AND ITS TEMPERATURE CORRECTION. 77 
The first values of g shew that periods of a week are insufficient for very 
accurate determinations. 
The mean for 1843 is almost identical with that for 1844. 
19. The differences for three periods were also summed without regard to 
days, but paying attention to the sign of ¢—Z,,,. The following table contains 
the results. 
TABLE VI. 
Determination of the Temperature Correction for the Balance Magnetometer, 
regard being paid to the signs of the differences of temperature. 




Preceding temperatures greater than the Preceding temperatures less than the 
succeeding. succeeding. 
Value of 









PERIOD. 
sa@ | 2am | ¢ | 24@| 24a | ¢ 2 
1843. 2 Mic. Div. Mic. Div. e Mie. Div. Mie. Div. Mic. Diy. 
June 1—30. 3350.2 29096.6 8.68 10970.2 85550.3 7.80 8.24 
1844. 
May 9—24. 5404.3 37559.9 6.95 3011.1 29061.8 | 9.65 8.30 
Aug. 3—Sept. 6. 4726.1 | 34249.3 6.68 | 16970.8 | 137211.2 | 8.09 7.39 
For all the periods, | 13480.6 | 100905.8 7.49 30952.1 | 251823.3 | 8.14 7.813 

The result No. 16, and the mean results in Tables V. and VI., for the whole 
periods, agree very closely. As the value of one micrometer division, in parts of 
the whole vertical force, is about 0.00013, the greatest difference of the three 
final results, 7.83, 7.90, and 7.81, is 0.0000012. 
The final results, from five days’ observations, by the method of deflections, 
were .000085, .000077, .000079, .000062, .000073, differing 0.000023. 
The results, from the comparison of daily observations, in parts of the whole 
vertical force, will be about .000134, the time of vibration being about 9 seconds; 
if 11 seconds were adopted, the result would be .000095 ; in either case consider- 
ably more than the result obtained by deflections. 
20. The satisfactory determination of q for the Balance magnetometer, led me 
to determine the correction for the Bifilar magnetometer by the same method. 
Besides the variation of the magnetic moment, temperature also affects 
the length and interval of the suspending silver wires; it probably also affects 
their elasticity. 
The determination of the correction from the daily observations, at once sums 
up all the effects of temperature. When the suspending threads are of silk, these 
sources of error are avoided; but I conceive that much graver errors are intro- 
duced, due chiefly to varying humidity affecting the torsion of the thread. 
I shall give simply the results of the comparisons of the daily observations for 
the Bifilar magnetometer. 
VOL. XVI. PART I. U 
