The bibliographical information content of research papers.

The occurrence, form, and content of the elements of research papers used in bibliographical control is examined in the whole range of disciplines---natural sciences, social sciences and humanities. Instructions given by primary journals to authors about these elements are considered in the light of requirements for bibliographic control. 1. The information content of titles is assessed firstly by counting the number of keywords in them, and variations with discipline, time and language are reported. A more refined measure depending on the probabilities of occurrence of title words is then used for samples taken from journals in chemistry, history and philosophy. 2. Suggestions are made for improving the presentation of author details to facilitate bibliographic control. 3. The information in abstracts of experimental papers is analysed into the four divisions traditionally used for the text: introduction, experimental/method, results, and discussion, and the proportion of each type is compared with that in the text. A wider range of abstracts are also analysed into informative and indicative content. Literature on the uses of abstracts is reviewed in a consideration of their purpose, and arguments are presented for indicative abstracts. 4. A comparison of authors' keywords with terms assigned by indexing services shows that such keywords are not usually adequate or accurate enough for indexing purposes. Instructions to authors to use a controlled vocabulary, e.g. Medical Subject Headings, did not seem to be observed very closely. 5. The average number of references per paper is reported for various subjects, broken down for experimental papers into the sections used in studying abstracts. The value of including the titles of cited papers in references is stressed. As the use of on-line services takes over from use of printed indexes, reliance is likely to be placed entirely on authors (and editors) for index terms. Titles and abstracts seem to be a more reliable source than separate lists of keywords.