OF THE NATIONAL PHYSICAL LABORATORY. 



351 



few hundred degrees, and that therefore an expression deduced empirically cannot be 

 applied outside the limits of actual experiment without risk of serious error. 



To obtain the most probable value over the whole range for the expansion of Berlin 

 porcelain, a combination was made of the values obtained directly by two independent 

 methods over the range to 100 C. by CHAPPUI.S and HARKKU with those obtained 

 for higher temperatures by HOLBOKX and DAY. As pointed out by CHAPPUIS in the 

 paper in the 'Philosophical Magazine' alluded to in the introduction, the formula of 

 HOLBORN and DAY undoubtedly gives too high values below -JoO" (.'. This has just 

 been confirmed by the experiments of SCHEEL, published in vol. 4 of the ' Wiss. Abb. 

 der Keichsanstalt,' whose results over the range I 4 to 100 obtained in the Fizeau 

 dilatometer are in close agreement with those of CHAPPUIS and HAKKEH. 



1 have therefore taken for this work the following values of the mean dilatation 

 between and T, and from them calculated a table giving the volume at T of a 

 reservoir, whose volume at (J J is unity, for temperatures up to 1.100 C. 



T. 







100 

 200 

 300 

 400 

 500 



whence volume at T of reservoir whose volume at I 



o 



100 

 1000 



1 000000 

 1-000897 



1-013834, ,tc. 



Vlll. Ptvwtre Coefficient of tin' Res 



In the work of CHAPPUIS and HAKKEH account was taken of the change of volume 

 of the bulb produced by changes of internal pressure. The value of the change of 

 volume was deduced from observations made by varying the external pressure on the 

 reservoir at ordinary temperatures. With the thick-walled reservoirs here employed 

 the whole effect is only very small, a change of 1 metre in the internal pressure 

 producing a change of about 4o~ooo *' the whole volume of the reservoir at ordinary 

 temperatures. The effect of high temperatures in changing the elastic constants of 

 the porcelain here involved being quite unknown, it was deemed justifiable to omit 

 this correction altogether for the present high -temperature experiments. 



