Institute for Scientific Information, 1122 Spring Garden St., Philadelphia 23, Pa.

Under contract with the Pharmaceutical Manufacturer’s Association and the U. S. Patent Office, the chemical literature has been screened for new steroid compounds. Now in its third year, the Steroid Literature Coding Project is a truly cooperative venture of industry and government. The Project has had some unique experiences in reading, extracting, diagramming, coding, punching, and microfilming over 7500 steroid compounds reported in the literature from 1958 to the present. The existing method of searching will be explained, as well as a new method of filing steroids by clerical procedures. The new method reduces the number of punched cards from 4 to 1 and eliminates difficulties in compositing, i.e. superimposed punching, used by the Patent Office for the steroid patent deck. By concrete example it will be shown why it is not possible in the immediate future to completely mechanize the encoding process, mainly due to the lack of standardized journal formats, and the brevity of scientific communications rather than variations in nomenclature. The latter could be overcome by rather sophisticated computer programming. Experiments along these lines will be discussed.