[wheeler] CUBICAL EXPANSION OF VITREOUS QUARTZ 147 



than was expected was obtained, measurements were taken with a 

 second dilatometer so designed as to be easily filled under a vacuum 

 in order to exclude the possibility of the presence of any air bubbles 

 in the bulb. Finally, a third form of bulb, containing concentric 

 silica tubes sealed inside to test for the possible existence of some 

 surface effect, was employed. The close agreement between the re- 

 sults obtained with the three dilatometers indicated conclusively 

 that the unexpectedly low value obtained was not due to either of 

 these causes. 



Harlow observed no permanent change in the volume of the 

 bulbs through repeated heatings to 184°C, as indicated by the ice- 

 point readings. 



The effect of change of pressure and of the change in the elastic 

 constants of silica, due to change of temperature, was shown to be 

 negligible. 



Consequently, in view of the fact that the value of the linear 

 coefficient of expansion changes sign at about — 80°C, Harlow was 

 led to conclude that the mean value of the linear coefficient as given 

 by Kaye was too high, especially for the range 0° to 100°C. 



In the discussions which followed the presentation of Harlow's 

 paper to the London Physical Society, Callendar expressed the con- 

 viction that the cubical expansion could not be deduced from the 

 linear because of the impossibility of properly annealing a silica 

 bulb, owing to rapid devitrification at 1,000°C. 



As Scheel 1 pointed out, in a summary of the measurements 

 which had up to that time been made on the thermal expansion of 

 quartz glass, had Harlow used the mean of the coefficient of expan- 

 sion of mercury between 0° and 100°C. obtained by the use of dilato- 

 meters of verre dur by Chappuis 2 and by Thiesan, Scheel and Sell, 3 

 i.e., 182-57 x 10- 6 , he would have obtained 1-52 x 10- 6 as the co- 

 efficient of cubical expansion of fused silica, a value in good agreement 

 with that calculated from the linear expansion. 



In view of the possibility of the existence of some undetected 

 error in the measurements of the linear coefficient by the interference 

 method, as suggested by Harlow's work, Mr. H. Donaldson 4 of the 

 National Physical Laboratory made a determination of the coefficient 

 of linear expansion of the standard fused silica meter, which is in the 



1 ZS. f. Instrkde 32, pp. 14-18, 1912. 



2 Trav. et Mini, du Bureau Intl. des Poids et Mesures 13 C, p. 31, 1903. 



3 Wissensch Abhandl. d. Phys.-Techn. Reichsanstalt 4, p. 4, 1904. 



4 Proc. Phys. Soc. London 24, pp. 186-194, 1912; Chem. News 105, p. 178. 

 1912; Nat. Phys. Lab. Coll. Res. 9, pp. 181-190, 1913. 



