MAGNETIC REGISTRA TZOZV. 181 



For this reason it became necessary either so to place the magnet 

 that it is not liable to changes of temperature, or else to provide 

 it with some means by which those changes can be compensated. 

 Unquestionably the best course is to place the magnets as is now done 

 in the Observatory at Greenwich, in an underground apartment, 

 which is liable to very small changes of temperature ; but if it be not 

 convenient to do that, it is then desirable to employ some means by 

 which the change of temperature can be compensated. In the bifilar 

 magnetometer that is accomplished by attaching to the magnet a rod 

 of glass to the ends of which two zinc tubes are clamped, and at the ends 

 of these near each other two hooks are placed to which the double 

 skein is attached. It is quite clear that owing to the greater expansion 

 of zinc over glass, heat will have the effect of approximating the hooks 

 by which the magnet is suspended by a minute quantity, and if they are 

 made to approach each other the torsion force is diminished. If the 

 interval between the hooks be diminished so as to diminish the torsion 

 force proportionate to the diminution of magnetic force by change of 

 temperature, it is quite clear that in that case the indications will be 

 unaffected by temperature : that is shown in the bifilar instrument 

 downstairs. In the balanced magnetometer the compensation is 

 effected in a different way : a thermometer tube is clamped to the 

 magnet, the bulb being on one side of the point of support and the end 

 of the thread of mercury in the closed tube somewhere on the other 

 side. It is quite clear that as the temperature rises and consequently 

 the thread of mercury travels along the tube, a minute quantity will be 

 transferred from one side of the balance to the other ; as the energy 

 of the bar diminishes a little additional mercury would be thrown over 

 from one side to the other, and the weight of mercury in the tube is 

 such that it just counteracts the diminishing force of the bar, so that in 

 spite of change of temperature, supposing the earth's magnetism 

 remained constant, the magnet will retain the same position. There 

 is a further advantage in this mode of compensation, which is this : 

 the amount of the temperature correction may be represented by a 

 formula of this kind (A t + B / 2 ), supposing / to be the number of 

 degrees of temperature at the time above the freezing point, and A and 

 B are coefficients which have to be determined. Those I have 

 determined in the instrument to which this correction has been applied 



