359 
that the differentiation is due to separation of crystals by 
gravity. 
Accepting this as a working hypothesis the recurrence can 
be more precisely understood. The complex now can be re- 
garded as made up of about forty layers which, originally, all 
alike have been each in its turn differentiated by a separation 
of its crystals under gravity so that each layer has given rise 
to three dissimilar sheets. Such a succession of processes may 
be accounted for in two different ways. 
(1). Each layer (of three sheets) may represent a separate 
intrusion. The objections we have raised in discussing the 
hypothesis of successive intrusions of the different rocks (‘in- 
trusion after differentiation’) disappear for the most part if in- 
trusion preceded differentiation. The remaining objections — 
absence of contact phenomena and gradual transitions between 
the sheets — may perhaps be met by the assumption that the 
intrusions followed upon one another rather quickly, and that 
the magma was not superheated, but it will be seen that the 
hypothesis of successive intrusions, applied in this way, simply 
expresses a particular mode of manifestation of the recurrence, 
and does not explain the recurrence in itself. 
(2). The entire kakortokite mass, in a tolerably homogeneous 
state, may have taken up its present position before consolida- 
tion. Outside influences may then have determined the re- 
current crystallization of a certain qauntity of magma. Thus 
crystals would be formed which would sink towards the bot- 
tom, and these when sinking might be to some extent sepa- 
rated according to specific gravity, after the manner of a 
coarse rock powder when treated with Txourer’s solution. This 
hypothesis of ‘intermittent crystallization’ does not explain the 
recurrence of the conditions which caused the crystallization, 
and in this the hypothesis is no more satisfactory than is the 
assumption of recurrent intrusion. It does however harmonize 
with the contact relations between the kakortokite and the 
