D. J. FINNEY observes1, in effect, that the data of a four-point assay in which the apparent response curves are not parallel do not allow us to distinguish between two conceivable causes of non-parallelism, namely, (1) non-linearity of the fundamental response curve, and (2) real dissimilarity between the response curves of standard and test material; and that while the first does not invalidate the assay2,3, the second does. To this we may add that the smaller the difference between the overall mean responses to standard and test material the more reason there is to attribute non-parallelism to the second cause. The whole matter is particularly relevant to the vitamin A assay, which employs a non-vitamin A standard and in which there are therefore no "strong a priori reasons for believing that the standard and test preparations have response curves of identical form"1. In this laboratory, statistical check is kept on the difference between the constants of slope for standard and test material (which is equivalent to a check on the value of T, as defined by Finney). No significant deviation has, in fact, been found, but it cannot be assumed that this would apply to other regimes.