xlviii. InrropuctTion To THE MAKERSTOUN OBSERVATIONS, 1845 AND 1846. 
difficulty, the method already described for the bifilar magnetometer was first em- 
ployed for the determination of g’ the temperature coefficient in micrometer divi- 
sions: the details of several of these comparisons will be found, pages xlv., xlvi., 
xlvii., xlviii., and xlix., Introduction, 1843. It was found from these comparisons, 
Ist, That the value of g’, when a sufficient number of comparisons had been 
obtained, was independent of the interval between the days compared. 
2d, That the value of g’ remained the same after various adjistments of the 
needle ; the vertical screw for adjusting the sensibility never having been touched. 
3d, That the value of g’ has remained constant while the time of vibration ina 
vertical plane has varied from upwards of 11° to less than 6°; from which result it 
has been concluded that the value of & also has been constant. 
4th, That the value of q’ is the same, whether the differences of temperature of 
the magnet have been due to natural or artificial causes, and whether the differ- 
ences of temperature of the magnet have had the same sign or an opposite sign from 
those of the temperature of the air. 
78. From the 1st and 4th conclusions, it follows that the variations of the ver- 
tical component of the earth’s magnetism are independent of the temperature of the 
air and of the temperature of the soil.* 
79. The mean of all the results in the volume for 1843, Introduction, pages 
xlvii. and xlviii., gave 
g = 7:90 micrometer divisions ; 
and adopting the value of k, obtained from deflections, No. 59, 
q = 0000079. 
Which result is only is more than that obtained from the hot and cold water ex- 
periments: it appears in the case of the Makerstoun instrument that the errors of 
the usual methods are found chiefly in the determination of k ; this, however, is not 
always the case. 
The observations for 1843, 1844, 1845, and 1846, in micrometer divisions, have 
been corrected by the value 
g = 7:90 micrometer divisions. 
80. Since this value was obtained, several other determinations have been made, 
by comparisons of observations in 1844, 1845, and 1846; all the results obtained are 
given in the Table below ; several of the results obtained more lately have been de- 
duced from periods ill fitted to give a good value; the whole, however, have been 
given in order to shew the amount of error that may be expected in using bad series. 
In one or two of these cases the amount of disturbance has not been very consider- 
able, but the greatest variations of the daily mean vertical force have happened to 
yu 
* See foot-note, p. 895 of the volume for 1844. 
