Sir

Publications not indexed in listings such as the ISI Web of Science are, these days, considered of questionable merit. In more collegial times, research performance not adequately represented by application of such standardized metrics could be evaluated fairly — for example, with allowances for lack of coverage of some disciplines, for citation behaviour in different disciplines, and for the existence of prestigious alternative forums. Your News Feature “The counting house” (Nature 415, 726–729; 2002) drew attention to some problems with bibliometric databases and their uses, and many of the 64 citations of this News Feature listed since then in the Web of Science provide further analyses of problems.

ISI has recently delisted a number of publications from the Web of Science without informing the affected publishers or editors, or publishing a full list of the excisions. The motivation seems to have been to focus the Web of Science on journals and to move conference proceedings to another, little-known product, ISI Proceedings — notwithstanding the fact that many journals have special issues containing conference proceedings.

Proceedings of the Combustion Institute, an important archive in the multidisciplinary field of combustion dating back to 1928, is one of the affected publications. Because its peer-reviewed papers are presented at the biennial International Symposium on Combustion, they will no longer be listed in the Web of Science. According to ISI, the decision to exclude this publication “was not based on an evaluation of its importance to the community of scholars it serves”.

This experience adds a new dimension to problems with excessive reliance on citation analyses. The Web of Science database itself is subject to unaccountable adjustments without scientific justification or regard to scientific importance.