1867.] 



of the Magnetic Declination. 



415 



The process employed in reducing the observations was identical with 

 that generally adopted in such cases. The disturbed observations were 

 first eliminated, by rejecting all that differed from the final normal belong- 

 ing to the same solar hour by more than a certain separating value, which 

 was taken at 3'61 minutes of arc. The ehmination of the larger disturb- 

 ances having been thus effected, from every remaining reading (R) of the 

 magnet's direction the final normal (N) belonging to that solar hour was 

 subtracted, so that the residue R— N is devoid of the influence of the solar- 

 diurnal variation. This residue is positive when the north end of the 

 needle is to the east of its mean position, and negative in the contrary case. 

 The number of observations at command amounted to 38,194, of which 

 4178 single observations were excluded from the discussion as being beyond 

 the assumed limit used for separating the greater magnetic disturbances, 

 leaving 34,016 available for the purpose of determining the lunar-diurnal 

 variation. 



The treatment of the residues with a view to classification according to 

 lunar hours presented no particular novelty. It may be mentioned, how- 

 ever, that before entering on any general discussion, every month's result 

 was calculated separately. The values for the various months were after- 

 wards arranged, irrespectively of the year, in two groups, viz. Sun South 

 (October to March) and Sun North (April to September). Thus the mean 

 lunar-diurnal variation was obtained separately for each half of the year, as 

 well as for the whole year. On examining the results, irregularities in the 

 lunar-diurnal variation presented themselves which seemed to show that 

 that variation depended in some degree on the moon's position with refer- 

 ence to the equator, on the circumstance whether her declination were 

 north or south. 



Accordingly the whole series of observations were rearranged in groups, 

 *' Declination of the Moon South " and " Declination of the Moon North," 

 so as to cause her declination to be divided between the hours of the day, 

 all those days being rejected on which the moon was close to the equator. 

 The 118 groups of lunar-diurnal variation thus formed were subsequently 

 classified according to whether the sun's declination was south or north. 



A Table was thus formed giving far each lunar hour the lunar-diurnal 

 variation of the magnetic declination separately under each of nine condi- 

 tions formed by combining each of the three conditions. Sun South and 

 North, Sun South, Sun North, with each of the three Moon South and 

 North, Moon South, Moon North. The results given numerically in the 

 Table were also laid down graphically in curves. 



A glance at the curves shows that the lunar-diurnal variation must be 

 regarded as being influenced by both the sun and the moon ; for it is seen 

 that in case the declinations of the two bodies are of the same name, the 

 curves show greater regularity than in the contrary case. 



The question whether during the winter season any lunar-diurnal varia- 

 tion can be traced at all can, the author conceives, no longer be entertained 



