392 Mr. Gilpin’s Observations on the Variation , 
not be considered stationary, but I would not from so short 
a period conclude that it really is so. 
From the observations of sixteen months, viz. from Sep- 
tember, 1786, to December, 1787, both inclusive, the variation 
may be considered as generally stationary at or about 7 or 8 
o’clock in the morning when it is least ; and about 1 or 2 
o’clock in the afternoon when it is greatest ; and therefore it 
has been the practice in determining the true variation put 
down in these tables, to take a mean of the two morning, and 
the two afternoon observations, made at those times, for the 
true variation. 
In March, 1787. The mean monthly diurnal alteration of 
variation was found to be i5',o ; in June ig',6 ; in July 1 g',6; 
in September 14! >8 ; and in December 7 ',6. But on a mean 
of 12 years observations, from the year 1793 to 1805, the 
diurnal alteration of variation in March was only 8', 5 ; in June 
1 1 / ,2 ; in July 10', 6 ; in September 8', 7 ; and in December 3', 7. 
Table IV. contains the differences for 12 years, viz. from 
if 93 to 1805, between the observations of the variation made 
in the months of March, June, September, and December, or 
at the times of the vernal and autumnal equinoxes, and 
summer and winter solstices ; by a mean of these 1 2 years, 
the variation appears to increase or go westward, from the 
winter solstice to the vernal equinox o', 80 ; diminishes or 
goes eastward from the vernal equinox, to the summer 
solstice i 7 5 43 ; increases again from the summer solstice to 
the autumnal equinox 2^,43 ; and continues nearly the same, 
only decreasing o', 14, from the said equinox to the winter 
solstice. 
These differences at the times of the equinoxes and sol- 
