I have two points in response to the authors. First, they were correct in their statement about disjoint databases. R. Needham misunderstood their paper and offers his apologies. My second point concerns clustering for recall or precision. In general, document clustering has been either intended or found to be most effective as a precision device. There is no solid evidence of its being a valuable recall device. Thus, while the authors suggest that clustering could serve recall, they do not demonstrate its utility for this, so my comments on retrieval performance stand.