of Air- and Mercury-Thermometers. 213 



52°. Each of these observations I subjected to the following 

 process of computation. Adding 274 to the temperaturej I 

 divided the corresponding tension by the sum, and extracted 

 the sixth root of the quotient. This was laid off by convenient 

 scale as ordinate to the temperature. The points thus obtained 

 ranged distinctly in straight lines ; each mixture presenting a 

 gradient of density. These gradients were found all to converge 

 to one point, which was situated on the base-line upon which the 

 temperatures were marked off, and from which the ordinates were 

 measured {i. e. the axis of temperature). The position of this 

 central point corresponded to the temperature — 77°. 



M. Eegnault's observations on the tension of the vapour of 

 pure water below 100° being treated in the same way, presented 

 also^a straight line directed to the same point. [See Appendix, I.] 



The same system of projection was applied to observations on 

 the vapours of other liquids collected fi'om scientific journals and 

 publications. In all cases they seemed to range in straight lines, 

 some of which were parallel to others, and some converged 

 towards a point in the axis of temperature, e. g. the gradients of 

 sulphuric aether, cyanogen, and ammonia vapour, show parallelism 

 to that of water. Alcohol radiates from the same point on the 

 axis as water. 



§ 2. The chart containing the observations thus projected was 

 submitted to the Royal Society in 1852. It will not be neces- 

 sary to reproduce that chart here, as I shall show independently 

 how exactly M. Regnault's observations on steam conform to the 

 straight line ; and the further proof of what has been stated with 

 respect to the vapours of the mixtures of sulphuric acid and 

 water, may be obtained by inspecting the Table of comparative 

 tensions given by M. Regnault at pp. 179, 180, 181, vol. xv. 

 Ann. de Chim. The convergence of gradients of density ob- 

 viously implies a constancy in the ratios of tension at the same 

 temperature [see Appendix, II.], and in the above-mentioned 

 Table we have columns in which this ratio has been computed 

 for each degree. 



Beginning with the last mixture, in which there are 18 equiva- 

 lents of water, the ratio, it will be remarked, varies irregularly 

 between "828 and '84], and the greatest divergence from the 

 mean value is equal to an error of j-^oth of an inch of the column 

 of mercury. In the next, the ratio varies between "077 and "701 

 and the implied maximum error, about ^^oth of an inch. The 

 same slight divergence characterizes them all; and although the 

 first and second show a considerable diminution of the ratio at 

 the higher temperature, yet the observed tensions are so small, 

 that an error of not more than yg^th of an inch of mercury 

 would accouiit for the diflfcrcnce. 



