xliv INTRODUCTION TO THE MAKERSTOUN OBSERVATIONS, 1845 AND 1846. 
Bifilar Magnet. 
65, The temperature coefficient for this magnet was also determined by means 
of hot and cold water experiments, Nov. 9 and 10, 1843. See page xli., Introdue- 
tion, 1841-2, for the details. The whole number of results was 30: the mean gave 
The correction for 1° Fahr., Q = 0-000294. 
66. If the 27th and 28th results (counting from the top of the last column of 
Table 19, p. xli., Introduction, 1841-2) be rejected, as it is believed that the great 
difference of both from the mean was probably due to one error in reading, we find 
the average difference of the 28 results from the mean = 0-000021, and the probable 
error of a single result was therefore about 0-000017. | Combining the results from 
high temperatures together, and similarly for those from low temperatures, we find 
The mean of 15 results for the mean temperature 48°:0 gave ( = 0:000292 
Mhejmean of WS. coweedeectoneas - scomeceeaerrereaas GSS 7a tas. 2- Q = 0-:000295 
so that for the bifilar magnet, also, the temperature coefficient is constant within the 
ordinary temperatures of 32° to 80° Fahr. 
67. The correction for the expansion of the silver wires and brass grooved 
wheel, =0°000010, being added to the value of Q above, we have 
The temperature correction for 1° Fahr., from hot and cold water experiments, g = 0:000304. 
68. As the observations in connection with the balance needle had shewn that 
there might exist variations due to temperature, other than those due to the varia- 
tion of the magnetic moment of the magnet, such as the varying elasticity of the 
suspending wire of the bifilar magnet, the temperature coefficient was determined 
in the following manner, which had at first been found to give consistent results for 
the balance needle. 
69. A series of days being selected in which the magnetic irregularities are 
small, and in which the variations of temperature are as considerable as possible, if 
we compare the mean instrumental readings for any two days, and if a R be the 
difference in scale divisions, this difference is due to change of temperature of the 
magnet, and to change of the horizontal component of the earth’s magnetism, let the 
portion of change of reading due to the former = A, and to the latter = a X, so that 
AR=A4+AX. 
If the difference of the mean temperatures of the magnet for the same two days 
be 4 ¢, then the correction for 1° of temperature in scale divisions 
