This work analyzes citations in the scientific literature published in the 104 years between 1900 and 2004. The authors were interested in determining the distribution of the age of work cited, as indicators of the state of science as practiced. An analysis of the age reveals some interesting facts and provocative interpretations of the state of science. The general conjecture is that scientific papers normally have a rather brief life. They are cited more heavily when they are recent. However, they become obsolete and ignored by the scientific community as research problems evolve and their relevance diminishes. Furthermore, the rate of obsolescence increases with the vigor of the research area; scientific work becomes more quickly outdated in the hotter areas. Using a century of data, the authors found important shifts in the distribution of citations in the natural sciences and engineering as one set, and in the medical literature in another. They were also able to work with a complete subset in astronomy and astrophysics, and another in nuclear and particle physics.

The authors determined both the median and mean ages of work cited. Even though these values differed, the trends were the same: during the two world wars, the distribution of ages of work cited tended toward age. The difficulty of scientific communication during wartime limited resources to those already at hand. Between 1945 and 1975, there was a trend toward more recent publications, with an exponential growth rate in publication. This was an era of vigorous scientific effort, fueled by Cold War science, the explosion of college and university enrollments from the G.I. Bill and from baby boomers going to school, and economic growth in the entire northern hemisphere. From 1975, there has been a gradual trend toward citations of greater age. The authors interpret this as a manifestation of Kuhn’s “normal science” environment of consolidation after a scientific revolution.

There may also be additional technological factors involved. Many older scientific papers are more accessible now through electronic databases, as older print journals are pulled off library shelves and stored. Journals not subscribed to in the past may be accessed through electronic means, extending search capabilities. Both of these factors may tend to increase the age of citations. The contrary technological trend is the existence of open-source journals and pre-print servers, like arXiv, that are venues for quick publication in some areas. This was the authors’ interpretation of the shift to younger references in astronomy and particle physics.

This paper had an interesting comment about the still exponential growth rate of papers in China, while other countries are experiencing only a linear growth rate. The geographical distribution of publication rates would be an interesting, although ambitious, study. China is investing heavily in science and engineering, and building research universities from scratch. As other countries build up their scientific capacity, there should be a shift in research production and vigor to these countries.