62 
Dr. Brinkley on the 
quiry, that dh only arises from an error in the place of the 
equinox. 
The mean of the latitudes thus found during the year, 
will be affected by an error s d L «*|— t d O + v d k, in which 
the coefficients s and t are so small, that the effects of dL 
and d O will be insensible. Thus the eighty-seven observa- 
tions give my latitude = 53 ° 23' 1 2", 39 + 0,04 d L -f- 0,2 1 d O 
-f- 1 ,42 d k. 
By circumpolar stars remote from the Pole 
Co-lat. == 36° 36' 47", 15 -J- 1,62 dk, 
making the sum = 90°, 
we deduce dk — o",i5 — 0,01 d L — 0,07 d O. 
This small value of dk appears to confirm the accuracy of 
the constant k that I had used. But if I relied on this I should 
deceive myself ; for on examining the series of latitudes de- 
duced, it is evident that this coincidence arises from the 
circumstance of more observations having been made while 
the sun was on the north side of the equator, than while on 
the south. The latitudes deduced show clearly, that had more 
observations been made nearer the winter solstice than the 
summer, the value of dk would have been much more con- 
siderable. 
This contradictory result, and some other circumstances 
that appear on an examination of the latitudes deduced, seem 
to point out that some new equation is required to be applied 
for the solar refraction. At least, that no conclusion can be 
drawn as to the exactness of a table of refractions, from its 
giving the obliquity of the ecliptic the same at the two 
solstices. 
