A s INCONCEIVABLE as a "boneless vertebrate" is a "nonquantitative geography." Estimates and measurements of terrestrial magnitudes-distances, directions, dimensions-have always constituted the framework of geographical knowledge. There are, however, degrees and standards of "quantitivity." Those attained by Eratosthenes and Ptolemy were comparatively high; those of the geographical writings of the Middle Ages in Europe were much lower; since the Middle Ages, Western geography, alotng with Western science in general, has become increasingly quantitative until by now geographers themselves are sometimes perplexed by the mathematical discussions in their own professional journals. Some years ago, in preparing for a conference on a proposed atlas, I had occasion to deal with the relation of quantitative data to map symbols.' Recently, in an unpublished historical study, I was led to consider some characteristics of the quantities on which maps and other geographical expositions have been based at different times in the past. For both purposes I made use of a technique of classification and cross classification that is somewhat akin to the techniques widely employed today in the framing of questionnaires and in analysis of the information collected through them. Although this technique has, no doubt, been used more or less unconsciously since the dawn of rational thinking, I do not know of its having been applied in geographical studies in the manner I have in mind; and since it offers possibilities that seem entertaining and may be useful, I propose to explain it and to illustrate it with reference to geographical quantities, more particularly from the historical and cartographic points of view.