﻿680 Prof. H. L. Callendar on the 



determination of the absolute expansion of mercury. The 

 remarks of Messrs. Scheel and Heuse compel us, in self 

 defence, to criticise these observations, and supply certain 

 details which they have omitted. In order to confine our 

 reply to a reasonable length, and to avoid thermometric 

 difficulties, our remarks will be limited to the expansion from 

 0° to 100° C. All differences will be expressed in terms of 

 the unit adopted by Messrs. Scheel and Keuse, namely, one 

 part in a million of the volume at 0° C. 



• Mercury Weight Thermometer Method. — This method was 

 applied by Thiesen, Scheel, and Sell between 0° and 100° C. 

 to five bulbs, two of Jena glass 16 m , two of Verre Dur, and 

 one of Jena glass 59 m . The cubical expansion of the bulbs 

 was deduced from observations of the linear expansion of 

 rods of the same glasses. The final results for the expansion 

 of mercury, as deduced from these bulbs, are printed in 

 heavy type near the end of their paper, and were as 

 follows: — 



?rona dilato meter 16 m No. 



1 



•0182327 



5 5 5 5 5 5 -^ O * 



2 



•0182091 



Verre Dur No. 



1 



•0181934 



No. 



3 



•0181941 



5.9™ 





•0182570 



The differences in the results, covering a range of 63 units 

 (neglecting the seventh decimal place as uncertain), greatly 

 exceed the possible errors of observation in so simple an 

 experiment, and may have arisen from three different causes: 

 (1) The bulbs and rods, though specially selected for the 

 experiment, may have differed in composition. (2) They 

 may have differed in expansion owning to difference of form 

 and treatment. (3) Even if the axial expansion of the bulbs 

 were the same as that of the rods, the radial expansion, which 

 is twice as important, might have been appreciably different. 

 The results from the 16 111 bulbs were discredited because 

 they differed by 23 units, whereas the verre dur bulbs were 

 in practical agreement, though much lower than either of 

 the others. There was nothing, however, to show that the 

 59 111 glass w T ould have given more concordant results than 

 the 16 m , if more than one bulb had been tested. It is hardly 

 fair now to lay special stress on the highest result in place of 

 the mean '018245 of the two highest results adopted in the 

 original paper. 



The result of Chappuis obtained with a verre dur bulb is 

 clearly entitled to much greater weight because it escapes 

 the first two objections above given, since the linear expansion 



