306 Dr. H. Herwig's Investigations on the Conformity 



tation, and the parallel to the axis of abscissa? MN the coefficient 

 of dilatation of an ideal gas. The curve, fig. 4, drops from a 

 value which is larger than that for an ideal gas, down to this 

 value, and in its further course arrives at still smaller values. 



In the same way, from the particulars of the curve i ) l (fig. 3), 

 it follows that the coefficient of dilatation of a vapour under a 

 constant pressure may depend on the temperature in the manner 

 shown in fig. 5, which is arranged after the fashion of fig. 4. 

 The curve (fig. 5) crosses the line which represents the coefficient 

 of dilatation of an ideal gas, so that, starting from greater values, 

 it meets this line, further on it arrives at a minimum lying 

 under it, then rises to a maximum which lies over it, and finally, 

 from a certain high temperature, runs into and along with it. 



Since the last part of the curve in fig. 4 lies always below the 

 line of the coefficient of gas, while the last part of the curve in fig. 5 

 coincides with this line, consequently, for such vapours as those of 

 chloroform and sulphuret of carbon, the coefficient of dilatation 

 at and from a certain high temperature is much smaller for a 

 constant volume than for a constant pressure, a property which 

 reminds us of Regnault's experiments on the so-called permanent 

 gases. 



On looking back, I find that some other observations besides 

 those here communicated on the coefficient of dilatation under a 

 constant pressure appear to point to such a behaviour of this 

 coefficient as is represented in fig. 5. That the curve has most 

 probably the maximum lying in the neighbourhood of B (fig. 5) 

 has been observed by Deville and Troost* in the case of vapour 

 of hyponitric acid under a pressure of one atmosphere. Any way 

 the want of any other explanation can no longer make it neces- 

 sary to assume a dissociation of hyponitric acid. Of course in case 

 such relations be assumed for hyponitric acid as I have found 

 for chloroform and bisulphide of carbon, and there also the 



. PV 



validity of the relation =c V a + t be supposed, the constant 



pft 



c must have a much larger value than *0595, since the den- 

 sity found at 26°- 7 of the vapour when all but saturated dif- 

 fers considerably more from the final vapour-density than it 

 would on the assumption of the constant c=*0595. I will not 

 make any assertion respecting this case; I wish only to suggest 

 the possibility of such an explanation of the results of Deville 

 and Troost. 



On the other hand, we may perhaps deduce the existence of 



the minimum of the curve in fig. 5 which lies at A, from Hirn's 



researches quoted in §1. Hirn gives, amongst other examples, 



the specific volumes of superheated aqueous vapour under a 



* Comptes Rendus, vol. l.xiv. p. 237. 



