“…The study of allometric relationships by ANCOVA has been severely constrained, however, by the logarithmic transformation, because such analyses require that the distribution for Y and X in each group conforms to that of a two-parameter power equation with lognormal, heteroscedastic error on the original, arithmetic scale (Packard, 2017c). Many distributions follow the paths of threeparameter equations instead of two-parameter equations (Packard, 2015(Packard, , 2016(Packard, , 2017a, and random error may be normal and homoscedastic or normal and heteroscedastic instead of lognormal and heteroscedastic (Packard, 2016). Such distributions are not amenable to analysis by ANCOVA on transformations.…”