426 
DR. T. R. ROBINSON ON THE REDUCTION OP ANEMOGRAMS. 
be satisfied, if the factors r r ^ ^ &c. were equal in every month. They, however, 
differ so little that I have thought it lawful to take their means for the twelve months. 
Though this is fairly warranted, yet it seemed advisable to test it by comparing for 
E the cosine constant, of the fifth order in the series for K, with the minimum square 
process. It gives for E T0525 ; the second approximation, using AE alone, gives 
T0305, which would be a little increased by using the corrections of the other constants, 
so that the agreement is sufficient. As the factors r r will answer for any 
year, I give their logarithms. 
A. 
O. 
B. 
P. 
C. 
E. 
0-00672 
0-00346 
0-02156 
0-02119 
0-04411 
0-04976 
D. 
s. 
E. 
T. 
G. 
U. 
0-09300 
0*08400 
0-16622 
0-11988 
Not determined. 
0-19731 
It does not seem necessary to give the constants A, O, &c. ; but instead the secon- 
dary constants of the formula w=K-{-K sin (r+0)-l-Ksin(r-|-2$)+&c., deduced from 
0 0 1 1 2 2 
their corrected values, are given as more convenient for computation in Table VIII. 
(p. 427). 
I have given the constants for the horary coefficients A, O, &c. to the 6th order for 
symmetry; but in fact I do not think any of them less than (hOS need be attended 
to. Even this limit is beyond what can be expected to be available when they are only 
determined by the observations of seven years, as is evident from what I have already 
said as to the P E of the quantities from which they are determined. Whether the 
diurnal variation of the coordinates follows the same law in different septennial periods 
remains to be determined ; probably it does. The constants belonging to K and K' are 
<p 
larger than the others, and, as derived from larger coefficients, merit more confidence. 
The effect of the terms of the first and second orders, which are the chief, are similar ; 
but the others present opposite phases, and would probably be modified by more 
accurate determination. It is here that I think changes in successive years will probably 
be found ; and were I to pursue this work further, I would combine the observations 
rather differently from what I have done in the present case. I would mean the 
homonymous hours of each month of each year, combine them in pairs, and mean them 
to get the K of each month. I would then compute the K constants, retaining the 
cosine and sine form ; and this should be continued through a few periods of the solar 
spots. This would decide the question whether the wind is affected by the conditions 
which modify those phenomena. 
At the same time the inspection of the horary means would show whether their laws 
vary with the time. Then the final constants could be determined for such intervals as 
might be considered sufficient. The sine and cosine formula, though requiring more 
