provide a criteria for rejecting the importance of substitution in ERP calculations. In addition, it is no less restrictive to assume a priori that all industries have the same value for a, whether it be 0.5, 2.0 or the traditional value of zero. It has been pointed out elsewhere that the realistic case to consider is what happens to the rankings when different industries have different values for a'.3 A meaningful approach would involve answering the question of what is the highest and lowest ranking an industry could obtain under any combination of different hypothetical values for o-. The data for the 135 industries in the investigation indicate that applying these criteria creates scope for the rank correlations to be somewhat less than the observed figures of 0.99, but the author's contention of a high correlation still stands. Nevertheless this result, while being interesting and useful, can however be misleading if applied as a general case. There is in fact considerable scope for variation in industry rankings when there is a wide disparity in the magnitude of the calculated ERP across the industries. This is the experience of some countries4 and it is also the case when, for a variety of practical purposes, industries are classified in smaller groups according to the intensity of their protection. For the group of highly protected industries, for example, there is traditionally a wide spread in the magnitude of the ERP and therefore considerable scope for variation in industry rankings. The result being that in the study under discussion for the 10 most highly protected industries there are 9 industries that could fill the top 4 positions and 7 industries that could fill the bottom 2 positions in the industry rankings.5