It has come to our attention that Fig. I in the article by David E. Sm
ith in the December issue of BioEssays (The tails of survival curves,
BioEssays 16, 907-911, 1994) is an idealized figure and does not rest
on 'actual data' as represented in the figure itself. Needless to say,
idealized representations of data should always be clearly and unambi
guously labelled as such. The following two letters deal with the phen
omenon of 'longevity outliers' described in the article. The first is
a comment by Robin Holliday; the second is from David Smith. Ed.