[Illustration: Fig. 113. Facial urns.] [Illustration: Fig. 114. Armadillo designs, Chiriqui.] [Illustration: Fig. 115. Designs of the Bushongo representing the head of the antelope and the beetle.] Here arises the important problem whether we are to assume that all these forms are developments from realism to conventionalism, as has often been assumed, or whether the converse process may not also have occurred, namely that a geometrical design existed and that a gradual development towards a realistic form took place, that a meaning was read into the geometric pattern and that in this way the significant forms originated. Unfortunately historical evidence is hardly ever available and we are compelled to rely upon indirect evidence. We cannot follow the excellent example set by Riegl in his detailed study of the history of the introduction of curved lines in Mediterranean art. However, we can apply the geographical method, the only one that has made it possible to unravel part of the historical development of people that do not possess written records, and for cultures the development of which cannot be traced by archaeological evidence. It is possible to establish with a high degree of probability the relationship of cultural forms and their gradual spread by means of a study of the distribution of ethnic phenomena and their variations in the sections of the area in which they are found. This method is strictly analogous to the one applied by biologists in their studies of the gradual distribution of plants and animals. In our case we must try to trace the distribution of designs together with the interpretations given to them by different tribes. If we should find consistent interpretations of the same form over large areas, perhaps even more realistic forms in a central district, more conventional ones in outlying parts of the country, but in all of them the same interpretation, we should have to consider this as plausible evidence of an origin of the conventional types from a realistic representation. If, on the other hand, it should be found that in the whole area realistic forms and conventional forms were irregularly distributed and that furthermore the meanings of identical or similar forms did not agree, then the origin of conventional forms from realistic ones would seem to be quite unlikely. Then we should have to assume one of two possibilities; either the form must have spread gradually over the whole area and must have been given a meaning independently by each people,—in other words the meaning must have been read into the pattern,—or it may have been that a dominating style has forced a diversity of realistic representations into the same geometric patterns. In the latter case we should have to consider the processes by which realistic forms should gradually change into unrecognizable conventional ones; in the former the processes by which from a conventional form a representative one develops.