TERRESTRIAL MAGNETISM AT KEW OBSERVATORY. 
77 
but the difference between the agreement in their case and in that of whole sunspot 
areas was small. In the case of H, the umbrae corrected gave as good an agreement 
as the whole spots, and the umbrae projected were very little behind. In D, the 
umbrae gave a distinctly inferior agreement; and in both D and H the agreement in 
the case of the faculae was very decidedly inferior to that from the other solar 
quantities. 
The natural inference from these results—though in the absence of trial it can oidy 
claim to be a probability—is that in any attempt to investigate a numerical relation¬ 
ship between sunspots and terrestrial magnetism it is not likely to matter mucli 
whether we consider Wolfer’s frequencies or the Greenwich areas of whole spots or 
of umbrse; and, if we take Greenwich data, it would appear unlikely to make much 
difference whether projected or corrected areas are employed. 
§ 3. The first attempt* which I made to investigate a relationship between sunspots 
and terrestrial magnetism on individual days employed Wolfer’s provisional 
frequencies, as published quarterly in the ‘ Meteorologische Zeitschrift.’ When 
discussing the Kew “quiet” day diurnal inequalities of the eleven years 1890 to 
1900, it occured to me that if any intimate relationship exists between sunspot 
activity and magnetic disturbance on the same day, we should expect sunspot 
frequency to be decidedly below its mean on the average magnetically “quiet” day. 
A comparison was accordingly made of the mean sunspot frequency derived from the 
“quiet” days selected by the Astronomer Royal, and that derived from all days of 
the year. The comparison was made for the eleven years 1890 to 1900 as a whole, 
and for two sub-groups of years, representative of sunspot maximum and minimum. 
It was also made, employing the whole eleven years, for three seasons of the year, 
winter (November to February), summer (May to August), and Equinox 
(remaining four months). In no single instance was there a difference between the 
mean sunspot frequencies from all days and from “ quiet ” days only, such as 
suggested any real relationship. The two mean frequencies derived from all months 
of the eleven years, viz., 41’22 for all days and 41‘28 for “quiet” days, were 
practically identical. 
§ 4. The next attempt! which calls for remark proceeded on different lines. It made 
use of the annual Greenwich tables, which give the daily values of projected sunspot 
areas—expressed in terms of the one millionth of the visible disc as unit—applying 
them to the 660 selected “quiet” days of the years 1890 to 1900, and to 209 days 
of the same eleven years selected for the large size of the magnetic disturbances 
shown. The days of each month were divided into three groups ; the first and last 
of these groups in a month of 30 days contained respectively the 10 days of largest 
and the 10 days of smallest sunspot area. It was then investigated how the 660 
“quiet” days and the 209 highly disturbed days distributed themselves between the 
* ‘ Phil. Trans.,’ A, vol. 202, jj. 433. 
t ‘ Phil. Trans.,’ A, vol, 208, p. 234. 
