Forms for Literature Citations

Science, 120(3129),

p. 1038-41, December 1954.


It is not entirely correct to assume that the "minimum information necessary in principle to specify a certain article" is the page, journal name, and year. Patents have long been identified by serial numbers. In theory, published articles could be so identified.

Another approach—different from Bishop’s or Reid’s—is to identify the individual article by a Serial number of two parts: the first part identifying the journal, similar to the identifying numbers used in the World List of Scientific Periodicals or the Bibliographie der Fremdsprachigen Zeitschriften Literatur; the second part a serial number as exemplified by those used in the Proc. Soc. Exptl. Biol. Med. or the reprint order number in the J. Chem. Soc. (London). The serial number has its obvious disadvantages as would an approach based on the use of secondary publication references, such as the volume and column number in Chemical Abstracts or the numbers in publications like the Current List of Medical Literature and the Bulletin Analytique. The disadvantages increase as one gets further away from the original complete citation. Communication engineers face similar coding problems. Both brevity and redundancy have shortcomings.

It is necessary to determine the function of any proposed citation system. If a unique identification tag is all we require, then the patent system is enough. If the tag is to contain journal identification, then the two-part serial number is sufficient and far more desirable for punched-card applications; where the sorting of numbers is less time consuming than the sorting of letters. In "The preparation of printed indexes by automatic punched-card equipment—a manual of procedures" (Medical Indexing Project Report; Johns Hopkins Univ., Mar. 1953) I have demonstrated the feasibility of using such numerical identification tags for the compilation of scientific indexes.

If the tag is to have mnemonic, as well as classificatory features, then the difficulties increase. It is questionable whether the space saved by reducing the already abbreviated citations compensates for the time lost in trying to identify the citations in published articles. The retention of the journal title in the accepted abbreviated form will probably continue for many years, since most libraries must ultimately deal with the journal title, volume, page, and even year for. verification. However, the addition to the established citation form of a unique tag such as a serial number would not be an excessively heavy burden on an author or editor. Such tags could help in speeding up communication between scientists and help particularly in facilitating the compilation of many types of compendiums, where uniqueness and brevity are compulsory.

Documentation Consultant
Smith, Kline and French Laboratories
Philadelphia 1, Pennsylvania
13 September 1954.