Simpson's Diversity and entropy are two good cases in point. Both indices were conceived outside the field of linguistics but they both relate satisfactorily, on a conceptual basis, tolr which can then be defined as an absolute entity. Moreover, the use of whole frequency distributions will be regarded by some as a guarantee of quality. Yet, these indispensable virtues are, in both cases, blemished by some shortcomings which rule them out for a linguist. Instead of borrowing directly from other sciences, it might be useful to collaborate with mathematicians in order to elaborate an index oflr which would combine linguistic consistency and mathematical reliability. This would have the added advantage of making inevitable a theoretical reflexion on the very concept oflr, especially as regards its relations to the structure of frequency distributions. In this case it might be necessary to question some of our thinking concerning the vision oflr in relational terms. From an epistemological point of view, an interesting situation might arise where, in a given field (in this instance, linguistics), progress in theoretical thinking is made possible not by the automatic incorporation of elements from other fields, but by the critical (and hopefully positive) study of such tools.