Bardsley and Harris (1987) test empirically the effects of changes In deterministic wealth and random income on the measure partial risk aversion. The paper, which is otherwise well written, failed to impose the relationship between the two effects and estimated them independently. Consequently, inconsistent estimates were derived for the elasticities of the measure of partial risk aversion with respect to wealth and to income. The purpose of this comment is to derive that relationship theoretically and to point out the resulting inconsistencies in the original paper.