394 Proceedings of Royal Society of Edinburgh. [july 18, 
ness of the effect at 254° - 3 is pretty strong evidence that it vanishes 
at most a degree or two from the directly found neutral point. 
The numbers in Tables IV. and V. agree fairly well with the 
fact that the nickel and German silver lines are parallel up to at 
least 150° C., and that between its two bends the nickel line is 
straight. Above 250° C. the deflections of the FeArg thermopile 
cannot be taken as accurately measuring the small temperature 
differences, for the Arg line is no longer parallel to the iron line. 
The numbers have not been corrected for this. The very small 
correction due to the resistance of the measuring thermopile in- 
creasing with the temperature has been neglected. 
In conclusion, I must express my thanks to Professor Tait, in 
whose laboratory these investigations were carried out, for kindly 
placing at my disposal much of the necessary apparatus, as well as 
for his ever-ready advice. I also desire to express my thanks to 
Messrs J. T. Morrison and A. H. Mackenzie for their most valuable 
aid, and to Messrs Shand and Buchan for their kind help in the 
determination of the neutral points. 
{Added December 1887.) 
§ 1. Description of Apparatus. 
The above investigations were continued by the writer and Mr 
J. T. Morrison, at the laboratory of the former, near Londonderry. 
As the galvanometers were arranged so as to give much more delicate 
measurements than those in the preceding experiments, a few words 
of description are here necessary. The galvanometer connected with 
the measuring thermopile had a lens of about 10 feet focal length. 
Of the galvanometers used for measuring the battery current, the 
Helmholtz had a lens of 1 2 feet focal length, and the other mirror 
galvanometer one of 6 feet. The scales, each 1 metre in length, 
were of translucent paper, stretched between two boards whose 
edges were circular arcs of the proper radius. The divisions were 2 
millimetres each, and could be read to y^-ths. In order that the 
battery current might be more steady, the commutator was so 
arranged that the moment the current was broken it was immediately 
short-circuited through a similar resistance. For greater accuracy 
