414 ANNUAL, REPORT SMITHSONIAN INSTITUTION, 1910. 



effect in the opposite sense would require a pressure of about 50 

 atmospheres. The experiment shows that neither the adhesion to 

 the glass nor the cohesion of the water is less than 50 atmospheres. 

 Berthelot's experiment has been variously misquoted (1) with regard 

 to the dilatation observed and (2) as to the effect of dissolved air 

 on the tensile strength of water. * * * 



Although a priori there seemed no reason to suspect that the pres- 

 ence of dissolved air would weaken the tensile strength of water, Dr. 

 Joly and the author ^ considered it necessary to investigate the 

 point specially. We used a cylindrical glass vessel with rounded 

 ends and provided at one end with a narrow tubulure. This vessel 

 was very carefully cleansed by washing it internally successively 

 with caustic potash solution, dilute acid, and distilled water. Half 

 filled with water, it was boiled for some time to make sure that the 

 walls were thoroughly wetted; then it was almost completely filled 

 with water which had been previously boiled to get rid of undis- 

 solved air and thoroughly to wet all dust particles which might have 

 been contained in the liquid. By subsequent exposure to air this 

 water was allowed to become saturated with dissolved air. During 

 exposure care was taken to shield the water from dust, which might 

 not have been completely wetted or which might have introduced 

 small bubbles. To fill the vessel a small quantity of water in it was 

 raised to ebullition, and, while steam was issuing from the attenuated 

 tubulure, the latter was submerged in the dust-free water. As the 

 steam within condensed and the vessel cooled, the latter became com- 

 pletely filled with water. A small bubble was then introduced and 

 the vessel was closed by sealing off the tubulure. 



If the vessel was then cautiously heated, the water expanded more 

 than its glass envelope and the air bubble was compressed. The 

 bubble became smaller and smaller as the temperature rose and the 

 contained gas was forced into solution. When the bubble had 

 reached very small dimensions and was about to disappear great 

 care had to be exercised in the further application of heat ; for if the 

 water expanded too much and strained the glass beyond its elastic 

 limit, the whole experiment was rendered abortive by the breaking of 

 the glass. But if the heating process had been carried on success- 

 fully and all the air had been dissolved so that the water had been 

 made to completely fill the vessel without breaking it, heating was 

 stopped and the water ceased to expand. 



At this moment the water in the vessel was either in compression, 

 being constrained by a tension in the glass walls, or it was quite un- 

 constrained, just exactly filling the envelope, and neither suffering 

 compression nor causing tension in the walls. As soon as cooling 



1 H. H. Dixon and J. Joly, On the Ascent of Sap. Phil. Trans. Roy Soc. London, vol. 

 186 (1895), B, pp. 568 et seq. 



