28 THE BIOLOGICAL LITERATURE 



field were not limited to two. Therefore the card could store 2^*^ (about 

 10^^) bits of information. 



The type of information stored can be described briefly. Authors' 

 names are given a code designation, a number from 00 to 99, by divid- 

 ing the alphabet into 100 more-or-less equal categories and using tw^o 

 fields for the name of the first author on the paper. A second pair of fields 

 gives the second or best-known author. The year of publication requires 

 one pair of fields plus one extra hole for a few papers before 1900. Jour- 

 nals have been coded by a modified alphabetical scheme. If a straight 

 numerical alphabetical code were used, some categories such as BIO 

 and JOU would be overloaded; for this reason certain journals are given 

 a separate numerical designation. 



Since most papers can be classified in more than one way, subject mat- 

 ter coding uses two sets of three fields each. A decimal system similar 

 to that used in the library has been adopted. A paper on "the fluorescence 

 of pigments" is likely to be coded under "fluorescence" in one set of 

 three fields and under "pigments" in the other. The subject matter file 

 is the most difficult to work out. It is necessary to decide which are the 

 broad general categories and then to decide what are the reasonable 

 subdivisions of each. The subject matter file must be reasonably detailed 

 and yet must be flexible enough to be modified when your ideas of 

 what is important change, as they will in the future. 



In addition to author, subject, journal, and date, several other bits 

 of information have been included as separate punches. One hole is 

 punched if the card concerns a book or monograph, another if a reprint 

 of the paper is on hand. A set of four punches indicates the language: 

 no punch for English, first hole for German, second for French, third 

 for Russian, and fourth for any other foreign language. 



The preparation of the cards requires a great deal of time, of course, 

 but then so does the use of any other card system. Once the cards have 

 been prepared, there is no need to file them in any particular order. 

 From among, say, two thousand cards, all the papers by one author 

 can be found with four or five passes of the needle, followed by hand 

 sorting of an average of twenty cards in that alphabetical category. 



I reiterate, this system can work, but everyone must develop his own 

 scheme of coding. Sorting by computer, as in the IBM system, would 

 be even easier but a good deal more expensive. Sorting by hand from 

 a set of cards arranged in alphabetical order might be faster, but the file 

 must be kept in alphabetical order and cards must be duplicated for 

 cross-reference. 



