492 T. C. CHAMBERLIN 



such a tortuous circulation as the case in hand seems really to 

 involve. Inquiry should, however, at least be put on the right 

 track by recognizing the later aspects of science and the physical 

 realities of the case. 



It is at least safe to say that isolated crystals are habitually 

 formed within magmas, not merely on their surfaces. In addition 

 to this it is particularly important to recognize that the order of 

 formation of minerals in magmas is not that of their melting- 

 points, but rather singularly at variance with it. Some of the 

 minerals commonly formed earliest, as magnetite, apatite, and 

 ziroon, are higher in specific gravity than the average minerals 

 formed later, and these are generally higher than the liquid from 

 which they were separated. It is quite reasonable to suppose, as 

 leading petrologists do, that the heaviest order of minerals, at any 

 rate, if not the majority formed, would tend to sink through the 

 mutual solution. The actual effectiveness of this tendency must, 

 of course, be dependent on the viscosity of the magmas, the vigor 

 of the circulation, and other conditions. Whether the heavy 

 minerals would remain solid and collect at once at the center or 

 be redissolved in the depths and continue longer in the circulation 

 is doubtless to be left an open question for .the present. But this 

 and other questions are to be considered under the conditions of a 

 tortuous circulation rather than those of a quiescent liquid. The 

 tendency of the circulation must certainly have been to equalize 

 the temperature and to favor a slowly progressive precipitation 

 affecting large portions, if not all of the mass, rather than the mere 

 surface. The heavier precipitates might then rather plausibly be 

 assumed to collect where the combined effects of current and gravity 

 offered them the most available resting places. If so, a core shaped 

 to fit such conditions seems more probable than a strictly 

 symmetrical sphere. 



If we turn now to the other type of nuclear evolution — ^in which 

 the sifting action not only went to greater lengths, but the sifted 

 residue was much more affected relatively by motions inherited 

 from the expulsory action — it is well to recall at the outset that 

 - the range of cases stretches from the largest solid bodies notably 

 affected by the sifting process downward to the very borders of 



