182 SECTION PHYSICS. 



by suspending the bar-magnet as a bifilar magnetometer without any 

 correction, the zinc tubes being clamped at their proximate ends so as 

 to prevent any alteration in distance. Then, while the registration 

 is going on, the temperature on the magnet is gradually raised by a 

 water envelope, the water in which is gradually heated by a jet of gas 

 burnt outside the apparatus. This goes on for a period of about six 

 hours, and the heat is so arranged that the temperature of the water 

 should be raised gradually from that of the atmosphere up to about 

 90, which is as high as is necessary. The jet is then extinguished, 

 and the water allowed to cool down again, in about the same period 

 of time. It is found that the line which would represent the actual 

 magnetic variation during twelve hours is deflected in the direction of 

 diminished force as the temperature rises, and goes back again as it 

 cools down, and thus it is found that the register line will again return 

 to its normal position. It is quite clear, therefore, that these changes 

 are due entirely to temperature. The temperature having been 

 recorded at intervals of a quarter of an hour or any convenient time, 

 there are means of identifying these epochs with the register line when 

 removed from the cylinder and developed, by shutting out the light 

 for a very brief period, and then letting it in again, when you have 

 a little interruption in the line which identifies the line with the 

 particular time when you have produced that spot. By equating the 

 formula A t+ B t with the differences of the ordinates of the normal 

 and displaced register lines, at the times at which the temperatures 

 have been recorded, and reducing these equations by the method of 

 least squares, the values of the coefficients A and B may be obtained. 

 With regard to the bifilar magnetometer the coefficient A only is taken 

 account of. In the balanced magnetometer B is also given by the 

 relation of the bore of the thermometer to the weight of the mercury 

 in it, because it is evident that the finer the bore the further the 

 minute quantity of mercury will travel for each degree of temperature, 

 the more will it affect the balance, and that will determine the 

 coefficient of t\ 



The set of instruments constructed by the authorities at Kew is not 

 yet arranged, but it will be shortly ; the principle is exactly the same, 

 the registration being obtained by reflected light. There is, however, 

 one deviation from the system which I adopted which I am free to 



