OF CRYSTALS AND OTHER BAD CONDUCTORS. 
487 
By inspection, we have the divisor for converting E.M.F. observations to temperature 
difference, since in every case q + t 2 is known approximately to within one or two 
degrees. 
The value of a (1 + b + h) is seen from the above table to change only about 
7 per cent, for 100° C. of change of t z -j- t l ; the platinum-silver-copper element is 
therefore well suited for thermo-electric observations of temperature. 
The determination of the constants of the element was repeated every few days 
during the progress of the work. The value of a was found to diminish slowly, the 
total change during the three months during which the couple was used being 
about 1 "5 per cent. 
The temperature of the enclosure in which the divided bar was suspended was 
determined by means of a thermometer graduated in degrees, placed about 7 cms. 
above the top of the water-tank and about 15 cms. below the bar. The bulb of the 
thermometer was protected from radiation from the hot bar by a small sheet of paper? 
just large enough to screen the bulb. 
Determination of the Thermal Conductivity of the Brass Bar. 
The thermal conductivity of the bars was determined, following the method of 
Forbes, by two series of experiments, the first to determine the law connecting the 
rate of loss of heat from a heated surface, with the temperature of that surface, and 
the second to determine the law of steady distribution of temperature along a bar 
heated at one end, and exposed to the air along its entire length. These tw T o obser¬ 
vations furnish data for determining the thermal conductivity required. 
Cooling Experiments. 
It is customary in making these experiments to use a short length of the bar used 
in the steady experiments. I have, however, shown that the law of cooling is the 
same for bars of different material, but of the same cross-section and surface; and 
as a knowledge of the specific heat of the bar is required, it seemed better to make 
use of a bar of a material the specific heat of which is well known, rather than a 
brass bar in these experiments. As the values found by different observers for the 
specific heat of copper are practically identical, this metal was chosen. As it is a 
better conductor than brass, it also enables the assumption that the temperature 
throughout the bar is the same to be made without much chance of error. The bar 
was of the same diameter as the brass bar, l - 93 cm. and 26 cms. long. It had a 
polished nickel-plated surface. To make the cooling experiment correspond more 
closely to that of an infinitely long bar, the ends of the bar were covered with a layer 
of sulphur about 5 mms. thick. This was done by placing the bar vertical, wrapping 
a sheet of paper round the upper end so as to project about 1 cm. above the end, 
