science Citation Index and ISI’s Journal Citation Reports:
Their Implications for Journal Editors

Eugene Garfield, Ph.D.
Institute for Scientific Information
325 Chestnut Street
Philadelphia, Pennsylvania 19106, U.S.A.

Presented at
Third General Assembly of the
European Association of Editors of Biological Periodicals
Paris, 10—12 May 1976.




The revised program of this meeting gave the title of my presentation today simply as, "Science Citation Index". Those of you who have had the time to scan the abstracts which accompanied the revised program will have noted an expansion. I should like to discuss with you today not only the Science Citation Index, but, as my full title indicates, "The Science Citation Index and ISI’s Journal Citation Reports, and Their Implications for Journal Editors."

If will be convenient if you will allow me to refer hereafter to the Science Citation Index as the SCI, and to the Journal Citation Reports as the JCR. That first set of initials --SCI-- is by this time, I think, well known to the greater part of both library and scientific communities around the world. I hope that in a short time the second set of initials --JCR-- will be equally familiar to both, and, more important perhaps, equally familiar to that special segment of the international scientific community of which you here today are members.


I should like at this point to assume that you are all familiar with the SCI. It would give me immense personal and professional satisfaction if I felt I could do that with full confidence. For, as some of you may know, it was I who more than twenty years ago began study of the use of citation indexing as a tool for information retrieval in the natural sciences. My early interest in problems of information retrieval was, fortunately, guided and encouraged by some eminent members of the scientific community, prominent among them one or two who were or had been editors of scientific journals.

The concept of citation indexing was widely regarded in those early days as too ‘different’, too ‘unusual’ an approach. Great numbers of the very scientists who insisted upon the useful necessity of complete and accurate citation in scientific reporting found it difficult to see in citations not only the most reliable written record of ‘information transfer’, but also the most reliable source of indexing technique. I do not deny that it required more time, more talk, more travel than I thought it should to convince them of the value of their citation practice. All that is past now, but I suppose the effort must have left its mark one me. The SCI has been published annually since 1961. Yet I confess I was momentarily and quite irrationally startled a year or two ago to find the SCI classified in some library’s brochure, along with Chemical Abstracts and Biological Abstracts and others as ‘one of the conventional reference tools.’


Even after fifteen years of annual publication, I cannot think of the SCI as a ‘conventional’ reference tool. Some people may excuse my inability to do so on the grounds that I am its originator and publisher, but they would be wrong. I also happen to fancy myself as one of the SCI’s most experienced users, and, as a working scientist myself, I know its capabilities in the little research that time allows me to do on the development and status of information technology and ——more important for this audience—— on the history and sociology of science. The latter, of course, includes that now much discussed and debated phenomenon, journal publication.

Rather than indulge myself to discourse on the capabilities of the SCI, I shall explain briefly how it is compiled and describe the three sections of each year’s annual edition.

The SCI covers about 2500 journals. Excepting genuinely ephemeral material such as news items related to a professional society’s or trade association’s corporate business, every item in all of the journals is processed for indexing original articles, technical communications, letters, correction notes, etc. AU authors, their affiliations and addresses, full bibliographic data (article title, journal title, volume, issue pagination, publication year), and all citations in the references of the articles are processed for indexing ——or, to use computer jargon, provide the input for the sections of the SCI.

There are three sections, the Citation Index, the Source Index, and the Permuterm Subject Index.

The first, the Citation Index, lists by first author, year of publication, and journal title and location (volume and pagination) every published and unpublished work cited in the references of articles published by the SCI’s 2500 source journals during the year. Each of these citations becomes an index entry. Under it are listed the bibliographical data needed to identify any paper published during the year that cited the earlier work.

The bibliographical data on the year’s citing articles is given in abbreviated form in the Citation Index. In the second section of the SCI, the user finds the complete bibliographical date. The second section of the SCI is the Source Index. It is an alphabetic listing by first author of all items published during the year by the SCI's 2500 source journals. Each entry provides the names of first and authors, the full title of the article (or an English translation of titles in other languages with a code for the original language), journal title, volume, issue, pagination, number of references, and nature of the item (original article, letter, book review, abstract, correction, etc.)

