353 
ness (2—150 meters) immediately above the naujaite, seems at 
first sight to present characters which are incompatible with 
the hypothesis advanced in the preceding paragraphs. The che- 
mical composition of sodalite-foyaite is that of undifferentiated 
agpaite, with some slight enrichment in sodalite (see table p. 
344). The rock, accordingly, is heavier than the naujaite (2°65 
as against 2°55). The sodalite contents is considerably lower 
(about 15—20 per cent. as compared with 30—60 per cent. in 
the naujaite) and the sodalite is largely allotriomorphic. In its 
jower portion the sodalite-foyaite passes into naujaite slowly and 
gradually — the only perceptible change being a gradual increase 
in the number of idiomorphic sodalite crystals — and both rocks 
must evidently have been solidifying in their present position at 
about the same time. 
The difficulty in interpretation is this: — If sodalite be the 
first mineral of the agpaitic magma to crystallize, as indicated 
by the structure of the rock (the minute crystals of ægirine 
etc. inclosed in the sodalite crystals are here for the moment 
left out), and if the magma, as explained above, differentiate 
either by fractional crystallization or by the movement of cry- 
stals under gravity, then the uppermost layer of the agpaite 
should be naujaite and not sodalite-foyaite. Since the condi- 
tions of differentiation in this case belong to the early stages 
of crystallization, no assumption that sodalite-foyaite crystal- 
lized before differentiation commenced, can remove this dif- 
ficulty. 
A key to the explanation is probably found in the obser- 
vation, made during the field work, that the sodalite-foyaite 
shows a certain structural and mineralogical resemblance to the 
horizontal pegmatite veins or benches of the naujaite. 
As more fully discussed in the descriptive section, the 
naujaite possesses a peculiar suite of pegmatites. Besides a 
number of pegmatitic veins which occur like ordinary peg- 
matites (exceedingly large size of grains, varying width, irregular 
XXXVIII. 23 
