MONO91 Indexes Compiled by Machine chapter Mary Elizabeth Stevens National Bureau of Standards An author not cited frequently enough or not cited witim a given time period will not appear in the citation index. Doyle points out that there are "many kinds of documents we would like to retrieve where it is not customary to provide citations at al]". 111n the bibliographic coupling method, both those papers which make no references to any other paper and those papers which do not share at least one reference with some other paper in the system are automatically excluded. 2/ Other disadvantages of the citation indexing technique relate to difficulties of the lack of standard practices in the citing of references and to problems of recognizing whether one citation is or is not equivalent to another. These are, of course, related to the normal difficulties arising from non-standardized formats and practices in descriptive cataloging, in use of journal abbreviations, in transliterations of foreign language titles and names, and the like, but they are now aggravated by the present prospects for direct machine processing. As Lipetz points out: !tAuthor1s names may be cited in somewhat different ways, and there is no simple mechanical procedure for bringing together the different versions. For example, an author's name may be cited both with and without initials; it would take a comparison of the additional information on the cited reference to establish that these authors are the same. Even more difficult are the problems of mechanically determining that a misspelling has occurred." Both the disadvantages of incomplete and disproportionate coverage and of failures to equate equivalent citations are quite readily obvious to the user of a citation index if he is reasonably familiar with the subject field or document set that is covered. Thus, the use of the citation index as the exclusive tool for literature search is subj ect to defects of both oversight and 1over-cite' which are cumulative and which are often easily recognizable. Atherton and Yovich emphasize that: "Knowledge of these weaknesses tends to prevent anyone from trusting the system's ability to retrieve the pertinent literature." 4/ In general, however, the citation index has not been proposed as an exclusive means for literature search and retrieval, but rather as one of a set of tools or as a supplement to other indexes. In this connection, it is of interest to note that a manual technique of literature search tested at The Therinophysical Properties Research Center 1/ 2/ 3/ 4/ 5' Doyle, 1963 [162], p. 8. See Atherton and Yovich, 1962 [26], p.39; Marthaler, 1963 [399], p. 23. Lipetz, 1962 [364], p. 262. Atherton and Yovich, 1962 [26], p.39. See, for example, Tukey 1962 [611], p.10: "The citation index, in its retrieval and pursuit uses, is not something to be used alone. Rather, it is the tool whose presence makes all the other tools more effective." 33