Using search engines available in our library, we reported surprisingly in-adequate recovery in test searches for lists of known references. Atkinson and Cunningham suggest that better results could have been obtained had we used BIOSIS Previews: perhaps, but BIOSIS Previews indexes more than 5000 serials, whereas CSA's Biological Sciences indexes more than 6000 serials. Regardless, the problem was not missed journals but inadequate coverage of the contents of journals. Atkinson and Cunningham also recommend Wildlife & Fisheries Worldwide, an engine that we cannot find, even in their own library; perhaps the reference is to Fish and Fisheries Worldwide (which contains a subset of ASFA, one of the engines we used). A third suggested alternative, Selected Water Resources Abstracts, is a subset of CSA, which we used in our searches.

We do not expect that use of the suggested alternatives would improve results; there was, in fact a very excellent match between the subject matter of the work in aquatic sciences in our list of known publications and the subjects covered by the search engines we used. Atkinson and Cunningham also point out that ASFA and Biological Sciences began work in 1971 and 1982; this does not explain why the inadequacy of the record was as prominent post-1970, nor why some papers published before 1970 appeared in our search. Atkinson and Cunningham also suggest that we should have added print indexes to our search, but our aim was not in a complete search, but rather on what normal working scientists would find in computer searches.

We reported a remarkable level of inconsistency, and below-par performance in tools that we all depend on every day. Atkinson and Cunningham try to explain the remarkable inconsistencies among databases as the result of different levels of full or selective indexing applied to different journals, and that commercial and other sorts of search engines differ in construction and serial selection. These may be the kinds of procedures that need review.

We have welcomed the attention that professional librarians have given to our article, as evident in many e-mails and in Atkinson and Cunningham's letter. We would like, however, that our results reach those who use citation indexes, searches, and related statistics to make decisions that matter to people's careers and lives. Such judgments need to be made, considering the current inadequacy of these ever more important tools.