Proceedings of ISSI 2005, pages 398–405. Edited by P. Ingwersen and B. Larsen. The Rhythm of Science, the Rhythm of SCIENCE 1 Liming Liang * ** , Ronald Rousseau ** *** and Fei Shi **** * pllm@public.xxptt.ha.cn Institute for Science Technology and Society, Henan Normal University, Xinxiang, 453007 (China) ** University of Antwerp (UA), IBW, Universiteitsplein1, 2600, Wilrijk (Belgium) *** ronald.rousseau@khbo.be KHBO (Association K.U.Leuven), IWT, Zeedijk 101, 8400 Oostende (Belgium) **** Thomson Scientific China Office, No 2 Kexueyuan South Road, Beijing 100080 (China) Abstract The rhythm of science may be compared to the rhythm of music. The R and T indicators studied in this article are complex indicators, trying to reflect part of this rhythm. The R indicator interweaves publication and citation data over a long period. T constructs an input-output relationship in knowledge production. In this way the R- and T-sequences can be used to describe the evolutionary rhythm of science considered from two different aspects. As an example the R and T sequences of the journal Science from 1945 on are calculated. Introduction What is science? How does science evolve? These questions refer to two of the most interesting problems in the philosophy and sociology of science. In general, people consider science to consist of the activities of knowledge production as well as the system of knowledge itself (Kuhn, 1962). Scientific documents, in printed or electronic form, are often the final products of knowledge production. They carry scientific information not only to contemporaries, but also to the next generations. Therefore, analyzing publication and citation data of scientific documents is a way of approaching the questions stated in the first sentences. Publication and citation data, being two basic scientometric indicators, have been playing an essential role in the study of science. As such, they reflect, to a large extent, the process of scientific evolution. Like the evolution of living things, the evolution of science has its own rhythm. In the history of western science, we see between the two prosperous periods of the ancient Greece-Rome science and the science of the Renaissance, the less frugal period of the Middle Ages. Then, more recently for four centuries we experienced a series of peaceful interludes punctuated by violent intellectual revolutions (Kuhn, 1962). Focusing on modern science, we can perceive its pulse by studying scientific documents and the time series of their publications and citations. For example, the publication of thousands of articles on superconductivity and the large scale of citing these articles represent a climax in the science of the end of the 1980’s and beginning 1990’s. Apart from global and average publication and citation indicators, relative indicators such as the (modified) impact factors (hereafter IF for short) form another class of important scientometric indicators. A time series of IFs can roughly reflect the evolutionary rhythm of a scientific field, country, or journal during a certain period. The traditional IF, sometimes referred to as the Garfield-Sher IF, has been generalized by information scientists after it was put forward in 1963 (Garfield & Sher, 1963). Whatever version is used, its calculation is always based on observed values of publications and citations. For this reason we call these impact factors observation-based IFs. In this paper we suggest another type of IF, namely expectation-based IFs (see further for their definition). Furthermore, we define and study ratios of observation-based and expectation-based IFs, leading to relative IFs. This relative indicator series has 1 The work presented in this paper was supported by the National Natural Science Foundation of China by grant no. 70373055.