Species that reach the end of a life-history stage (transform) at rela
tively fixed sizes may often grow nonasymptotically before transformin
g. Many species of squid and at least some larval frogs, fish, and ins
ects appear to follow this pattern. When data on body size at a range
of ages are available for such taxa, they are often described well by
exponential curves or by power curves that are concave upward. When su
ch data are transformed to mean sizes for ages or age classes, they ar
e likely to fit asymptotic growth models such as the logistic and Gomp
ertz curves. These curves are good descriptions of the behavior of the
population mean but poor descriptions of the pattern followed by any
individual in the population. Analyzing and presenting data on size at
age using mean sizes can thus lead to incorrect interpretations of gr
owth patterns and should be avoided.