AT KEW OBSERVATORY, 1890 TO 1900. 
211 
appreciable uncertainty, for, as I have shown elsewhere, declination amplitudes at 
Kew and Falmouth are practically identical. Only once during the whole 11 years 
—viz., on February 14, 1892—-was there loss of trace owing to the light going beyond 
the edge of the sheet during a magnetic storm. In this case the maximum was taken 
as at the edge of the paper, so that the range deduced, 79', is almost certainly an 
underestimate. This is, I think, the sole occasion during the whole 11 years in 
which there was any appreciable uncertainty as to the range. 
Mean Annual Values. 
§ 9. A question of interest is whether any sensible difference, systematic or other¬ 
wise, exists between mean yearly values derived from all ordinary (“undisturbed”) 
days and from quiet days. In the case of Pawlowsk (St. Petersburg) Muller* found 
that the mean annual values of the declination derived from Wild’s “normal” days 
(which are very few in number and exceptionally quiet) were throughout the period 
1873-85 invariably higher (more westerly) than those derived from all ordinary days, 
the average excess being 0'’24. Again, W. Ellis! found for the seven years 1889-96 
at Greenwich that the yearly means from the Astronomer Royal’s quiet days were on 
the average 0''08 higher than those from ordinary days. Only one year, 1891, showed 
the opposite phenomenon. Recently, however, Ellis | has found that while the quiet 
day mean was the larger in 1903 by O'T, it was smaller by the same amount in 1904. 
To make the comparison absolutely fair, the mean quiet day in each month ought 
to come exactly in the middle of the month. This is only approximately true of the 
Astronomer Royal’s quiet days. This being so, it seemed hardly worth while 
attempting an accuracy of the order 0 r '01 in individual years, as this would have 
entailed the recalculation of the quiet day mean values, which are given in A. only to 
the nearest 0'T. Only the last seven years of the period were considered individually. 
The results obtained were as follows :— 
Algebraic Excess of Quiet Day Mean. 
1894. 
1895. 
1896. 
1897. 
1898. 
1899. 
1900. 
Whole 11 years. 
-O'-l 
O'-O 
+ O'T 
+ O'T 
+ O'T 
O'-O 
-0'-l 
+ 0' • 02 
For Greenwich Mr. Ellis, using the same quiet days, got +0'T in all three years 
1894 to 1896. 
Taking everything into account, all we seem entitled to infer is that the 
Astronomer Royal’s quiet days give a yearly mean in very close agreement with that 
obtained when only days of marked disturbance are omitted. 
* 1 Repertorium fiir Meteorologie,’ vol. 12, No. 8, 1889. 
t ‘Brit. Assoc. Report for 1898,’ p. 80 (see especially p. 108). 
t ‘ Roy. Soc. Proc.,’ vol. 79, p. 15. 
2 E 2 
