80 DRS. J. A, IIARKER AND P. C1IAPPU1S ON A 



The second table gives the values of 



for all the values of T in the comparisons. 



XXXIII. FILLING OF THE GAS THERMOMETER. 



The nitrogen employed was prepared by the following method ; into a solution of 

 100 grams of potassium bichromate in 900 grams of distilled water were introduced 

 100 grams of nitrite of soda and 100 grams of nitrate of ammonia. When gently 

 heated, this mixture gives off a very regular stream of nitrogen, which is collected 

 in a large bottle over distilled water. To destroy any oxides of nitrogen which the 

 gas may contain, it was passed through two tubes containing caustic potash, then over 

 copper, heated to dull redness in a combustion tube, and finally through a series of 

 drying tubes containing baryta and phosphoric anhydride. The gas, after remaining 

 a long time over the drying agents, was introduced into the reservoir of the gas 

 thermometer by a series of glass tubes, leading on the one hand to the tap on the 

 manometer-limb and on the other by a side-tube to the mercury pump. 



The reservoir was then heated for some time to about 250, being meanwhile 

 thoroughly exhausted by the mercury pump. Dry nitrogen was then admitted, and 

 the alternate evacuation and filling with gas were repeated several times. Our first 

 definite filling was made on February 2, 1898. 



As the comparisons were to extend between the limits 100 and 200 the initial 

 pressure at C. was adjusted to be approximately 800 millims. of mercury, the 

 pressure at 200 corresponding to this being about 1,387 millims. This is nearly the 

 highest pressure which can be measured on the manometer. 



XXXIV. DETERMINATION OF THE INITIAL PRESSURE. 



It is essential to measure repeatedly the pressure of the gas at the temperature 01 

 melting ice, in order to make sure that no leakage takes place at the joints, and to be 

 in a position to take into account the inevitable small variations in capacity which 

 take place when a glass reservoir is employed. As we have previously mentioned, 

 prolonged heating produces a permanent contraction of the glass, therefore we may 

 expect an increase in the initial pressure after the comparisons at high temperatures. 



For observation of the initial pressure the zero apparatus, previously described, is 

 used. We give, as an example of a determination, the second series of observations 

 of May 3, with their reductions, in the form adopted throughout the whole of the 

 experiments. 