The third section of the SCI is the Permuterm Subject Index. Briefly, it uses words from the titles of articles listed in the Source Index as subject headings referring to the first author in the Source Index entries. The Permuterm Subject Index is thus a so—called ‘natural language’ index. It uses the current language, that is the current terminology, of science and technology as its subject headings, not a terminology that has been chopped or stretched to the arbitrary demands of some Procrustean hierarchy, classification, category listing, etc. Furthermore, each of the main—entry title—words is paired with every other word from the same title. The other words appear as subentries. The user matches up main—entry and subentry words for quick reference through first author’s name to appropriate entries in the Source Index.


These brief descriptions of the sections of the SCI hardly do more than suggest the usefulness of the SCI and its many unique capabilities. To those of you who do not know the SCI and who have not used it, I can only urge you-—not as its publisher but as a fellow scientist and, in our different ways, as a fellow communications specialist in science-—to let your librarian, or an ISI representative if necessary, introduce you to this index.

I should like at this point to make myself the fuller introduction which the program and time denies me. Let me say only that the SCI fulfills, as no other index does, the ‘retrieval’ requirements of today’s interdisciplinary and multidisciplinary research. Coverage of the SCI spans all of science and technology, as well as sociology and the behavioral sciences. Its use demands knowledge of no particular language, not even the international English of science —though one of course must be able to read in some language that uses the Roman alphabet. One need not even be familiar with the technical vocabulary of any area of science in any language to use the SCI in searching the literature of that area —-on as broad or as specific as may be necessary. The SCI empowers the user in this rather unique fashion because of two things: first, as I have mentioned, the breadth of its coverage across the whole of science, and, second, because of its organization of the literature upon the structure of citations. For citations are the true and immutable language and terminology of scientific bibliography. They transcend time, language, all lesser terminologies, and —-most important to the effectiveness of an SCI search-— even the author’s (and certainly any indexer’s) statement or evaluation of the significance of his work.

Because I cannot discuss this further now, I encourage your requests from me here or later of any of the plentiful literature on the subject, and perhaps questions during informal meetings during this assembly.

Before going on, however, I should like to stress the SCI's timeliness, and the business of dates. The SCI is a calendar—year index. Its 1975 Annual, for example, includes appropriate entries for all items published in all 1975 issues of the journals it covers. A date on an SCI annual means something. Because I have so often stressed the timeliness of the SCI, a librarian colleague has recently chided because our 1975 SCI was distributed a few days later this year than another index of some size and repute. I won’t mention its name. I will, however, report to you that that 1975 index contains nothing for material published during the last three months of that year.


For two centuries or more, it has been customary to include in a scientific publication a list of references to earlier works. References are, of course, given for a variety of reasons: to identify the groundwork, so to speak; to indicate sources of fuller explanation; to establish authority for a point of view; to validate or to refute another viewpoint; and so on. The reasons for referencing are various, but whatever reason may apply in a particular case, one thing remains true in all cases -—there is always a subject relationship, a matter of common substance or some aspect of common substance, between the citing work and the work that it cites.

Thus, if one should want to identify work proceeding from, or related to the research reported by J. D. Watson and his colleagues in an article beginning on page 737 of volume 171 of the journal Nature in 1953, it is an absolutely simple job to find it all. The article in question is of course that which announced the proposed structure of the double helix. All of the relevant literature in this case can be found by looking up the citation of that article, because the articles sought will have cited Watson’s —or to go a step further in time— will have cited articles that cited Watson’s. The first term consulted is the citation formula itself: WATSON JD, 1953 NATURE 171 737. One particular capability of the citation index should be mentioned here. The SCI identifies, as no other index can, applications of research that have crossed one or more of those arbitrary disciplinary boundaries that are becoming yearly, it seems to me, more ideal than real. A not especially unusual example is the citing of various articles by Einstein in the references of articles published in journals devoted to dairy science and obstetrics.