134 Dr. C. Chree on the Law 
accord practically identically with the value *805 assigned to 
p by Borgen; but the mean from the last 13 instruments 
would answer to p='82. | 
From the last 18 pairs of Cooke magnets, I find 
mean value of P! (at 30 and 40 ems.) = + 8°00. 
Allowing a contribution of —*18 from Table X., we should 
infer from Table XII. that p slightly exceeds °86. 
From the last dozen pairs of Hlliott magnets, I find 
mean value of P’ (at 30 and 40 cms.) + 6°77. 
Allowing a contribution of —'18 from Table X., we deduce 
=°80. 
af For the Cambridge Instrument Co’s. magnets the average 
P! (at 30 and 40 ems.) has been —1°50; which allowing for 
the correction in Table X. answers to p=‘81. 
The above mean values accord pretty closely with Bérgen’s 
value *805 except in the case of the Cooke magnets. In 
their case the collimator has a central transverse groove 
which affords a grip for a screw determining the position of 
the magnet in its stirrup. This groove presumably exercises 
some influence on the pole-distance, and a very small influence 
would suffice to explain the observed high value of P!. The 
combination, for instance, of the value *815 in the p of the 
collimator magnet with °805 in that of the “ mirror” magnet 
would raise P’ to 8:0. 
§ 22. To test the influence of the magnetic moment, m, I 
divided each maker’s instruments into two equal groups, 
containing respectively the strongest and the weakest colli- 
mator magnets. The mean results appear in Table XVI. 
TasLE XVI. 
Cooke. | Dover, | Elliott. 
7 Moe | te Ppl |B 
Group with largest m ...... 977 pede 884 +588 | 1031 +6-04 
| x 2» smallest m ...| 907 48-24 788 4689 888 |+7-49 
| 
Nothing is known of the strength of the mirror magnets, 
but we may reasonably suppose that a strong mirror magnet 
is at least as likely to accompany a strong as a weak colli- 
mator magnet. Assuming the average mirror magnet: the 
same for the two groups, a larger P! means a larger p—i.e., 
an increased pole-distance—in the collimator magnet. Thus 
the results seem all in favour of Bérgen’s law that the pole 
