Small sample sizes in the study of ontogenetic allometry; implications for palaeobiology