f 156 ) 

 from A\ liic'li follows that — q^ being supposed to be podtlve — 

 the value of ( — ) can only be positive if — should be greater than 



unity. Let us therefore determine the limiting value of — . With 

 7^= 7\, x = 0, .v' = we may derive from the equations (2): 



T= T 



and we have 



1 + ^ (/J - /J') 

 9^ 



R2\ x' 

 1 H -log — 



92 *'o 



log — = 



.^■„ R\ T. 



Therefore the vahie of — remains smallei- than unity, and tlic 



meltingpoint-cur\'e continues to (Jcscenil, as long as we have : 



<l,fT. 



In the following we will always assume T^ > T^ or ^ — 1 



positive. The above condition will then the sooner be satisfied, accor- 

 ding as /3' in the solid phase has a /w/her positive value. Now 

 probably /? will nearly always have a very small {)Ositive value and ^' 

 a ratiier large positive value. The concUtion will therefore probably 

 be nearly always satisfied. If wo i»ut (> := 0, {hen we get simply: 



-^^^-9.^'<9.{j~^ 

 If /?' (or a') is positive, i. e. if heat is ahsorbed iu mixing the solid 



phase, then we shall ahnags have — <[ 1 and therefore the melting- 



point curve will always descend on the side of the /////Ar^s-/ temperature. 

 An initially ascending part and in connection with this the occurrence 

 of a niaximum-meltingtemperature is therefore (di)iost totally ex- 

 cluded. The possibility of a maximum exists only in the exceptional 

 and nearly inconceivable case, that ^' has a nuich smaller positive 

 value than /?, or even a negative value. 



If we determine ( — J at the side of the /of'/Ms-nemperature quite 



iji the same way, then we fmd, denoting 1 — x by y. 



