The following refers to Magnet No. 55:-
299
Year.
(-)
An
} (~')
двор
Year.
| (8~4')
(4-4')
P,
!
1886,......
99.99 5.286 93.28 5.463
*"
"
1892,
97.20 4.948
60",68 4.977 1892, 60.00 4.924
1898,- 62.20 5.113 66.90 5.498
48".20 104.37 104.50 5.426
97..75
4.722
35".25
5.297
5.411
09 .62
5.794
63.75
5.299
»
5.068 73.87 6.112
We therefore have for magnet No. 55 μ =5.189±0.055 and μ, obtained at Kew in 1883 was 4.3. Similarly we have for magnet
5.377±0.093 while the value No. 83 5.151±0.084 and p, 5.480±0.151, while the value obtained at Kew was 4.962; and for No. 83A 6.160±0.084 and 6.482±0.151 while the value obtained at Kew was 6.047. The probable error of a single observed
На
is 0.146, and of a single μ, it is 0.262, so that the latter is about double the former because the vertical force is so small here. It will be seen from the table that a smaller value of the coefficient results from using smaller deflections, though the effect of this is so small that it may almost be attri- buted to chance. The induction coefficient of No. 55 shows no sign of change in course of sixteen years. No. 83 appears to have been made from the same steel tube, but No. 83A was made of another tube of different steel and shows larger temperature and induction coefficients. The values obtained at Kew, where the vertical force is large and where only vertical induction is observed, agree best with the values obtained at Hongkong, where the horizontal force is large, with the magnet hori- zontal. The differences between the values obtained in the two positions are larger than the probable errors of the differences and must be ascribed to the fact that the induced magnetism is not distributed over the magnet in the same manner as the permanent magnetism. Magnets destined for use near the magnetic poles ought therefore to be examined in the horizontal position at Kew.
The determination of P is very precarious and probably the difference between the results obtained with different magnetometers depends mainly upon errors made in the determinations of that constant which affects the third decimal of the horizontal force. Its determination is extremely liable to be vitiated by even small magnetic disturbances in horizontal force and still more by disturbances in declination. It does therefore not help matters much when P is separately determined each day, and the changes in P obtained in that way are evidently not to be trusted, even when the deflections have been observed with the greatest care, when the torsion has been reduced to a minimum, and precau- tions have been taken to protect the silk fibre from changes in humidity during the progress of the observations. According to Lamont's theory the lengths of the magnets are so selected as to make P very small, while no steps are taken to reduce Q, which being divided by the fourth power of the distance has an only minute effect on m: X. P and Q depend upon the "lengths of the magnets." If the inagnet loses more magnetism near its middle than elsewhere, this "length" increases and vice versa, but when the progressive decrease of the magnetic moment has become so steady as in case of No. 55 it is not likely that changes in its P constant cau amount to as much as the observations directly indicate. The effect of the P correction is very much decreased by observing at greater distances, for instance at 36 and 48, but at least in Hongkong even if i were kept as large as pos- sible, the angles of deflection are then so much decreased that the observations must be indefinitely multiplied in order to attain to any accuracy especially on account of small magnetic disturbances. On the other hand it is scarcely legitimate to approach the magnets as close to each other as 24 centimeters.
+
From observations made with No. 55 at 5 different distances between January 19 and February 7, 1899, it follows that the correction to the horizontal force obtained with distances 30 and 40 only (P+7.614) requires a correction of +.00011, while with 3 distances between January 3rd and 5th follows-.00018 (P 4.997). The correction to No. 83 from 5 distances, January 9 to 12, is +.00015 to the force obtained from 30 and 40 alone (P +8.91). In case of 83A there seems to be no correction.
From 7 comparisons between 83 and 55, each used on its own magnetometer, it follows that the horizontal force obtained with the former exceeds that with the latter by .00069±.00008, while from 3 comparisons between 83A and 55 the former exceeds the latter by .00035.00004. The former were compared throughout the year, the latter on only a few days, but the latter comparison was made