176 Holborn and Day — Gas Thermometer at 



remaining from the grinding, agree quite as well for porcelain 

 as for the metals. These observations were all made with a 

 rising temperature, however, and it is not impossible that dif- 

 ferent values would be obtained for the expansion of porcelain 

 if the temperatures observed were approached from the other 

 direction, i. e. in the order 1000°, 875°, 750°, etc. It also 

 remains to be investigated what differences exist between dif- 

 ferent bars of the same manufacture. 



The expansion of Berlin porcelain formed the subject of an 

 earlier investigation by Holborn and Wien,* who heated por- 

 celain plates 9 cm in length in a gas furnace and measured them 

 at the temperatures 550° and 1050° nearly. The mean value 

 then arrived at, 4:4:*10 -7 for the linear expansion coefficient, 

 agrees with the present value at the higher temperature within 

 the limits of accuracy then obtainable; at the lower tempera- 

 ture the earlier value is too large. 



More recently Bedfordf determined the expansion of French 

 porcelain (Bayeux) between 0° and 800°. His formula is 



X — [3425 1 + 1-01 1"] 10- 9 



A comparison with our results is not instructive, however, 

 for the expansions of porcelains from entirely different sources 

 are not properly comparable. 



Our measurements with the gas thermometer and porcelain 

 bulbs, for the calculation of which the Holborn and Wien value 

 was used, require only a slight correction for the new determi- 

 nation of the expansion, partly because the absolute expansion 

 is very small and partly because the use of too large a value for 

 the bulb expansion between 0° and 100° makes it also necessaiw 

 to correct (diminish) the coefficient of expansion of the gas 

 about ^-q. The required correction, therefore, remains less 

 than 1° for the whole range from 500° to 1000°. 



Jena Glass 59 lu . — For the measurement with the gas ther- 

 mometer below 500° a bulb of Jena glass 59 UI and a nitre bath 

 were used. We have, therefore, undertaken to measure the 

 expansion of this material up to 500°, but the results which 

 have so far been obtained only serve to emphasize how much 

 depends upon the way the glass is treated. We were able, 

 however, to establish the fact that the mean error of the gas 

 thermometer observations in this region, ± 0*5°, is not exceeded 

 in assuming for all temperatures up to 500° the constant value 

 18*10" 6 . For more accurate measurements it is certainly 

 advisable to employ the platin-iridium bulb for these tempera- 

 tures also, the expansion of which is much better defined. In 

 the nitre bath it will be necessary to protect it with a thin glass 

 tube, however, as the bath becomes alkaline with use. 



* L. Holborn and W. Wien, Wied. Ann., xlvii, 121, 1892. 

 f T. G. Bedford, Phil. Mag. V, xlix, 90, 1900. 



