TREE RINGS AND HISTORY — SCHULMAN 471 



These considerations do not, of course, apply to fluctuations from dec- 

 ade to decade. 



Three critical regional indices for the western United States, based 

 on the ring growth of dry-site conifers of the type just discussed, are 

 compared with appropriate water-year flow data in figure 3. The 

 strong parallelism to be noted in this figure suggests that variations 

 from year to year in the pattern of climate — distribution and intensity 

 of storms, anomalous temperatures, the varying ratio in different years 

 of runoff to rainfall, etc.— introduce only a relatively small error in 

 such regional indices. It is highly probable that local biotic and 

 other factors are to a substantial degree canceled out in these large- 

 scale means. 



Pronounced differences in some years in the march of the compared 

 variables will undoubtedly be corrected somewhat with more repre- 

 sentative tree indices which await development; minor errors in the 

 river-gage data no doubt also exist and may be found and corrected. 

 For example, reexamination of the early gage measurements at Fort 

 Benton, ISIont., recently led to somewhat reduced annual totals for the 

 flow of the Missouri Eiver at this station fi'om 1891 to 1918, as shown 

 in the figure. Despite such improvements in the data, however, it 

 cannot be hoped that the observed correlation between river flow and 

 the best regional tree indices will ever substantially exceed about 

 4-0.85, the presently observed value for some of the comparisons. 



Certain features of the probable past rainfall in the western United 

 States suggested by figure 3 may be specially noted. A strong though 

 by no means perfect parallelism is evident between the Colorado 

 River basin and southern California. The final 25 or 30 years of 

 the 1500's in these regions seem to have been characterized in general 

 by deficient growth, rainfall, and river flow of severity substantially 

 greater than that in the recorded "dry spells" near 1900 and 1934; 

 data based on the oldest trees appear to show, in fact, that this was 

 the worst drought since the century-long dryness of the 1200's. The 

 pronounced deficiency of the late 1500's seems to have affected the 

 Missouri River basin also. The occurrence of two extremely deficient 

 years in succession, however, is apparently quite rare, except during 

 the major general minima. The bearing of inferences of this nature 

 on water and power reserves, such as those at Ploover Dam on the 

 Colorado, is perhaps obvious. 



The minimum near 1900 in water supply in the Southwest, evident 

 in the figure, is less severe in the upper Missouri River basin and, 

 in fact, is replaced still farther north by what appears to be the 

 greatest maximum in three centuries, at latitude 51° N. in Banff 

 National Park. 



370930—56 31 



