trees is probably not as good in details as Huntington's samples from young and sensitive trees. His material is well worth cross-identifying and dating with care, and then comparing with any records of snowfall which can be obtained from the sequoia groves. It is greatly to be regretted that Fresno, 65 miles away and at 5,000 feet lower elevation, is the nearest point where precipitation records can be obtained for a period long enough to be of value.
Future work.— It will be very interesting to find whether the characteristics of the correlation at Prescott are general in arid climates and dry soils and whether practical formulas for conservation in moist soils or climates can be worked out. When this is done the significance of the study of annual rings will be greatly increased.
The study above described raised emphatically the question as to the extent of the region or district from which comparative rain records should be selected. Such a meteorological district could be defined as one in which homogeneous weather elements are found. But we immediately ask ourselves the questions: must all weather elements be alike in it or is it sufficient to have only rainfall (for example) essentially the same throughout; will the district remain constant through indefinite time or will it change; is the district for short-period weather changes the same as the district covered by secular changes. In the present discussion I have understood by meteorological districts such regions as may show similar or identical variations in some one weather element. It seems likely that a region which may show unity in small or rapid variations may not do so in large and slow variations, or more likely may be a small fraction of a region which will show unity in large variations.
Meteorological districts and growth of trees. — The cross-identification of trees over large areas has already suggested the use of annual rings as a possible aid in delineating meteorological districts. This function of the rings has received some exemplification in the present study. For instance, the pine trees of Norway differed in such a way that it was necessary to divide them into two classes, one of which came from the outer coast near sea-level and the other from the inner fjords and mountains. The trees from these different regions show strong reversal with reference to each other. Again, the trees from the lowlands about the Baltic Sea show marked similarity in their variations and indicate, as we would expect, a homogeneous district. Furthermore, groups from near the Alps show strong differences from the other European groups, as we might expect from our experience with the five groups from the mountainous country about Prescott. A rugged and mountainous region is very difficult to divide satisfactorily