358 
different compounds alternately, a recurrent deposition of three 
different kinds of rock sheets might perhaps be produced. With 
the kakortokites supersaturation is extremely improbable, for all 
the sheets contain the same minerals, and the zgirine and 
the arfvedsonite, which were generally the last to crystallize, 
occur also as minute crystals inclosed in all the earlier minerals. 
Moreover, the succession of the sheets bears no apparent rela- 
tion to the order of crystallization. 
The question remains to be discussed whether the hypo- 
thesis of separation by gravity can account for the peculiarities 
of the banded kakortokite or not. This problem may be con- 
veniently considered under two heads. The differentiation of 
the kakortokite into white, red, and black sheets is readily ex- 
plained by the assumption of gravitative separation, but the 
recurrence of the sheets is not explicable in this way. 
That gravitative separation is able to account for the dif- 
ferentiation will appear from the following consideration. The 
black kakortokite sheets (sp. gr. 3°12) are characterized by the 
abundance of arfvedsonite (sp. gr. 3°4); the red sheets (2°85) 
which overlie the black ones abound in eudialyte (2°9); and the 
white rock (2°76) which covers the red sheets has alkali felspar 
as its dominant mineral (2-6). The arrangement, thus, agrees 
with what should be expected if it were due to gravitation. 
Further it must be remembered that the difference between the 
individual sheets depends almost exclusively upon the relative 
quantities of the different mineral components, and that each min- 
eral exhibits the same habit and the same size of grains through- 
out each set of sheets, and generally in contiguous sets also. 
Finally the microscopical examination of the kakortokitic rocks 
has shown a much less marked order of crystallization for the 
mineral components than is commonly the case with abys- 
sal rocks. It will, thus, be seen that if each set of kakor- 
tokite sheets (a white, a red, and a black) is considered sepa- 
rately, the observations tell decidedly in favour of the view 
