BETWEEN SUN-SPOT FKEQUENCY AND TERRESTRIAL MAGNETISM. 155 



large disturbance. Corresponding " quiet " day data appear in Mr. ELLIS' second 

 paper for the 8 years 1889 to 1896. Mr. ELLIS specifies several sotirces which may 

 have introduced some heterogeneousness into the earlier data as compared to the 

 later. Prior to 1848 there were only eye readings at 2-hour intervals, whereas 

 subsequently hourly data were available. Prior to 1864, when the magrietographs 

 were transferred to a new building, some uncertainty seems to have prevailed as to 

 the temperature correction for H, and the data for 1864 itself seem to be interpolated. 

 The data subsequent to 1864 would seem to be strictly homogeneous. 



6. Mr. ELLIS employed no formula, and, whilst his graphical method appeals 

 readily to the eye, it does not lend itself immediately to the present investigation. 



I have accordingly calculated values for a and b for each month of the year in both 

 D and H for the following periods: 1841 to 1896, 1865 to 1896, and 1889 to 1896 

 for both "all" and "quiet" days. In treating the first period, use was made of a 

 group of sun-spot maximum years composed of 5 sub-groups each of 3 years, viz., 

 1847 to 1849, 1859 to 1861, 1870 to 1872, 1882 to 1884, and 1892 to 1894. The 

 corresponding sun-spot minimum group consisted similarly of 15 years, made up of 

 5 sub-groups, viz., 1842 to 1844, 1854 to 1856, 1865 to 1867, 1877 to 1879 and 1888 

 to 1890. 



For the period 1865 to 1896 the groups of sun-spot maximum and sun-spot 

 minimum years were composed in either case of the last 3 sub-groups specified above. 



For the period 1889 to 1896 the groups were : 1892 to 1895 for sun-spot maximum, 

 and 1889 to 1890 for sun-spot minimum. 



The values thus found for a, b and b/a appear in Tables II. and III. In addition 

 to values for the individual months, the tables give values for winter, equinox and 

 summer each comprised of 4 months, as explained in 1 and for the year. These 

 seasonal and yearly values of a and b are simply arithmetic means of the individual 

 monthly values ; the seasonal and yearly values of b/a are derived from the seasonal 

 and yearly values of a and b. The tables also supply corresponding data for Kew as 

 given in (A), Table XL. 



X 2 



