EARTH’S MAGNETISM AT PAVLOVSK AND POLA (1897-1903). 
305 
Hence the probable errors of the results for vertical force are distinctly less than 
those for the other two elements, but here again, an account of the much smaller 
magnitude of the variations in vertical force, these are 'proportionately the least 
accurately determined .* These differences between the probable errors are, however, 
small compared with those between the errors relating to the different harmonic 
components. The means of the successive columns of Table V. (after the application 
of the small amplitude corrections of § 5) are 
d\. b\. cio . bo. d%. d±. b±. 
117 9*9 67 6'6 4’4 4*3 4’2 3*4 
Hence the probable errors of either resultant of the harmonic components of 
frequency 1, 2, 3, 4, are respectively 10'8, 6*6, 4'3 and 3'8, in units of 10~ 7 C.G.S. 
These results apply directly to Table VIb. 
§ 11. In Tables VIa. and VIb. are given the results of this paper in their complete 
form, the necessary corrections to amplitude and epoch having been applied to the 
mean values of a n and b n in Tables II. to IV., after which these were converted into 
the geographical components of force, with a common unit 10“ 7 C.G.S. The C, 6 and 
a, b results are given separately in Tables VIa. and VIb. The latter is the more 
useful for the ultimate purpose of this work, viz., for the investigation of the potential 
of the lunar diurnal variations; the C, 6 table, however, gives a better idea of the 
relations between the various components at a single observatory, and between 
corresponding components at the two observatories. 
* The following comparison with the values deduced for the a 2 , & 2 coefficients of the lunar diurnal 
variations at Pavlovsk and Pola by van Bemmelen (‘ Meteorologische Zeitschrift,’ May, 1912) may also 
be of interest, van Bemmelen divided his data into the summer and winter half-years, and therefore 
the only comparison that can be made is between the mean variations throughout the whole year. 
Moreover, as he did not separate the different lunar phases, but took the mean of a number of whole 
lunations, the values of a i, bi which he gives represent accidental error and are no true lunar effect (as he 
fully recognized); the real diurnal lunar component simply disappears from the result when whole 
lunations are dealt with. Further, the method of computation used by him was different, as also the 
period of observations dealt with (1894-1903 in the case of Pavlovsk, and 1899-1906 in the case of Pola). 
While exact agreement is not to be looked for, therefore, the following comparison has some interest in 
the present connection:— 
Element. 
PayloTsk. • 
Pola. 
a 2 . 
b,. 
• 
K 
VAN 
Bemmelen. 
Chapman. 
VAN 
Bemmelen. 
Chapman. 
VAN 
Bemmelen. 
Chapman. 
VAN 
Bemmelen. 
Chapman. 
North force . . . 
ii 
0 
15 
48 
52 
42 
59 
56 
West force .... 
53 
61 
- 2 
7 
74 
6L 
9 
30 
Vertical force . . . 
10 
0 
12 
0 
-20 
-27 
37 
27 
YOL. CCXIY.-A. 2 R 
