ÖйúAƬ

Three observations on your list of "Highly cited papers in classical studies since 2000" (Research, 23 July). First, the ten papers attracted on average 1.5 citations a year. Such small numbers are not useful: they are too sensitive to random factors, including data errors.

Second, the data errors are gross. Web of Science records 14 citations for "Willink" (not 11: the discrepancy is due to the inconsistent abbreviation of journal titles), of which four are phantoms (papers on nutrition that cite W.C. Willett, not C.W. Willink); the rest are self-citations. "Brosman" has 13 recorded citations: 10 are phantoms (papers on materials science that cite W. Brostow, not P.W. Brosman).

Third, half of the papers are on Greek drama. That's unsurprising: publications on intensively researched topics accumulate citations more quickly because they have more opportunities to be cited. It follows that citation counts for papers on different topics within the field of classical studies are not comparable. So even large citation counts based on reliable data would not provide a valid instrument for comparative evaluation.

Malcolm Heath, Professor of Greek, University of Leeds.