400 
MR. CHRISTIE ON THE MAGNET1CAL OBSERVATIONS 
tures, we shall obtain the values of that is, the measures of the intensity of the 
needle II. corresponding- to these temperatures. The following- Table contains the 
7)1 
values of ^ thus deduced, and in the last column are given the differences in the 
7YI 
value of ^ corresponding to a change of one degree in the temperature of the needle 
No. II. 
Therm. II. 
Differences. 
Therm. N. 
Values of 
m 
~M' 
Differences. 
Difference in 
value of — 
M 
corresponding 
to difference of 
1° in Therm. II. 
47*0 
59*8 
72*1 
,83*7 
72*0 
60*3 
46*6 
O 
12*8 
12*3 
11*6 
11*7 
11*7 
13*7 
58*6 
59-8 
60*3 
61*5 
61*0 
60*9 
60*5 
•7784982 
•7743520 
•7691917 
•7655492 
•7694908 
•7740618 
•7784982 
•0041462 
•0051603 
•0036425 
•0039416 
•0045710 
•0044364 
Mean. . . . 
•0003239 
•0004195 
•0003140 
•0003369 
•0003907 
•0003238 
•0003515 
There are here, certainly, considerable discrepancies in the differences in the value of 
7)Z 
corresponding to a difference of 1° in the thermometer II. These, I consider, have 
arisen in a great measure, if not wholly, from the thermometer II. not indicating, in 
all cases, precisely the temperature of the needle No. II. : and this is one of the prin- 
cipal difficulties which occurs in an inquiry into the effects produced on the intensity 
of magnets by changes of temperature. From experiments which I long since made*, 
and from others which I have made more recently, I consider that the differences in 
the values of ^ corresponding to a change of 1° in the temperature of a magnet in- 
crease with the temperature ; but the foregoing results can scarcely be considered to 
indicate such a law. For the purpose, however, at present in view, this is not of 
great importance ; and with regard to the discrepancies I have noticed, I may 
remark, that if these have arisen from the cause I have assigned, their effect on the 
mean result will be extremely small ; for in this case the errors in the results arising 
from errors in the divisors would tend to destroy each other. In order that the 
results should be accurate, it is necessary, either that the temperature of the trial- 
needle should be the same throughout the observations, or that a correction should 
be applied for the changes which may have taken place in that temperature. This 
correction might be determined by observations similar to the preceding, but in this 
manner the final correction would form an infinite series. In these observations the 
* Philosophical Transactions, 1825, p. 63. 
