in California and Arizona he finds that the grasslands must once have been continuous across the Mohave and Colorado deserts when they had an annual rainfall of about ten inches in contrast to a rainfall of but two inches a year, as on the floor of the Mohave desert today, or an amount so small as to be incapable of supporting even the most xerophytic of the grasses. He concludes that the region has passed from a condition of mixed prairie at the close of the Pleistocene or Glacial Period to the desert conditions of the present under the pressure of shifting climate. The effect of having the grassland of California in contact with the mixed prairie and the losing of that contact, would be the development of relict groups of grassland in favorable situations though now far removed from similar tracts with which they once enjoyed contact. Our whole thought of these outliers has been that they represented a sort of advance guard, an invasion of the drier area, whereas the more closely analytical field studies have shown that they are remnants of a once more extensive grass cover. Migration without the agency of man or water is practically always local, and the wide distribution of these forms of grasses and shrubs requires, as a prerequisite, a natural disturbance or broad climatic change.
Were such studies of grassland and woodland to be carried out in northwestern Argentina, there would be offered a chance to apply the results to the sites of former habitations in the valleys that border the Puna de Atacama. It is not to be supposed that an exact relationship can be established, for the type of culture with which we are dealing was not suf- ficiently far advanced to permit the drawing of definite con- clusions regarding the time of its existence. The point of chief interest would be to discover if the sites selected for habitations were now uninhabitable in fact and not merely inconvenient or difficult to inhabit. That would be the first point. In the second place one should set out to discover if possible what changes in the water supply and in the zones of vegetation would be required to make the sites of former settlement habitable wherever these occur about the borders of the mountain country. If carried out on broad lines over
