CRYSTALLIZATION IN IGNEOUS ROCKS 183 
If a solution crystallizes, is a recurrence of the first component 
to solidify possible, after complete crystallization of the other 
dissolved substances? We must concede as true that a magma 
represents a solution at high temperature, and that its crystalliza- 
tion must be analogous to the crystallization of any solution. 
Chemical precipitation due to chemical reaction caused by con- 
centration, or by the elimination of some dissolved substance, may 
take place, but on the whole the analogy to a freezing salt solution 
should be striking. This has been pointed out repeatedly by 
various authors. 
For comparison, let us recall Usiglio’s classical experiments on 
the evaporation of sea-water. In this case we have a complex 
solution of such compounds as Fe,O,, CaCO;, CaSO,, NaCl, KCl, 
MgsO,, MgCl, etc. Pressure and temperature were kept constant, 
but the degree of concentration was varied. The following order 
of crystallization was: obtained: Fe,0,, CaCO,, CaSO,+2 HO, 
NaCl, MgSO,, MgCl, NaBr, KCl. There was an overlapping of 
the periods of crystal- i 
° : . CZs 
lization, but in no case’ ¢&, C0, 
was there a recurrence 
of the crystallization of 
any member after a later CN A 
product had formed (see HSL 
fig.1). When asalt solu- Hee 
tion freezes, ice crystal- Wait: 
lizes out. and the solu- Fic. 1.—Order of crystallization upon partial 
3 , evaporation of sea-water. 
tion becomes more and 
more concentrated until saturation is reached (23.6 per cent 
NaCl); when at a temperature of —22° C. both crystallize 
together, forming a cryohydrate. Salt does not crystallize out to 
be followed later by ice. Other examples of crystallization of salt 
solutions might be cited, but the above are sufficiently familiar and 
characteristic. In all cases (see Figs. 2, 3), however, no overlap 
of the first crystals formed can occur with regard to later products, 
as far as order of cessation of crystallization is concerned. In 
the majority of cases the end product should be a cryohydrate 
(eutectic). 
Ca$o f,0 
