374 BOTANICAL GAZETTE [NOVEMBER 
these values do indeed vary, but also that their variations are in 
accordance with the geographical positions of the stations con- 
cerned. It is clear-from fig. 3 that the portions of the United 
States characterized by ratios below 9.00 lie mainly near the 
margins of the country. The lowest ratios of our series occur in 
western Washington. Other areas with ratio values below 9.00 
are indicated for northwestern California, southeastern California 
and southwestern Arizona, southeastern Florida, northeastern 
Maine, and north-central North Dakota. It thus appears that a 
line representing a ratio of 9.00 may be passed, according to the 
indications of fig. 3, around the area of the United States, entering 
within its boundaries only sufficiently to include the areas just 
mentioned. 
On the original chart, from which fig. 3 is taken, were drawn 
lines representing the ratio values by tenths, from 9.0 to 10.0. 
From this chart (as of course also from the numerical data given in 
fig. 3) it is evident that the ratio values with which we are dealing 
are consistently distributed, with few or no exceptions. An area of 
high ratios occurs in the Appalachian Mountains, having the 
highest value at Anniston, Alabama (10.33). Another similar 
area occupies the Rocky Mountains and appears to extend north- 
ward to the Canadian boundary, with maximum values of 9.81 
(Williston, North Dakota and Durango, Colorado). A small area 
of high values lies about Moorehead, Minnesota (9.84), and 
Huron, South Dakota (9.82). A less pronounced but strongly 
and consistently indicated area occupies the southern peninsula 
of Michigan and eastern Wisconsin, the maximum here being at ” 
Port Huron, Michigan (9.69). Finally, a somewhat questionably 
indicated area of high ratio values centers about San Luis Obispo, 
California (10.44). Interior areas of low ratios require no dis- 
cussion; they are few, and are generally based upon the evidence 
of single stations. ; 
From the study of this chart it seems clear that the ratios of 
direct temperature summations, for the normal frostless season, t0 
the corresponding efficiency summations as here derived, must be 
considered as a measure of some climatic characteristic. Just 
what this characteristic may depend upon we are not here attempt- 
