fter rising steeply through most of the 1990s, the number of papers with multiple authors—particularly with 50 or more coauthors—appears to have leveled off in recent years, according to a new Science Watch analysis.

This report follows up on a previous study that tracked multiauthor papers published between 1981 and 1994 (see Science Watch, 6[4]: 1-2, April 1995). For this update, Science Watch tracked multiauthor papers published in Thomson ISI-indexed journals from 1990 through 2003. The graph on the right shows the number of papers published in each of those years with, respectively, more than 50, 100, 200, and 500 authors. (The numbers are cumulative in that papers with more than 100, 200, and 500 authors are included in the statistics for those papers with more than 50.)

Immediately noticeable on the graph, as it was in the previous study, is a dip in the number of multiauthor papers in 1991 (when the 50+ contingent, for example, fell from 190 papers in 1990 to 81). This decline was never entirely explained, although Science Watch theorized that certain installations devoted to high-energy physics—a highly collaborative enterprise that is a principal source of multiauthor papers—may have been relatively inactive that year for equipment refitting or other reasons.

In any event, as the graph shows, the number of multiauthor papers ascended sharply after 1991, with three of the four lines spiking noticeably around 1997-98. Papers with 50 or more authors, in particular, reached a zenith of 602 in 1997, while those with 100+ authors rose to 316 in 1997, and the 200+ cohort reached 227 in 1998. Another spike follows around 2000, with, for example, the more-than-100-author group reaching its highest count of 324 papers. After 2000, the lines appear to level off, although the figure for 500+ authors, after holding at 22 papers for both 2001 and 2002, shoots up to a high of 40 in 2003.

The graph on the left provides a breakdown of multiauthor papers (those listing 100 or more authors) for each of the last 10 years according to two major disciplinary groupings: physical sciences and biomedicine. The physical-science papers climb in a fairly regular series of steps towards a high of 279 in 2002, falling off to 226 the next year. (This combined "physical sciences" designation includes not only the main Physics category but Space Science, as well as a handful of experimental-physics papers appearing in journals indexed in the category of Engineering.) By contrast, a similar graph in the 1995 Science Watch study showed that multiauthor physics papers barely exceeded 30 per year between 1985 and 1989, before increasing sharply in 1990. By 1994, as the present graph shows, such papers numbered upwards of 150, and increased significantly thereafter.

The multiauthor papers graphed under the heading of "Biomedicine," meanwhile, represent primarily Clinical Medicine and Molecular Biology/Genetics (with comparatively small representation from such fields as Neurosciences and Pharmacology). As with the graph on Page 1, multiauthor papers in biomedicine seemed to peak around 1997, subsequently falling off in number. The year 1997 did happen to feature a number of large-scale trials in cardiovascular medicine; in fact, the 1997 paper identified by Science Watch as having the greatest number of authors of any report published that year—with more than 650 authors listed—was a trial reported in The Lancet of warfarin with aspirin after myocardial infarction (V. Fuster, et al., Lancet, 350[9075]: 389-96, 1997). Incidentally, the all-time champion for multiple authors in the period covered by this survey proved to be another Lancet paper, from 2000, reporting a study of ramipril in people with diabetes (H.C. Gerstein, et al., Lancet, 355[9200]: 253-9, 2000); this paper lists more than 900 investigators.

In 1990, the mean number of authors per Thomson ISI-indexed paper was 2.6; by 2003, it had climbed to 3.6. Certainly, the single-author paper is an endangered species: in 1990, 38% of Thomson ISI-indexed papers listed a single author, a figure that had slipped to just over 25% of papers in 2003. Nevertheless, the graphs here suggest that the trend toward multiauthor papers, while not exactly reversing, may be moderating, or, at least, stabilizing.