40 Dr. ©. Chree on the Bending of 
of the magnet and carriage causes a slight tilting. In con- 
sequence of this, the point of suspension of the deflected. 
magnet moves towards the deflecting magnet, thus reducing 
the horizontal distance between them. This compensating 
effect increases with the length of the suspension of the 
deflected magnet; it also depends on the pattern and mas- 
siveness of the magnetometer. The tilting is objectionable 
for several reasons, and suitable counterpoises exist in some 
instruments. In others, however, there is no regular counter- 
poise, the only equivalent being a thermometer, usually con- 
siderably lighter than the magnet and carriage, whose position 
varies according to the ideas of the observer, who may even 
put it on the same arm as the magnet. I have investigated 
the tilting effect in one or two cases, but refer to it at present 
only to put observers on their guard. If it exists, but is 
overlooked, the corrections deduced from a pure bending 
experiment are not strictly applicable. 
Since the publication of my first paper on the subject 
measurements of the bending effect have been made on over 
twenty magnetometers at the National Physical Laboratory; 
and it is now the regular practice to take the effect into 
account in framing the certificates, as these record lengths to 
0°001 cm. 
The earlier measurements were made by Mr. F. E. Smith, 
the bar being symmetrically loaded, and the consequent 
increase in distance between two carried points being read by 
microscopes. 7 
After regular work was commenced at Bushy House, 
the measuring-apparatus was transferred there, and for some 
time magnetometers were sent over from Kew to Bushy, so 
that the measurements might be made by Mr. Smith on the 
bars supported asin use. The transport of heavy magneto- 
meters being troublesome, and the degree of accuracy abso- 
lutely necessary being comparatively small, I devised the 
following method which enables the measurements to be taken 
at Kew. As actually carried out, the method makes no claim 
to the highest precision ; butit is, I believe, novel and capable 
of development, and it could easily be applied by the owner 
of any of the older magnetometers which have been verified 
at Kew, the great majority of which are probably situated 
where there is no ready access to physical laboratories. 
The magnetometer is set up exactly as in an ordinary 
deflexion experiment, with the deflecting magnet on its 
carriage at any convenient position on the deflexion-bar, and 
equal weights are hung up symmetrically, one on each arm. 
The consequent increase of distance between the two magnets 
