A.— MATHEMATICAL AND PHYSICAL SCIENCES 33 



a control suitably imposed by gravity, by elastic torsion, or by an artificial 

 magnet, the daily variations of the earth's field, amounting to as much 

 as 35Y, make themselves manifest. Temperature changes also, by causing 

 variations of dimensions, elasticity or magnetic moment, disturb the 

 balance, and thus affect the observations to an extent corresponding to 

 15Y per degree centigrade. Moreover, where, as is most usually the case, 

 the method of mounting the indicating magnet is on knife edges, friction 

 adds uncertainty of the order of 5y. If we could escape the necessity of 

 applying the corrections which these important effects involve, we should 

 feel much safer in attaching significance to anomalies only a few times 

 larger than the limit of measurement of the apparatus. 



A year ago I thought I saw the way to do this, and brought the pro- 

 posed method to the notice of this Section. It was to make use of the 

 essential principle which gives to the Eotvos gravity balance its extra- 

 ordinary sensitivity, namely, to measure the space-variation only of the 

 forces in question. I found later that Eotvos himself had worked on these 

 lines, and actually constructed an instrument partially fulfilling the con- 

 ditions ; although it is not clear that he realised the full significance of 

 complete success. I have to confess that unexpected practical difficulties 

 of construction have so far prevented realisation, but I have not given up 

 hope that a magnetic instrument can be constructed to operate in the same 

 way as the proved gravity instrument. Accordingly it may be worth 

 while to indicate what a device of this kind might be expected to achieve. 

 Also, if I present the difficulties, perhaps someone more able than myself 

 may show how to surmount them. 



The chief virtue of such a magnetic torsion balance is that it would 

 discriminate between time-variatioji and space-variation of the earth's 

 magnetic field. The variation with time of a magnetic field remaining 

 spatially uniform would not affect it ; it would respond only to a sufficient 

 distortion in space. (Even if this distinction of space and time is repugnant 

 to relativity, it is practically of real importance.) Calculation shows that 

 with the magnets and suspending wires now available we could anticipate 

 an instrument which would be just about sensitive enough to respond, 

 in the average magnetic latitude, to the non-uniformity "^ of the earth's 

 main field. The additional lack of uniformity arising from diurnal 

 variations, or even magnetic storms, is by comparison small, because the 

 amplitude of the variations is only a small fraction of the total field, and 

 they are very widespread in character ; consequently they would fail to 

 disturb the instrument appreciably. We should therefore be able to 

 attribute the distortion observed solely to local magnetic features, apart 

 from a nearly negligible correction for general earth's magnetism. The 

 effect of changes of temperature also would be comparatively small, for 



' I.e. the non-uniformity implied by the change in the resultant intensity 

 from about 064 gauss vertically at the magnetic poles to about 0-32 gauss 

 horizontally at the magnetic equator. 



Unlike the gravitational ' horizontale Richtkraft ' of Eotvos, its magnetic 

 equivalent turns out to be zero everywhere on the surface of a uniformly mag- 

 netised earth. But the analogue of the gravity gradient, i.e. the northerly 

 gradient of the earth's vertical intensity, is zero at the magnetic poles only ; 

 it has its maximum value, about 1-5 x io~° gauss/cm., at the magnetic equator. 



