346 
Proceedings of the Royal Society of Edinburgh. [Sess. 
to be tested was in circuit with the battery B, the rheostat R x being 
adjusted so as to give the desired temperature. EF was connected in 
multiple arc with the galvanometer G. The resistance R 2 in the 
galvanometer circuit was one of 5000 ohms, its purpose being to reduce 
the deflection, and also to make the resistance so large that the current 
in EF would not be appreciably diminished by closing the galvanometer 
circuit. CD was a resistance in the circuit EF which was varied with 
the different wires ; it never differed much from the resistance of the 
wire which was under examination. By placing it in multiple arc with 
the galvanometer instead of EF, and comparing the deflections, the 
resistance of EF could be determined. 
With each wire preliminary experiments had to be carried out for the 
purpose of finding suitable weights to remain permanently in the pan to 
stretch the wire, and also what weights should be used to put on and take 
off so as to get a suitable elongation. If the stretching weight was too 
small the wire was not horizontal, and when the additional weight was put 
on the mark was drawn out of focus, and the reading could not be 
accurately made. 
For most of the determinations four sets of readings were taken, each 
set consisting of the changes in length produced by adding the elongation 
weight and then removing it, thus making the result the average of eight 
determinations. When the wire was heated, care was taken to see that 
it returned very closely to the original length when the weights were 
removed ; if there was much variation, it was concluded that the tempera- 
ture had altered, and the set was discarded. 
For each measurement, the microscope was set twice, and the average 
reading taken ; if the readings differed from one another by more than one- 
tenth of a division, more observations were made. The weights were placed 
in position a minute or two before the readings were taken, the aim being 
to make the conditions of determining the modulus as nearly identical in 
all cases as possible, so that the results at different temperatures might be 
strictly comparable. In order to bring it to the state of ease, each wire 
was stretched for about twenty-four hours by a weight somewhat greater 
than the maximum weight to be used in determining the modulus, before 
any observations were made. 
The procedure in each case was first to determine the modulus at the 
temperature of the room before any current had been passed, then to pass 
a weak current by which the temperature was raised through two or three 
degrees and again to determine the modulus. After this, the current was 
gradually increased and readings taken at various temperatures. When it 
