( 229 ) 
being solved have already been hydrated to a certain extent in the 
latter case. In consequence of this Q, n can have become so strongly 
negative that it predominates over, the other positive caloric quantities, 
of which Qsl now relates to the anhydride. 
So, seen in this light, we may expect that if a hydrate splits olf 
much water at its transition point, a retrograding melting point line 
will arise, because then the probability that the change in the heat 
of melting is smaller than that in the heat of mixing, is very great. 
From the 7\X-figure it appears further, that the retrogression of 
the melting-point line of rhombic anhydride becomes less and less 
Proceedings Royal Acad. Amsterdam. Vol. XII. 
