352 
of the lujavrite body, in all places where the marginal decrease 
in the size of grains indicates that the magma has been in 
contact with a cooler wall rock. As a matter of fact, however, 
the naujaite nowhere occurs as a border zone to the lujavrite; 
and the whole form of the naujaite mass is in favour of the 
view that gravity has been the dominant factor controlling the 
distribution of the rocks. It seems, therefore, very probable, 
that the low specific gravity of sodalite has played an important 
part in the process. Indeed, as lujavrite has a specific gravity 
of 2:7—2'8, compared with the 2°2—2-3 of sodalite, it will be 
seen that even if the lujavritic magma has been considerably 
lighter than the solid lujavrite, it may yet have been heavier 
than sodalite. 
To sum up, the above discussion indicates that it is ex- 
tremely probable that naujaite and lujavrite are differentiation 
products of the same magma and have separated in situ. The 
cause of the differentiation may be referred partly to fractional 
crystallization — sodalite crystallizing at the upper border of 
the magma chamber —, and partly to the floating away of 
sodalite crystals from the heavier residual magma under the 
influence of gravity. Both causes have acted in the same di- 
rection, and it is, therefore, not possible to decide which was 
the more important. 
Conditions of this kind — fractional crystallization and 
gravitative separation acting with it — are unusual with most 
magmas, for nearly all minerals will no doubt tend to sink in 
any magma in which they grow. The co-operation of the two 
processes explains the considerable thickness (about 300 meters) 
of the naujaite layer. 
The structural differences between naujaite and lujavrite, 
and the character of their mutual contact have been discussed 
in earlier sections (pp. 322, 328). 
Relation of sodalite-foyaite to naujaite. — The coarse- 
grained sodalite-foyaite occurring as a sheet of varying thick- 
