8 
Mr. Christie on the effects of temperature on 
case. If, instead of expanding the fractions, we computed 
them in the form which they have in the equation (B), we 
could hardly be supposed to obtain the absolute values of F 
more nearly than this ; although in either case the relative 
values would be obtained to a much greater degree of accu- 
racy. In the observations which I made, the values of 
were seldom much less than 8o°, and in such cases the error 
would be considerably less. In an instance where <p was 
82° 37 ^ the value of F was 222.5630 M, employing the terms 
containing cos.'^ cp, and 222.5640 M, neglecting them. Seeing 
then that no sensible error would arise from neglecting these 
terms, I have in all cases made use of the equation (C), for 
determining the values of F. I now proceed to the experi-, 
ments which I made for this purpose. 
On this occasion I made use of the same compass which I 
had already used in the greater part of the observations 
detailed in my former paper, and distinguished there as 
No. 1 ; the magnets were also the same that I had used with 
this compass. The length of the needle is very accurately 
6 inches. In order to determine the distance between the 
points which I ought to consider as the poles of the needle, 
I fixed it at right angles to the meridian ; and bringing 
another needle, freely suspended, near to it, I moved the centre 
of this needle along a line parallel to the axis of the first, 
and noted the points opposite to which the axis of the second 
was exactly in the magnetic meridian ; these points I con- 
sidered as the poles of the first needle. The distance be- 
tween the points thus determined was 4.28 inches. 
In my former paper I have stated the length of each of the 
magnets to be 12 inches ; more accurately, the length of the 
