AT KEW OBSERVATORY, 1890 TO 1900. 
207 
altogether too arbitrary a process. It was decided to omit such disturbed days 
entirely when calculating the regular diurnal inequality. The days thus omitted 
numbered 209, or an average of 19 a year. 
§ 4. It should be clearly understood that in classifying a day as “ disturbed,” regard 
was paid exclusively to the nature and not to the mere magnitude of the disturbance. 
If the irregular movements were mainly of the type seen at QRS in the diagram on 
p. 206, or if the declination showed an abnormally high or low value for several 
successive hours, the curve was classified as disturbed, though the range might be less 
than in a neighbouring “ordinary” curve where the disturbances approached the 
type illustrated by ABE of the diagram. That the method of choice is open to 
criticism, I freely acknowledge. It introduces a personal element, and something 
unquestionably depends on the individual’s freshness and nerve at the moment. If 
in his best form, he may at once make up his mind how to smooth a disturbed curve, 
even when heroic rectifications are necessary, whilst if he is tired and hesitates he 
probably in the end relegates the curve to the disturbed class. The selection of the 
disturbed days was in every case made by myself, and the curves for a single year 
were always considered together. Thus I regard the number of disturbed days as 
more appropriate for determining the relative amount of disturbance at different 
seasons of the year than for comparing one year with another. I have discussed this 
question at some length because other criterions for disturbance have been applied. 
Thus Mr. Ellis has classified days as disturbed, and as of greater or less disturbance, 
mainly according to the amplitude of the range, and his classification has been 
followed by Mr. Maunder in his interesting researches into the relationship between 
sunspots and magnetic storms at Greenwich. At first sight, a reference to amplitude 
seems a simpler and more satisfactory method than the one that I have adopted, but 
it is in reality, as I have explained elsewhere, highly arbitrary. This will, I think, 
be recognised on referring to Table XIV., showing the mean of the absolute daily 
ranges (maximum less minimum) for each month from 1890 to 1900. The mean, it 
will be seen, varied in individual months from 4' - 73 in December, 1900, to 24 /, 02 in 
March, 1892, and taking the mean of the twelve months it varied from 9'T7 in 1900 
to 17 /, 70 in 1892. Even restricting ourselves to the Astronomer Royal’s quiet days, 
the mean ranges for August, 1892, and December, 1899, were respectively 15 /- 20 and 
3''12. A range of 15' at mid-winter at sunspot minimum may imply much more real 
disturbance than a range of 30' at the equinox near sunspot maximum. 
§ 5. Mr. Maunder’s list for the years 1890 to 1900 included 150 disturbances. His 
figures, however, denote not the number of disturbed days, but what he believed to 
be the number of separate magnetic storms. Disturbed conditions usually last for a 
good many hours, and not infrequently for two or more days. Thus the number of 
disturbed days naturally exceeds the number of separate storms. When disturbed 
conditions last for several days it is sometimes doubtful whether one is dealing with 
one or with several storms separated by comparatively quiet interludes. If we classify 
