Porcelain ivitti Rise of Temperature. 91 



to determine how the expansion of porcelain varies with the 

 temperature. 



Direct determination s 7 of the linear expansion of porcelain 

 have been made by Deville and Troost (Comptes Rendus, lix. 

 p. 162. 1864), and by Holborn and Wien (Ann. Pliys. Chem. 

 xlvii. p. 107, 1892). In the experiments of Deville and 

 Troost on Bayeux porcelain the temperatures were measured 

 by means of an air-thermometer heated in the same enclosure 

 with the experimental rod, and having its bulb of the same 

 kind of porcelain. Headings were taken at 0° C. and again 

 at temperatures between 1000° and 1500°. The conclusion 

 reached was that the coefficient of cubical expansion is con- 

 stant, and between 000016 and '000017. In the experiments 

 of Holborn and Wien the expansion of a length of 9 cm. of 

 Berlin porcelain was measured, the temperature being deter- 

 mined by a thermo-electric junction in contact with the 

 porcelain. Measurements were taken at the temperature of 

 the room, and again at about 1000° in some experiments, at 

 about 500° in others. As the values found for the mean 

 coefficient for one of these intervals agreed as well with those 

 found for the other interval as they did amongst themselves, 

 these experimenters also concluded that the coefficient is 

 constant. The mean value of the coefficient of linear expan- 

 sion for the kind of porcelain used for their air-thermometers 

 is taken as '0000044. 



The method adopted in the experiments described in the 

 present paper was essentially the same as that employed in 

 the case of glass by Callendar (Phil. Trans. 1887 A, p. 167), 

 and by Callendar and Griffiths (Phil. Trans. 1891 A, 

 p. 123). 



A tube of glazed Bayeux porcelain about 1 metre long and 

 1*7 cm. external diameter was used. Two fine transverse 

 scratches ran round the tube at a distance of about 91 cm. 

 from each other ; and the distance between these marks was 

 directly compared at a series of temperatures from 0° C. to 

 830° C. with a standard length, by means of a pair of reading 

 microscopes. 



The mean temperature of the portion of the tube between 

 the marks was deduced from the resistance of a platinum 

 wire (doubled for the sake of obtaining a convenient value 

 for R 100 — P ) which passed straight down the tube from one 

 mark to the other and back again. The wire was fastened 

 down at intervals to a strip of mica which fitted the tube 

 fairly tightly. At the upper mark the wires were attached 

 to others of thicker platinum, which were themselves joined 

 at the pyrometer-head which closed this end of the tube to 



