518 Mr. E. Buckingham on a 



It is clear that the order of discrepancy bears no relation 

 to the order of slope at any of the temperatures or pressures. 



With a theoretically "perfect^' gas, in the ideal state in 

 which its molecules occupy no space and exercise no attraction 

 on one another, the curve representing Pr/T should run 

 parallel to the pressure-axis. This condition is nearly satisfied 

 by the gases examined by M. Daniel Berthelot, all of which 

 were weighed at temperatures much above their critical 

 points. One would have imagined that at sufficiently low 

 pressures the distance between the molecules of vapours such 

 as those examined would be so great that neither molecular 

 attraction nor the size of the molecules would influence the 

 result ; and it might have been expected that the curves 

 showing variation of the value of Pz;/T with pressure should 

 become parallel to the pressure-axis at sufficiently low- pres- 

 sure. It is, of course, possible that at pressures lower than 

 40 millimetres — the low^est measured — the curves might 

 become parallel. A glance at figure 5 (PI. XX.) will show 

 this. But it is to be noticed that no such change in the slope 

 of the curve in extrapolating to zero pressure would have had 

 any important influence on the discrepancy. We have for all 

 these reasons been unable to establish any connexion between 

 the volatility of the substances examined and their divergence 

 from the '^ theoretical " molecular weights. 



The last supposition, viz. that it is possible that the atomic 

 weights of the elements may depend on the proportion in 

 which they are present in the compounds which contain 

 them, is added only for the sake of completeness. Even if it 

 be considered, there is no regularity in the cases examined 

 which would lend probability to the hypothesis. 



It must therefore be concluded that the determination of 

 the density of a vapour does not serve as a means of arriving 

 at a conclusion regarding the accurate atomic weights of the 

 elements present in the compound. 



LXI. On a Modification of the Plug Experiment. 

 By Edgar Buckingham"^. 



IN spite of various advances in our knowledge of the other 

 quantities concerned in the computation of the relation 

 of gas scales to the absolute scale of temperature, we are, as 

 regards the Joule-Thomson effect, almost exactly in the same 

 position as when Joule and Thomson finished their plug ex- 

 Deriments. Xo mathematical ingenuity can tell us anything 



* Communicated by the Author. 



