] INTRODUCTION TO THE MAKERSTOUN OBSERVATIONS, 1843. 
On three days the signs of 3 a (T) and of 3 (¢) are the same, namely, on January 23, 
March 22, and March 23 ; if the comparisons with these days be rejected, we have 
=A (T) = — 2533, 3 A (¢) = 4191, 2 A (y) = — 39:9 mic. div.; error of adopted coefficient = — 0-09 mic. div. 
In all the previous comparisons the signs of 3 4 (T) and ¥ a (¢) are the same ; in this 
series they differ, and the coefficient is unaltered.* 
70. In correcting the observations by this coefficient, no attention has been 
paid to the varying times of vibration in the vertical plane ; this, as has already 
been noticed, cannot be done. From the results in Table 16, it seems very doubt- 
ful if it should be done, as the correction deduced at various times does not seem to 
differ with the time of vibration. After correcting the observations made in the 
years 1844 and 1845 by the same coefficient, I was led, from the results, to suspect 
that some source of error still remained. Investigations for the temperature cor- 
rection at different times throughout these years have shewn me that the coefficient 
in micrometer divisions is not constant, and that it varies from some cause which I 
have not as yet determined ; certainly, however, it does not vary directly with the 
time of vibration in the vertical plane, as theory would shew (49, 51.), but rather 
inversely ; this, however, and other facts in connexion with the 6th conclusion (50.), 
will be considered in the Introduction to the Observations for 1844. It will be 
enough to mention at present, that the results for the value of q’, in 1844 and 1845, 
vary from 7-0 to 10-0 micrometer divisions ; by the usual method, at the end of 1845, 
it would not exceed 3-0 micrometer divisions. 
Bifilar Magnet. 
71. The value of the temperature coefficient for the bifilar magnet, determined 
by temperature experiments on two days, which gave very consistent results, is 
q = 0000304, 
See Table 19, Introduction, 1841-2 ; or if 
k = 0:00013, g = 2:34 scale divisions, 
this includes the theoretical correction for the expansion of the suspending silver 
wires and brass grooved wheels; it did not seem improbable, however, that other 
sources of error might exist. The results for the balance needle shewed that this 
might be determined by comparisons of the daily observations ; such comparisons 
have been made, and the following Tables contain the results. The differences of 
the daily sums of the bifilar magnetometer readings are indicated by 4 (#), the 
= A (2) 
=o as given in scale divisions whose value = 000013. 
2 
results 
* Tf other evidence of the accuracy of this method of determining the temperature coefficient should 
still be desired, I would refer to the coincidence of the results for the value of %, No. 51, and end of 
note, p. XXXVii. 
