of Action between Magnets. 135 
distance is greater the weaker the magnet. There are 
grounds, however, for some reserve of judgment. To account 
for the differences between the mean values of P’ in Table XVI. 
would require changes in p varying from ‘007 to fully -02. 
Now 01 seems the limiting departure from the mean allowed 
by Borgen for the weakest magnets, whereas the difference 
between the stronger and weaker magnets in Table XVI. is 
comparatively small. 
§ 23. Table XVI. gives but an inadequate idea of the 
apparent variability of P’ in different magnetometers. How 
far this apparent variability is real is somewhat doubtful, in 
view of observational and other uncertainties. We are, how- 
ever, on surer ground when we consider changes in the mean 
annual values found for P’ at a fixed observatory, where 
numerous deflexions have been taken with the same two 
magnets, at the same two distances, in a single magnetometer. 
At Kew, the same instrument and magnets have been in use 
for a generation. During the last twenty years, the extreme 
values found for P’ (at 30 and 40 ems.) from a year’s 
observations have been —1°899 and —0°603, the mean value 
tor the whole period being —1°238. The difference between 
the extremes might be accounted for by changes of ‘01 in 
opposite directions in the values of p for the two magnets. 
There is, however, no reason to think that either magnet has 
increased in strength, and whilst the moment of the collimator 
magnet has gradually diminished, it altered by less than 
2 per cent. between 1889 and 1892 when the extreme values 
of P’ presented themselves. If the cause lay with the mirror 
magnet, whose moment has not been recorded, its p must 
have altered nearly ‘02. As a matter of fact, the variation 
of P’ has been so irregular as to suggest that the principal 
cause must have been something other than change of 
magnetic moment. 
At Batavia Observatory, values were calculated for the P’ 
of a magnetometer from each two months’ observations, from 
1885 to 1893. A table recording the mean for the year, and 
the greatest and least of the 2-monthly values, appears in 
the Batavia “Mag. & Met. Observations” for 1893. The 
extreme 2-monthly values are —5:073 and +1°638, while 
the mean annual values vary from —0:920 to —3°786. The 
magnets are, I think, of English make, and the figures 
calculated here for the old Jones magnetometers will prob- 
ably apply fairly. The view expressed at Batavia is that 
“the values of the constant P...show rather large differences 
which are to be ascribed to variation in the momentum 
(magnetic moment) of the deflected magnet ; since 1891 this 
