Pirsson—Complementary Rocks and Radial Dikes. 119 
is especially true in many cases of these systems of radiating 
ones. | 
If they were formed, so to speak, initially we should expect 
to find them filled with the same magma as that which forms 
the central stocks. In that case they would form apophyses 
from it. A study of them in several regions has shown, how- 
ever, that they are of later origin than the filling of the central 
stock, that they represent indeed the very latest phase of erup- 
tive activity in the district and that the rocks composing them 
are much more highly differentiated than those of the central 
masses. 
The proof of their later origin is not that they represent 
more highly differentiated material (though this would be indi- 
cative of it) but in the fact that they frequently possess extremely 
dense or even glassy saalbands although they are found in the 
immediate neighborhood of the central stock, cut through 
apophyses of it or are even found in the stock itself. 
It is in such systems of dikes that the most typical examples 
of complementary rocks, lamprophyres and oxyphyres are to be 
found. 
When a large mass of molten magma is intruded into a 
region of sedimentary beds, a vast amount of heat is spread out- 
ward among the strata in a zone surrounding it. Besides effect- 
ing contact metamorphism this heat must cause a considerable 
expansion in the surrounding material, and expansion which it 
may readily be conceived will seek re-adjustment in various 
ways. 
_ As the mass as a whole gradually cools, it must also contract 
and this contraction will not cause a reversal of the adjust- 
ment process but will seek relief in cracking. Oracks will thus 
occur not only in the cooling mass of igneous rock but in the 
heated zone of sedimentaries as well, and if these are homogene- 
ous we should expect the cracks to assume a more or less radial 
position. 
For the same reason one would expect these systems of 
cracks to be most typical where intrusions have taken place 
into areas of approximately homogeneous and undisturbed strata 
and to be obscure or even entirely wanting when the area has 
been greatly disturbed, faulted, etc., before intrusion and when 
it consists of dissimilar unhomogeneous rock masses. Such 
facts as have come under our observation tend to comfirm this 
view. 
If this hypothesis be admitted it is easy to see how more or 
less radial systems of dikes may arise which are much later than 
the central stocks they surround. 
It is generally considered at the present time by most petrol- 
ogists that the process of differentiation in molten magmas 
