288 DR. CHARLES CHREE: SOME PHENOMENA OF 



These inequalities are not deduced directly from the curve measurements but are 

 calculated from the corresponding inequalities in D, H, and V through the 



relations 



AN = AH cos D-HAD sin D, 



AW = AH sin D + HAD cos D, 

 Al = 1 sin 21 (AV/V-AH/H), 



where A represents departure from the mean value for the day. 



The numerical formulae actually employed, accepting their mean values for D, H, 



and V, were 



AN = 0-955AH-1-58AD, 



AW = 0-298AH+5-08AD, 



Al = 0-0278AV-0-0667AH. 



In the numerical formulae AD and Al are expressed in terms of 1' as unit, while ly 

 is the unit for AH, AV, AN and AW. 



11. As already explained, there is reason to expect a not inconsiderable pseudo- 

 element in the diurnal inequalities of individual months. When the number of days 

 available is large, the most satisfactory way of distinguishing between the normal and 

 the accidental is to compare results based on two or more independent sets of days 

 belonging to the same month of the year. But the same object is usually achieved 

 pretty satisfactorily by comparing results from different months of the same season. 

 This second method, of course, will break down if the seasonal variation in type is 

 very great, but that is rather an unlikely contingency in temperate latitudes. 



The results for H in Table VI. present irregularities in the successive hourly values 

 which are obviously of an " accidental" character, but the main features are clearly 

 not " accidental." In summer there seems to be only a single maximum and minimum, 

 the former about 7 p.m., the latter about 10 a.m. In equinox, and still more in 

 winter, there is evidence of a double oscillation, and in December and January the 

 forenoon maximum and the evening minimum are apparently the principal ones. A 

 closely similar difference between summer and winter was described in (A) for the 

 case of quiet days. 



The range of the diurnal inequality in H from the disturbed days bears to that 

 from quiet days a ratio varying from T26 : 1 to 3 '24 : 1. The mean of the twelve 

 ratios 2'08 is a little in excess of the corresponding mean 1'95 found in the case of D. 

 But whereas the average of the monthly ratios for the three seasons was in the case 

 of D for winter 279, equinox l - 62, and summer 1 '43, the corresponding figures in 

 the case of H are winter 2'08, equinox T42, and summer 2'74. The position occupied 

 by summer is thus exactly reversed in the two cases. The same phenomenon appears, 

 but in an even more striking way, in the seasonal results in Table IX. for N and W. 

 In the case of quiet days the range in W bore to that in N a ratio varying from 1'25 



