186 DE. C. CHREE: AN ENQUIRY INTO THE NATURE OF THE RELATIONSHIP 



Results calculated for Milan from the period 1890 to 1900, which is the period 

 chiefly utilised in the present paper, differ but little from those found by RAJNA for 

 the periods 1836 to 1894 and 1871 to 1894. 



The tendency for b to be small in winter and large in summer, described at Kew, is 

 also, in general, conspicuous elsewhere ; but there are exceptions, especially at tropical 

 stations. 



The tendency in b/a to be large in winter as compared to summer, so prominent at 

 Kew, is also, in general, prominent at other northern stations, but the phenomenon is 

 comparatively inconspicuous in the case of the declination range at Greenwich. At 

 the tropical stations the seasonal change in b/a appears much reduced and is some- 

 what uncertain. 



There is no conspicuous difference between the "all" and the " quiet" days' mean 

 yearly values of b and b/rt for the ranges of the D and H diurnal inequalities at 

 either Greenwich or Pawlowsk ; but at Pawlowsk there is a somewhat notable 

 difference between "all" day and "quiet" day D results in winter, and the difference 

 between " all " and " quiet " day V results is very large throughout the whole year. 



If we exclude Mauritius, the values of W*b/a for the ranges in the mean diurnal 

 inequality of declination for the year at the several stations vary only from 65 to 73. 

 The corresponding values of b show also a pretty close agreement at the northern 

 stations, but the values for the tropical stations are much smaller. 



In H there is no very conspicuous difference in the values of b or b/a for the ranges 

 from the mean diurnal inequality for the year at the northern stations ; but the values 

 found for b/K, at Batavia and Mauritius are considerably smaller, while the value 

 found for b is smaller at Mauritius, but very materially larger at Batavia. 



When the formula (l) is applied to any ordinary measure of magnetic disturbance, 

 it gives much too high values for 1893 the year of sun-spot maximum and much 

 too low values for 1892. Thus the application of (l) to disturbances has not the same 

 justification as its application to ordinary diurnal inequalities. It may, however, 

 serve a useful purpose in giving a greater degree of definiteness to the comparison of 

 contemporaneous disturbance phenomena at the same or at different stations. 



In the case of results obtained by the application of (l) to individual months of the 

 year a considerable latitude must be allowed to chance, especially in winter months 

 when the diurnal range is small, unless an exceptionally long series of observations is 

 available. liesults obtained from arranging months in seasons are much less exposed 

 to numerical uncertainties, but they are insufficient for the reason that there are 

 conspicuous differences between months which have to be grouped under the same 

 season. Tins remark applies more particularly to winter and equinoctial months in 

 higher latitudes. 



[June 8, 1904. The following additional data all obtained by the method of 

 least squares apply to the ranges of the mean diurnal inequalities for the year at 

 Irkutsk (" all" days) and Colaba (" quiet" days), and to the mean difference between the 



