435 
cond; and so on to the twelfth. This is continued until the 
lunation is completed ; so that the same solar hour falls on 
each of the lunar hours in succession, and thus the diurnal 
variation due to the sun is completely eliminated in the 
monthly means. 
As the lunar day exceeds the solar by 0"8, or by four 
hours in five days, there will be thirteen observations in two 
out of every five lunar days. In all such cases the observa- 
tion at 1 a.m., being the first of the solar day, is omitted. 
There are thus twelve observations omitted in each lunation, 
one between each pair of successive lunar hours in the Table ; 
and it will be easily seen that the effect of such omission is to 
alter the mean interval from 2" to 2" 4™, corresponding to 30° 
of the lunar hour-angle. 
On account of the smallness of the periodical variation 
sought, it is indispensably necessary in this investigation to 
diminish, as far as possible, the effect of magnetic distur- 
bances, the presence of which would altogether mask the re- 
gular change. This has been done, as in the corresponding 
investigation of the ordinary diurnal variation, by omitting 
altogether days of disturbance, such days being defined to 
be those in which the sum of the differences between the se- 
veral scale-readings, and the monthly means of the corres- 
ponding hours, exceeds a certain assumed limit. 
The monthly means of the scale-readings, in each luna- 
tion of the three years, having been calculated in the manner 
above described, the results of the twelve lunations in each 
year are again combined, and their means taken. The fol- 
lowing Table contains the differences between the yearly means 
corresponding to each hour, and the mean yearly mean, reduced 
to angular value, one division of the scale of the instrument 
being equal to 0°7205. The positive numbers correspond to 
easterly deviations, and the negative to westerly. The num- 
bers in the first column are the lunar hours reckoned from 
the upper meridian passage, each lunar hour being 1* 2™. 
