21-2 



Mr. \V. JJn.-niiaiul. 



Columns 1 and 2 arc copied from lloscoc, I.e. ; column 2 gives my 

 Qz = 2Jo (ir sin a, + 2 cos ). In column 3 I give /a, calculated from 

 this equation. In column 4 I place the values of ia obtained from 

 table B, the " sky alone " column, by the aid of the formula at the 

 end of Art. 24. 



In the 5th column the values in column 4 are brought up by pro- 

 portion for comparison with those in column 3, taking the observa- 

 tion at altitude 42 13' as the best ; i.e., increasing all the numbers 

 in the ratio of 121 to 160. 



Table E. 



The discrepancies do uot appear at first sight great between the 

 results of Sir H. Roscoe and my own. But his observations would show 

 the maximum value of ia attained when the Sun was at or near the 

 zenith, mine that this maximum occurs when the -Sun is about 45 OP 

 50 altitude. 



It is true that in the Dacca Table B, the actual observations ext 

 only to 45 or thereabout, and that the values for altitudes of tl 

 Sun above 45 are only filled in hypothetically ; but my 

 established observations at Dacca, for altitudes of the Sun from 30 

 to 45, show directly that at altitudes of the Sun of 45 or 50 

 value of ia would reach a maximum. 



In my Dacca observations, each additional 5 to the Sun's altitude 

 brings into effect an additional 5 gore of the sky. It is therefore 

 clear (apart from the law ig = ia cosec 6 and the integrations con- 

 sequent thereon) that ia will have a maximum value when , the Sun's 

 altitude, is about 50 or 60. 



I am not surprised that so considerable a discrepancy results from 

 a comparison of the observations. In a single series of observations, 

 the incidental errors of reading, &c., would introduce into the small 

 numbers given in column 3 sufficient differences to alter entirely the 

 law indicated for ia. 



