184 Dr. Harrison on Variation with Temperature of 
By connecting C! and M with a brass strip D, the current 
returns by the solenoid tube, and thus the direct action 
of the former on the magnetometer is toa great extent 
avoided. 
The heating-current, which was obtained from accumulators, 
was varied by means of a series of german-silver rheostats 
with plugs. The maximum current “used was 20 amperes, 
and this raised the platinum tube to a temperature of about 
1000° C. 
In order to avoid oxidation at temperatures above 500° 
the specimen must be kept in a high vacuum, and _ this 
necessitated a slight modification of the apparatus. When- 
ever the platinum changes length, there is a corresponding 
movement in the rod ©’, so that it was found impossible to 
seal up the apparatus before exhausting. 
To avoid this difficulty, a second and very “‘slack ” 
spring, S (made from a bundle of fine hard-drawn copper 
wires), is inserted between one end of the glass and the 
corresponding end of the brass solenoid tube ; thus any 
movement of C’ is taken up by 8, and the portion B of the 
lead passes out of a cork Q (which closes the solenoid tube) 
and is fixed by the binding-screw T. The joints are then 
sealed up with pitch and wax. 
Leads, ll', of very fine platinum wire leave the platinum 
tube at 5 cms. from each end, and pass out along capillary 
glass tubes G G, the ends of which are sealed round the wire. 
A mercury gauge, set up in the same reservoir as a 
barometer, indicates the pressure, and exhaustion is carried 
out by means of a mercury pump. 
A small hole (2) in one of the copper end- -pieces EH puts the 
interior of the platinum tube into communication with the 
space to be evacuated. The platinum-tube resistance is 
obtained by comparison with that of a standard manganin 
wire, 2 mms. in diameter, which is provided with two fine 
copper leads, one near each end, and is connected in series 
with the platinum. The manganin standard is immersed 
in oil and surrounded by a coil of lead tubing through which 
cold water can circulate ; with this arrangement, the temper- 
ature of the manganin did not vary more than half a degree 
during an experiment. 
The resistance of the platinum at various temperatures Is 
compared with that of the “standard” by tapping off part 
of the current, by means of the fine leads Jl’ and gg’, and 
putting each in turn in circuit with the magnetometer-coil C 
(fig. 1). The deflexions produced are proportional to the 
resistances of the portions AB, CD (fig. 3), provided that 
