Mr. W. Brennaud. 



strip," and divided by 30 are the numbers in column 3. 

 the numbers in Column 4 represent [Uj. 



Now [V] = [W] should be- . [U]. 



Similarly, 



These observations do not give [U] large enough. 



Also, the observations of 12th August would show the value of u 

 when = 20 to be about 0'018 or 0'019. But the Table D shows the 

 true value of ia when a = 20 to be O0107 ; that is to say, the read- 

 ings of 12th August, 1890, with the octant actinometer are altogether 

 too high. This may easily be so without any fault in the instrument 

 or error in the observations, and on two reasons. First, the presence 

 of any bright cloud may have given the readings [V] and [W] too 

 large. (Bnnsen and Roscoe, in ' Phil. Trans.,' 1859, p. 905 : " These 

 observations prove that the presence of a thin film of cloud increases 

 the amount of chemical illuminating effect in the most striking 

 manner." The clouds "act as mighty reflectors of light.") Secondly, 

 a very slight haze over the Sun would give the sunstrip too low, and 

 thus largely increase the results of columns (3) and (4) read by it. 



I do not consider these observations to decide anything as to the 

 merits of the octant actinometer, which can only be satisfactorily 

 tested by the sky of Dacca or some similar subtropical or tropical 

 station. 



37. It is difficult to determine which method of resolution of the 

 sky and Sun gives the most useful measure of the general total effect, 

 whether for determining the time of exposure of a photographic plate 

 or for estimating the effect on vegetation. Sir H. Roscoe has taken 

 (for the sky) the resultant action on paper exposed horizontally ; 

 append, therefore, in Table G the chemical action similarly measui 

 so that column 2 is exactly = the " diffuse daylight " of Roscoe, anc 

 column 4 = the " total daylight " of Roscoe. This table is deduced 

 by calculation from the Dacca Table B, by the aid of the law i* = 

 ia. cosec 0, i.e., from the value for I H I u in Art. 24, and the value 

 Q. = 2i a (ir sin + 2 cos at), which are directly derived from that law. 



This table, as far as * = 45, is a direct consequence of the Dacca 

 observations. The values given from 50 up to 60 are a theoretical 

 extension, perhaps as near as would be given by interpolation between 

 known extremes. I do not think the numbers for a. from 60 to 90, 

 which might be arrived at in a similar way, would have any real 

 value. 



This table, equally with the Dacca Table B, shows how large 

 sky effect is in comparison with the Sun effect, especially for altitnc 

 of tbe Sun below 30. This may be the explanation of the reason why 

 trees close to the north side of greenhouses exercise a prejudicial 

 influence. 



