Derek Price's Puzzles: Numerical Metaphors for the Operation of Science

The allocation of tasks among the speakers, now writers, gave me the opportunity to focus on Price's work on the measurement and modeling of science. Because so much of the present work in scientometrics and bibliometrics is directly based on his work, I decided not to review his writings or their impact, but instead to be highly selective, perhaps to the point of being eccentric. My paper examines two of Price's most complex models, both appearing in the 1965 Science article, "Networks of Scientific Papers." The choice of these models was made on the basis of several considerations. Each speaks to important features of science and imposes order on a substantial amount of data. Each exhibits a major talent of Price in displaying data graphically. The key feature is that each model displays the data so that visual mechanisms-human perception-do the bulk of data analysis without statistical procedure, or even much need for words. (It surprised me that the Yale economist Edward Tufte had to ferret out nineteenth-century French engineers and their graphs for a book on graphical display of data when there was such a clever fellow as Price right in New Haven.) Finally, both of these models remain problematic even after nearly a quarter-century; for me, they still pose interesting questions and offer unexplored possibilities. In Science since Babylon, Price has analyzed the history of N-rays and their discoverer Blondlot as a case study of the breakdown of normal mechanisms to avoid self-deception in science. That had apparently not satisfied his curiosity about this specialty. Figure 1 shows the pattern of