A. Holmes & H. F. Harwood — Picrite, Mozambique. 155 



"Washington has recently pointed out the general sympathetic 

 relation between potash and magnesia, and between soda and iron 

 oxides in igneous rocks. 1 The former relation is exemplified with 

 one exception (C) by the analyses quoted : — 



indicating that the Ampwihi picrite is not different from its 

 analogues in this respect. 



The soda-iron relationship does not hold within the narrow limits 

 of the five analyses A-E. If soda and potash be compared with the 

 MgO/FeO atomic ratio, however, it will be noticed that with one 

 exception in each case potash increases, while soda decreases, with 

 the ratio. 



It may not appear that there can be any meaning in such results 

 as these, drawn as they are from rocks without apparent genetic 

 relations in either time or place. Such rocks may, neverthless, have 

 genetic relations in virtue of the processes by which they were 

 formed. By comparing similar igneous rocks, correspondences and 

 discrepancies of the kind to which attention has been drawn may 

 come to be used, when their significance is understood, to suggest the 

 origin and differentiation of the magmas from which the rocks 

 have crystallized. It is possible that magnesia-rich magmas do not 

 readily part by crystallization with their potash, and that they may 

 selectively absorb and accumulate j)otash from the rocks through 

 which they pass in approaching the surface. Similarly, in proportion 

 to their iron content, magmas may not readily part by crystallization 

 with their soda, and they may selectively absorb and accumulate soda 

 from the rocks through which they pass on their upward or lateral 

 journeys. This digression has led us far from the picrite dyke of the 

 Ampwihi River, to which we must now return to pick up afresh the 

 lines of thought suggested by analytical results. 



The radium content of the rock, only - 44" 12 grams per gram, is 

 very low. Peridotites appear to average more than this amount, 

 a composite analysis of ten varieties giving 0*51 XlO -12 grams per 

 gram. 2 Two dunites analysed for radium by Professor Strutt, 

 however, gave - 33~ 12 and 0-34 -12 grams per gram respectively. 3 

 The slightly higher result for the picrite is probably due to the 

 presence of felspathic constituents, which are generally far richer 

 in radium than olivine or enstatite. The result is of more than 

 numerical interest merely, for it shows that the dyke cannot have 



1 Proc. Nat. Acad. Sci., p. 574, 1915. 



2 A. Holmes, Science Progress, 1914, No. 33, p. 16. 



3 R. J. Strutt, Proc. Roy. Soc, 1906, A. 77, p. 479. 



