34 
MR. S. W. J. SMITH ON THE THERMOMAGNETIC ANALYSIS OF 
ordinary type. The resistances of the ring secondary and primary were measured by 
means of a previously calibrated Callendar and Griffiths bridge. The demagnetisa¬ 
tion circuit contained a shunted voltmeter similar to that used in the primary circuit 
(see below), and the current strength was varied by means of wire resistances and a 
zinc sulphate trough. The inductive effects in the secondary circuit were measured 
by the apparatus already mentioned in connection with the preliminary experiments. 
The primary currents were measured by a Weston voltmeter (reading to 5 volts) 
shunted by a resistance of 10 ohms, which was not heated appreciably by the 
currents traversing it. It was found, by calibration, that a deflection of 50 scale 
divisions on the voltmeter (30 divisions = 1 volt) corresponded very nearly to a 
current of 0*017 C.G.S. units in the primary circuit. The adjustment of the primary 
current to a constant value at different temperatures, notwithstanding variation in 
the resistance of the ring primary owing to temperature change, was accomplished by 
means of resistance boxes, and a stretched wire of variable length included in the 
primary circuit. 
§ 2. The temperature of the ring at the time of each observation of permeability 
was deduced from the resistance of the secondary. The method of deduction was to 
assume the relation 
R„ = R 0 (1 + ct 6 — b $ 2 ) 
between the resistance of the ring secondary and its temperature. The values of a 
and b were determined, in the case of the meteoric iron, by constructing a platinum 
thermometer out of wire taken from the same reel as the secondary, and in the other 
cases by constructing a thermometer out of the secondary coil itself, re-wound for 
that purpose. In each case the resistances of the thermometer at the freezing- and 
boiling-points of water, and at the boiling-point of sulphur, were taken in the usual 
way, a previously calibrated bridge being used for the purpose. 
Each of the three platinum thermometer observations, with each of the wires used, 
of corresponding values of temperature and resistance, was estimated to be measured 
correctly within O'l per cent. With this degree of accuracy the probable error in 
the estimation of a was about 1 per cent., and in the estimation of b about 15 per 
cent. The error in the estimation of a temperature of 700° C. would, under these 
circumstances, amount to about 5° C. The values of the higher temperatures given 
in the subsequent tables may therefore be regarded as subject to an error certainly 
less than 10° C. 
§ 3. In the first experiments the heating coil was placed between the primary and 
secondary coils, but in the later experiments the heating coil was wound outside the 
other two. The method of winding was as follows :—Two strips of mica were cut of 
such size that one could be bent into the form of a hollow cylinder fitting inside the 
ring, and the other into a cylinder fitting outside. The depth of these cylinders was 
slightly greater than that of the ring, and equidistant notches were made along 
