6 Messrs. Richardson, Nicol, and Parnell on the 
order to keep the temperature of the heated tube at all 
constant, it was necessary to protect it and the adjacent parts 
from draughts. This was done by building a cylindrical box 
of asbestos card around them; the position of the box is shown 
by the dotted lines in the figure. To enable the tube to be 
seen the asbestos was provided in front with a mica window. 
Much trouble was also caused at first by the gradual heating 
of the gas in the tubes FL MEF’. This was ultimately got 
rid of by wrapping the glass tubes F and F’ closely round 
with spirals of composition tubing, through which a constant 
stream of water was driven. Under these circumstances it 
was found that turning on the heating current produced a 
sudden expansion due to the heating of the gas actually 
inside the tube, but there was no gradual drift in the sulphuric- 
acid gauge due to the adjacent glass parts F F’ becoming 
slowly heated. 
The thermocouple was standardized directly by determining 
the melting-points of the. pure sulphates of potassium and 
sodium. To do this a minute quantity of the powdered salt 
was placed near the middle of the platinum tube and watched 
carefully through the mica window witha lens. The melting- 
point was quite sharp, and constant within the error of reading 
of the galvanometer. Two such determinations were made, 
one before and the other after the measurements had been 
carried out, and they agreed absolutely with each other. 
The melting-points were taken to be 1066%5 and 883°2 C. 
respectively *. 
In carrying out the experiments the apparatus was first 
filled with pure hydrogen. The caustic potash in the generator 
was first warmed, and a large volume of hydrogen allowed to 
blow off into the air to make sure that all the air was driven 
out of the generator. The apparatus was then exhausted, 
first with the water-pump and finally with the Toepler pump 
down to about =; mm. It was then filled slowly to 
atmospheric pressure with hydrogen from the generator. 
After standing some time the hydrogen was pumped out and 
new gas let in. This operation was repeated before the final 
filling took place. The barometer and the mercury-gauge in 
the apparatus were then read ; the difference of the readings 
gave the pressure of the hydrogen. 
A number of measurements of the rate of diffusion were 
taken at any one pressure corresponding to a series of 
temperatures. As a rule the temperatures were those given 
by deflexions of 80, 100, 120, 140, and 160 mms. on the 
* Cf. Heycock and Neville, Journ. Chem. Soe, vol. lxvii. p. 160.3 _, . 
