RICHARDS AND MARK. — THERMAL EXPANSION OF HYDROGEN. 127 



This is very near the average value 0.000025 found by others with 

 soft glass of this kind. Therefore the volume of the two bulbs together 

 at 32.38° becomes 



305.174 (1 + 0.0000252 f) = 

 305.174 (1 + 0.000817) = 305.423 milliliters. 

 It is well known that glass when heated assumes au increase of volume 

 which is not at once given up on cooling. Thus the zero point of a 

 thermometer made of hard French glass is lowered for a time by 0.03° 

 after heating to 33°.* This corresponds to an increase of volume of only 

 one in two hundred thousand, and is therefore on the limit of the 

 accuracy of the above volume. Error from this source, slight as it might 

 be, was partly if not wholly eliminated by determining the volume under 

 conditions as nearly as possible like those used in the actual expansion 

 experiments. The small temperature interval of 32.4° used in the pres- 

 ent experiments is a very favorable circumstance, as regards this correc- 

 tion, because the error increases greatly with rising temperature. In a 

 research aiming at greater precision than the present one, the point 

 should receive careful consideration. 



Measurements of Temperature. 



The necessity of recording the temperature of the bath in units of the 

 international standard has already been pointed out. The temperature 

 of melting ice under atmospheric pressure is one of the bases of this 

 standard, and the position on it of the transition temperature of sodic 

 sulphate has been accurately determined by Richards and Wells as 

 32.383° on the hydrogen scale. f The temperature of the gas becomes, 

 then, a question of the purity of the substances used in the bath. 



The bath was filled with very clear pure washed commercial ice and 

 distilled water. After standing half an hour the water was replaced 

 with freshly distilled water which had been cooled to nearly 0° C. in Jena 

 glass. By this means very little impurity from melting ice was intro- 

 duced. As a matter of fact, careful comparison of the melting point of 

 this ice thus measured was found to be within a thousandth of a degree 

 of that of the purest ice made from boiled distilled water, wholly in 

 platinum. Moreover, the higher temperature was read in such a way as 

 to eliminate any error from this source, so that the results cannot be 

 affected by any impurity in the ice. 



Because ten kilograms of sodic sulphate was needed to keep the bath 



* Notice sur les Thermom. a mercur. Bureau Internat. Poids et Mes. (1896), p. 9. 

 t Proc. Amer. Acad., 38, 431 (1902). 



