517 



shine-recorder not being small with respect (o the time-scale, ma}- 

 give rise to a smaller value of IT than is derived from the estimated 

 cloudiness, because a more or less continuous discoloration can be 

 observed when light and shadow are in reality varying in a dis- 

 continuous way. It will be difficult to take into account either factor 

 as they are dependent on the velocity of the cloudmotion by which the 

 possibility of a registration of relatively feeble rays of light is determined. 

 The influence of these factors can be determined only by experiment, 

 which however would be difficult to carry out. It would be necessary 

 to extend it to a great number of ditferent cases and, in order to 

 inquire in how far compensation happens to occur between too large 

 a percentage of sunshine in the case of slowly moving clouds and 

 too small a percentage when the clouddrift is more rapid, the ex- 

 periment would have to be extended to a great number of ditferent cases. 

 Finally it may be noticed that the fact that clouds generally show 

 a flat base and a pointed apex, might be taken into account by 

 considering the upper half of ellipsoids, the equatorial plane being 

 situated in the plane Ms M^ . . M,, . Then the duration of sunshine 

 corresponding to a gi^en value of W increases, but as, on the other 

 hand, the apparent cloudiness decreases, the sums remain nearly 

 the same. 



T A B L E VI. 



Calculation for spherical clouds. 



De Bilt, June + July, ^ = 37°42'.5 



