72 



VEGETATION OF A DESERT MOUNTAIN RANGE. 



temperature of 32 or less was registered at a station during an interval 

 of two or three weeks, the date of such minimum could be determined 

 by finding the exact date of the lowest temperature for the same period 

 at the Desert Laboratory, and such determinations undoubtedly have 

 a very slight possibility of error. Taking into account the number of 

 direct observations and the larger number of estimations, the limiting 

 dates of the frostless season, given in tables 10 and 11, may contain 

 errors of as much as 7 to 10 days. The swamping of these errors by 

 averaging the dates for the 6 years of observation reduces the probable 

 error to about 5 days. 



TABLE 11. The altitudinal shortening of the frostless season in the Santa Catalina Mountains, 

 as shown by the dates of the last spring occurrence and the first autumn occurrence of a 

 temperature of 32 at 5 altitudes in 1911, 1912, and 1913. 



[Data for 9,000 feet are partially interpolated.] 



In figure 2 are shown separately the curves for altitudinal shortening 

 of the frostless season for the years 1908, 1909, and 1910, and the years 

 1911, 1912, and 1913. In the latter group of years the advent of spring 

 was nearly a month earlier at the middle altitudes than it was during 

 the former years, and the advent of winter was correspondingly later, 

 in spite of the fact that the frostless season was of approximately the 

 same length at the Desert Laboratory during the two periods. A con- 



