384 Mr. Christie on the diurnal deviations of the 
influence on the extent of the deviations, I did not consider 
that such changes of temperature as they were liable to would 
sensibly affect the energy of the magnets themselves. 
In the experiments of Canton, on the effects of temperature 
in increasing or diminishing the forces of magnets, the pre- 
cise change of temperature in the magnets could not very 
well be ascertained from the manner in which it was pro- 
duced, but it could not be less than 8o° or 90° on Fahrenheit’s 
scale, and yet this only caused the needle to move from N 
45° E to N 44^° E, when applied to a single magnet retaining 
the needle in this position ; and when two magnets were 
applied on contrary sides of the needle, so that it pointed due 
north between them, an increase of temperature to this ex- 
tent, in either of the magnets, only caused it to deviate 2|° 
towards the other. From this we might be led to suppose 
that in such observations as I had been engaged in making, 
the effects of small changes of temperature might be neglected 
without sensible error, and that even in extreme cases the 
deviations would not be materially affected : the latter, how- 
ever, I found was far from being the case. 
I had continued to observe in the open air under the same 
impression which I had when the compasses were in a room, 
but, on the day in question, the effects of a hot sun upon the 
magnets were too unequivocal to be doubted. So early as 
seven o’clock in the morning I had found a deviation of i3°qo { 
of the western point, and n°48' of the eastern, towards the 
north, from their situations at £ h 15® in the compass No. I . ; 
and in No. II, 6° 5 5' of the westerly and 4°20 / of the easterly 
point : whereas on the preceding day, the corresponding 
changes from the night before had been, for No. I, 2 ° 54 ' and 
