124 



Proceedings of the Royal Society of Victoria. 



telescope and a scale at some distance. I soon found that there 

 was no necessity for such a sensitive arrangement, as the needle 

 was constantly going off the scale. It was also necessary for me 

 to measure the resistance of the circuit each time it was altered, 

 i.e., each time a wire was fused or broken and had to be renewed. 

 This was troublesome and took up a lot of time, and I soon found 

 it more convenient to use a high resistance though less sensitive 

 galvanometer, which I arranged to give direct readings as a volt 

 meter and save the trouble of reducing the readings. The lamp 

 and scale was at a distance of forty inches from the concave 

 mirror on the needle which formed on the scale an image of a 

 lens with a dark vertical hair-line immediately in front of the 

 lamp, the lens serving to concentrate on the mirror a larger 

 amount of light from the lamp than it would otherwise have 

 received. The scale had 350 divisions on each side of zero. The 

 galvanometer resistance was 7.400 ohms; in series with this I 

 added 2,600 and another 90,000, which could be short circuited, so 

 that neglecting the resistance of the wires under observation! could 

 have a resistance of either 10,000 or 100,000 ohms. The galvano- 

 meter was al.so provided with three shunts ^, t^j-, and -^\^ of its 

 own resistance. Now, putting a Leclanche cell of 1*45 volts into 

 circuit with the 100,000 ohms and the ttV shunt I adjusted the 

 height of the control magnet till the deflection was 145 scale 

 divisions. With this arrangement the readings were always 

 very approximately in decimals of a volt whatever shunt was 

 used. Thus with 10,000 ohms and no shunt, 100 scale divisions 

 indicated -001 volt; with 100,000 ohms, -01 volt; with same 

 resistance and \ shunt, -1 volt ; thj shunt, 1 volt ; and with ^\^ 

 shunt, 10 volts, so that by adjusting two plugs the one instrument 

 would indicate '00001 volt and measure 35 volts. I found this 

 arrangement very satisfactory. For perfect accuracy the external 

 resistance should have been slightly different for each shunt, but 

 neglect of this caused an error of only two or three per cent., and 

 I aimed at quickly getting the magnitudes of the effects involved 

 rather than a very accurate measure of them. It may perhaps 

 be convenient to those who are not familiar with the magnitude 

 of the ordinary thermoelectric phenomenon to quote a few 

 figures, so that the relative amounts of ordinary thermoelectric 

 forces and those which I am describing may be readily compared. 



