350 
sodalite will give a rock of the same composition as naujaite (see 
table p. 351). It is true that the values which have been cal- 
culated according to the supposed process of differentiation 
differ a little from the average composition of the naujaite as 
calculated from the analyses. But as all the rocks contain 
varying quantities of alteration products, mainly zeolites, the 
agreement may be regarded as close enough. 
It will thus be seen that the chemical as well as the min- 
eralogical composition of naujaite is sufficiently explained by 
the hypothesis that this rock has originated from the agpaitic 
magma by a differentiation set up by the local concentration of 
the first crystallized mineral (sodalite). 
In the calculation given in the table (next page) the start- 
ing point is the average composition of the entire agpaite mass 
which is regarded as made up of the mixture; 
Sie NE ob IK 
the letters signifying respectively sodalite-foyaite, naujaite, lu- 
javrite, and kakortokite (comp. p. 344). It might perhaps be 
more correct to deduce S before calculating the mean com- 
position, but since in this particular case S has approxi- 
mately the mean composition of the agpaite mass, and forms 
only one tenth of the whole, the result will be practically the 
same. 
‘As the calculated composition of the ‘mean agpaite’ de- 
pends mainly upon the naujaite and the lujavrite, it follows that 
the hypothesis set forth in the preceding paragraphs will ac- 
count for the mean composition of the lujavrites (including ka- 
kortokites) just as well as it does for the naujaite. 
Passing now to examine the relative position of naujaite 
and lujavrite from the point of view of the supposed differen- 
tiation process, we find that the evidence agrees with the sup- 
position of differentiation in situ. The differentiation, if due to 
fractional crystallization, must give rise to an upper layer 
