118 Barrerr, Brown, anp HAaprisup—Researches on the 
water; from these mercury cups wires led to the galvanometer. A moving coil 
dead-beat reflecting galvanometer was employed with an extremely good concave 
mirror, whereby scale-readings of a fraction of a millimetre could be made. As 
the coil of the galvanometer had a high resistance, any error due to variation in 
the resistance of the leads was eliminated. The following is an extract from 
Mr. Edmonds’ report :— 7 
‘‘ After various trials, an effective way was found of securing good contact 
between the thermo-couples and the rod; six couples were employed, fixed at 
definite and exactly similar distances along the rod under experiment. Each 
thermo-couple was beforehand carefully calibrated by a standard mercurial 
thermometer ; within the limits of temperature used the calibration curve was a 
perfectly straight line; from this curve the temperature difference corresponding 
to any given deflection was read off. From the temperatures given by the couples 
fixed along the rod a curve was drawn for each rod, having excess of temperature, 
above the air as ordinates and distances from the source of heat as abscisse. 
From these curves the relative conductivities were obtained. A nearly pure 
copper rod, having an electrical conductivity 96 per cent. of Matthiessen’s 
standard pure copper, and corresponding in length and diameter to the specimens 
of iron alloys used, was employed as a standard of reference. ‘The absolute 
thermal conductivity of this copper standard was subsequently determined, 
both by the method of Forbes and by Angstrém’s method of the periodic flow 
of heat. The latter method involved much troublesome reduction of the obser- 
vations; the value found for the absolute conductivity of this specimen of pure 
copper in ©. G.S. units was 1-041. 
‘‘ Assuming the temperatures along any two rods to be given by 6= Ae’ + Be” 
and @= Ae” + Be", and the temperatures at points 7, 7+ /, +21, along one rod 
0,, 02, 93, and at the same points along the second rod @,’, 0,’, @;, then it can be 
shown that 
pe _ log. (a +7 = 1) 
Bw log. (n! + V ay: 
— 0, + 0; 
where n= 
: and == 
DG 2 Os 
‘Fyrom the equation for the flow of heat along the rod we have 
Le 
5 ale and an 
WC AGN aa 
‘“where Z is the surface emissivity of the first rod, 7 = its radius, A = its 
