477 



In this case is the mean of n 1 observations, of which 24 give 

 the true means for the total solar influence, and the remaining n 25 

 being equally distributed through the hour-angles also give the mean 

 approximately. 



Instead, however, of combining the observations in this way, the 

 following method has been preferred. Let, in the quantities (B), 



Then H' +A' + ^ _ (H )= h' + )=< 



H" T + A" + x\- (H T ) = A" + (^ I )=^'o 



H -\-h -4-x CH )=A -\-(x ) =G? 



23 23 V 23 X v 23 ' I 



Summing the last two columns, we have 



, 



nl nl 



Similarly we obtain 



l , 



n 1 w 1 



It will be observed that in these summations there are two assump- 

 tions ; one, that the lunar diurnal law is constant throughout the 

 lunation, or series of lunations, for which the means are obtained ; or 

 that the quantity ([ in the expressions (B) is constant. If this be 



not exact, then the quantity will contain the variation due to this 



cause, and depend in part on the lunar hour-angle ; so that the mean 

 (H) which is employed in taking the differences will eliminate part 

 of the lunar action and partially distort the law. The other assump- 

 tion is that the mean solar diurnal variation, represented by (HJ/H^ 

 . . . . , is nearly constant throughout the period ; for, if not, the dif- 

 ferences due to such changes might be sufficient to mask any lunar 

 law, the latter having a small range compared with the former. 

 Also it should be remarked that the means h , h lt &c. are combined 



with the irregular effect 23 ^ . This effect, as far as it is due to 

 n 1 



