( 2:14 ) 



Two cases, however, are possible, if beginnino' with ?> =^ c6 we 

 diniiuisli the volume gradually. 



When diminishing t;, the value of v' — v may either decrease 

 continually down to 0, or it may first increase to a certain maxi- 

 mum, and then decrease. Which of the two cases occurs will depend 



on the value — . If we put by first approximation 



..■-, = (.'-*--)(, + -^) 



or 



a' {a'— 2 b) 



V — V 



= a' — h + 



we see, that when o' > 2 /*, v' — v will begin to increase and vice 



7), 16 27 



versa. The condition n > 2 i leads to — > • When T^ — T^.^ 



v' — V will begin to increase and vice versa. The condition for the 



27 

 limiting value of v' — v being positive, is «' > Z> or T^—Tt. 



Our result is therefore that for half of the temperatures, at which 

 V — Ü is positive, this quantity will begin to increase still more. 



This result will be modified to some extent, if we take also the 

 variability of h with the volume into account, — for this particular 

 problem it is sufficient to know the first correction term — but as 

 long as the accurate value of b is not known, it may be better to 

 consider b as invariable — , also because the laws found in this 

 supposition are generally much simpler. Not before a sufficient 

 number of reliable observations are at our disposal, at which the 

 result of this and similar calculations may be tested, it will be 

 possible to investigate, whether this and the following calculated 

 correction-terms of b can account for the deviations. 



If we consider b as independent of the volume, we find by diffe- 

 rentiating v' — V. that a maximum is found at 



b a' — 2h 



In order to find v positive, a' must be greater than 2 b^ which agrees 

 with the preceding result. 



