190 I. P. Косн. 
of observations for the months May—October I have made a mean 
series. The observations in the single series are performed at hours 
corresponding to the full hours, but in the three cases at even hours, 
in the other three at odd hours. On account of the variations from 
month to month, this circumstance would cause the curve of the 
mean series to zig-zag from hour to hour. In order to avoid this 
picture, so fatal to the sense and corresponding in no wise to the 
actual conditions in nature, I have in every single series interpolated 
the value of the twelve hours on which no observations have taken 
place, and from the values obtained in this manner, from hour to 
hour, in the six series, I have made the mean series, the twenty- 
four single values of which are indicated above. 
The significance of the irradiation and radiation in the case of 
=. is in all probability somewhat different over land and over water, 
as is also suggested by the curves shown in Fig. 15. In order that 
this may cause the smallest disturbance possible between the daily 
variation of the actinometric difference and the co-efficient of re- 
fraction, I have from the four curves in Fig. 15 made a mean curve, 
though at the computation of the ordinates from the table on 
pp. 186—187 I have left out the quite abnormal and disturbing 
values of June 15th at 10P10. 
Ak До 
+ 0,100 |+10,0 
+0.080 | = 8,0 
+0,060 | +6,0 
+0040 | = 4.0 
le | | 2 
Er SH 
Li QP АР GP oP о I = GENE #102 
Fig. 16. The co-efficient of refraction and the actinometric difference. 
In the table above the actinometrtc difference, a, is expressed in 
centigrade. In order to be able to use, in Fig. 16, for a the same 
