Differences between tools' original and discarded sizes bear on classificat
ion, curation and other interpretive issues in Palaeolithic archaeology. Re
cent experiments demonstrated significant relationships between platform si
ze and original flake size but also generated controversy about the relativ
e importance of empirical and theoretical platform variables in determining
flake size. Depending on the nature and strength of the relationships, ori
ginal size might be calculated from platform variables, often retained in d
iscarded tools. We examine the platform-flake relationship in Pelcin's cont
rolled data and two replicated assemblages. As Pelcin suggested, theoretica
l platform variables predict flake size only with stringent assumptions. Em
pirically, log-size regresses consistently upon log-platform size in hard-h
ammer but not in soft-hammer flakes. A universal relationship between varia
bles exists in the assemblages, but only in hard-hammer flakes which, in an
y case, were more likely to be used as tools. Copyright 2000 Academic Press