SCHOLARS conducting research on diversification have generally not had access to accurate data on the actual vector of outputs, by industry, of an enterprise.1 Instead, researchers have had to approximate these vectors from commercially available services that either simply list plant outputs2 or present the rank ordering of the outputs of a plant.3 In the former instance an equal weight is given to each output (the rectangular approximation)4 while for the latter, the weights, from the lowest to the highest ranked output, are assigned in accordance with the geometric series I, 2, 4, 8, I6 . . . (the geometric approximation) .5 A third approximation, commonly used by census authorities,6 is to assign the whole of the output of the plant to the highest ranked output (the primary approximation). No analysis has been conducted to test the accuracy and implications of the application of these three approximations, at either the plant or enterprise level.7 This note attempts to fill that void.