196 N. L. BOW EN 



according to the conditions under which it cools. This is equally 

 true of the complex reaction series that dominate the crystalliza- 

 tion of magmas an outhne of which, for the subalkahne magmas, 

 is suggested in Table II. As a result of the existence of the reaction 

 relations there indicated, it may come about that a magma pre- 

 cipitating olivine may, at a later stage, react with the olivine and 

 convert it into pyroxene, and, according to the extent to which 

 controlling conditions facilitate the reaction, the whole future 

 course of the magma is modified. If conditions are particularly 

 favorable the reaction may be complete and the olivine may dis- 

 appear, its place being taken by pyroxene. The liquid then cools 

 further with deposition of later members of the crystal series. If 

 conditions are not so favorable some ohvine may be left unchanged 

 and, in consequence, a somewhat different liquid is left to pass on 

 down the crystal sequence. So it is with the reaction, pyroxene- 

 amphibole. Not only that, but within each mineral group the 

 reactions may be variously facilitated under different conditions. 

 This is true not only of the members of the olivine-pyroxene- 

 amphibole series but also of the plagioclase series, and the liquid 

 may be entirely used up by reaction, sometimes earlier, sometimes 

 later. It need not be surprising, therefore, that the differentiates 

 derived from a gabbro magma may vary both in scope and quality. 

 We may have only gabbro and diorite in one case, but a longer 

 sequence including granite in another. We may have a potassic 

 granite in one case and a sodic granite in another. Rapid crystal- 

 lization at an early stage, with subsequent slowing-up, seems to 

 bind up much of the potash feldspar in the early formed plagio- 

 clase and give a sodic granite. On the other hand slow cooling 

 throughout seems to avoid this factor and give potash granites. 

 Too little is known as yet of the details of the various reaction 

 series to make possible any very definite statements on these points. 

 In offering an outhne of crystallization series in rocks (and 

 therefore of differentiation) the Hst has been carried only as far 

 as the constituents of granite. This should not be interpreted as 

 indicating that there is anything final about the granite with respect 

 to differentiation series. We have seen in the system KjSiOj- 

 Si02-H20 that when any mass is cooled to 200° and no farther 

 there will commonly be a liquid left over that never crystalKzes. 



