A New Look at Reference Scattering

It was first observed by Bradford that, for a large collection of journal references on a given subject, most of the articles are derived from a small proportion of the total titles. Bradford listed the journals concerned in order of decreasing productivity and, by plotting the logarithms of the cumulative totals of titles against the cumulative totals of relevant articles produced, he obtained a straight line. (Similar results have been obtained by many later workers. The pattern is illustrated by Table 1 which shows the distribution of references among journal titles obtained by the author during a study of literature usage in the petroleum industry.) On the basis of these results Bradford then formulated a simple mathematical model to describe reference scattering. Vickery later pointed out that this ‘law of scattering’ predicted not a straight line but a curve. Kendall has now provided a more refined statistical explanation of the straight line observed by Bradford.