NO. 3 RADIATION OF THE ATMOSPHERE ANGSTROM 39 



chosen the observations at Indio during the nights of July 23 and 

 24, and at Lone Pine on August 5 and August 11. I have preferred 

 these nights to the others because of the constancy of the humidity 

 and the relatively great temperature difference between evening and 

 morning values. By means of the formula connecting radiation and 

 humidity obtained from the Algerian values at constant temperature, 

 a small correction may be applied to these Californian observations, 

 in order to reduce them to constant humidity. The logarithms of 

 the radiation values thus obtained are calculated and also the loga- 

 rithms of the corresponding temperatures, tables III and IV. If log 

 E at is plotted along the y-axis, log T along the #-axis, it ought to be 

 possible to join the points thus obtained by a straight line, if the for- 



civ 



mula (2) is satisfied. The slope of this straight line ( — =con- 



dx 



stant = a) ought in such a case to give us the value of a. 



I have applied this procedure to the observations mentioned and 

 found that within the investigated interval the logarithms of radia- 

 tion and of temperature are connected to one another by a linear 

 relation. Figure 4 gives the logarithm lines corresponding to the 

 Indio observations. The deviations from the straight lines are some- 

 what larger for the Lone Pine values, but the discrepancies seem not 

 to be systematic in their direction and I therefore think that one may 

 regard the formula (1) as satisfied within the limits of the variation 

 that can be expected as a result of the many atmospheric disturb- 

 ances. The following table gives the values of a obtained from the 

 observations on the four nights selected: 



Place Date a Weight 



Indio July 23 3.60 4 



Indio July 24 4.27 4 



Lone Pine August 5 4.4 1 



Lone Pine August 11 4.4 1 



Weighted mean : a = 4.03. 



The table shows that the value of a is subject to considerable varia- 

 tions, which is a natural consequence of the great variations from the 

 average conditions, to which the atmosphere is subject. In the fol- 

 lowing pages, when I have used the value 4.0 as an average value for 

 a, in order to reduce the various observations to a constant tempera- 

 ture (20 C.)f this procedure is held to be justified by the preceding 

 discussion, as well as by the fact that, in applying this method of 

 reduction, we obtain an almost constant value for the radiation 

 during the night, if we, reduce it to a constant humidity. For 

 all other values of a, we shall get a systematic increase or de- 



