18 
DR. S. CHAPMAN ON THE SOLAR AND LUNAR 
The epoch of the data used is modern, the years 1902 and 1905* being chosen ; these 
were years of sunspot minimum and maximum in their eleven-year cycle, two such 
periods being considered in order that the influence of solar activity on the 
phenomenon might be definitely determined. Also, so that the seasonal changes 
might be studied, each year was divided into four quarters of three calendar months 
each, beginning with February, March and April as the spring quarter. In calculating 
the mean variation for each of these eight periods, all days were used except a very 
few which were so highly disturbed as of themselves to be able to modify the 
quarterly means appreciably. The published data in most cases gave the variations 
of horizontal force and declination instead of North and West force, to which they 
had to be transformed. In some cases changes of phase had also to be made, to 
reduce the data to the adopted time-origin, which is here the local mean time of noon 
at each station. In the formula (l), t represents local time reckoned in angle at the 
rate of 15 degrees per hour. The first four harmonics (n — 1 , 2 , 3, 4) have been used 
throughout, the coefficients a n , b n being expressed in force units of amount O'l y (10 -6 
C.G.S.), and reckoned positive to North, West, and radially (or vertically upwards). 
The initial data of this paper, relating to the solar diurnal magnetic variations, are 
given at the end of the discussion in Tables I. and II. (a X (/3),(l)-(4), (pp. 74, et seq.). 
§ 8 . General Outline of the Analysis of the Data. 
The data in the Tables I. and II. exhibit a considerable degree of regularity and of 
constancy in type ; thus, save for the increased amplitude in the later year, the 1902 
and 1905 values are closely similar : they show a general independence of longitude : 
and they are nearly symmetrical, or anti-symmetrical, with respect to the equatorial 
plane. These features are little less apparent in the terms of lower than in those of 
higher frequency. 
Besides this, however, there is an irregularity about the numbers which seems to 
represent something peculiar to each station, persisting from year to year, and also 
affecting different elements unequally, the North component, perhaps, being the one 
most affected. In order to assist in eliminating this local part of the phenomenon 
from the analysis, nine group means of the data of Table I. have been formed, the 
groups being indicated and numbered in Table A. It would probably have been 
advantageous had each group included a still larger number of separate stations. 
For the discussion of seasonal influence the method adopted was as follows : the 
mean of the spring and autumn data was taken to represent the main part of the 
phenomenon at the equinoxes, at which times there is a general similarity between 
* Or rather 1902, February, to 1903, January, and similarly for 1905. It may also be noted that the 
Batavian vertical force data for the 1902 summer quarter are drawn from June and July observations 
only, no records being available for May. Wolfer’s sunspot numbers for these years were 5 (1902) and 
64 (1905). 
