736 PROFESSOR TAIT ON THERMAL AND ELECTRIC CONDUCTIVITY. 
tions. Several points must be thoroughly cleared up before more definite 
statements can be made. Meanwhile the MSS. of the whole work is placed at 
the disposal of the Society. 
§ 17. To determine the electric conductivity of the bars above described, I 
employed, in succession, three different methods. The results of these separate 
methods agreed with one another quite as well as did results by any one method 
made on different portions of the same bar. The German silver bar is the least 
uniform of the metallic bars, portions of it of 10 inches length at different parts 
varying through a range of as much as 5 per cent. in their conductivity. Slight 
defects in the casting, some of which are visible at the surface, of course easily 
account for this. I give the average value. 
Neither the absolute nor the relative electric conducting powers of these 
bars were found to agree at all well with those of wires (said to be of the same 
material) which were furnished to me along with them. Hence some of my 
earlier statements to the British Association (especially with regard to copper) 
were inaccurate. The fortunate circumstance that I had no wire said to be of 
the same material as the ForBEs iron bar, led me to test all the thick bars 
themselves for their electric conductivity. 
§ 18. The first process I employed was that described by Sir W. THomson 
(Proc. R. S. 1861). The principle of the method will be easily seen from the 
following diagram. 






A 
EI a WLLL ILL N ILL LLLLLL = 
ap EEE 
LLL LLL LLL LLL ETE 
eX 
The bars to be compared are placed parallel to one another, and connected by 
a small resistance D at one end, while the poles of a single cell E (sometimes 
short-circuited) are applied to the other ends for a period usually very short. 
Points A, A’, B, B’, are joined by resistances, s¢milarly divided in C, C’; and these 
latter points are connected with the terminals of a sensitive galvanometer whose 
coil has a resistance, large in comparison with that of any other part of the 
arrangement. 
Under these conditions, if 7 be the current in the battery, the current in the 
galvanometer coil is (to a sufficient approximation), 
a { aq—bp a a 
ne a+b +(agi-aeae } 
Here the resistances are AC=a, CA’=b, BC’=a, C’B’=B, AB=p, AB=G 

