DISCUSSION OF ROCK ANALYSES 401 



biotite and is coarser grained. The prominence given to the biotite 

 by Williams is apparently due chiefly to its abundance in the altered 

 form of protovermiculite in the soil of the " Cove," especially at the 

 " lodestone bed." Judging from the large size of the protovermiculite 

 plates and of the masses of magnetite and schorlomite found here, it 

 seems very probable that they are derived from a much coarser grained 

 and more basic rock, more nearly allied to the jacupirangites. A deep 

 boring at this point would be of the greatest interest. 



VI* is of Williams' " eleolite-mica-syenite (dark Cove type) " from the 

 small area northeast of the Cove. It is a dark brown and very coarse 

 grained rock, composed for the most part of a violet brown augite (ev- 

 idently titaniferous), some brown biotite (often in poikilitic crystals in- 

 closing augite and magnetite), with magnetite and some interstitial 

 nepheline. Hornblende, olivine, and feldspars % are wanting. 



This rock varies much in the relative amounts of augite and magne- 

 tite, the latter sometimes being present in large quantity. It is evidently 

 closely allied to the pyroxenites, and seems to be identical, or almost so 

 (except that it is not schistose), with the jacupirangites of Brazil de- 

 scribed by Derby ,f so that it may be called by that name. From the 

 minerals found at the central "lodestone bed" it is probable (and is 

 here assumed) that this analysis represents fairly the composition of the 

 central patch of the Magnet Cove complex. 



GENERAL MINERALOGICAL FEATURES 



It is evident that mineralogically these types are all closely related and 

 grade into one another. Nepheline is present in all — abundantly in the 

 intermediate types, but more sparingly in the most acid and most basic 

 extremes. Melanite, or a brown garnet, is also very common. It may 

 be mentioned incidentally that from the low Ti0 2 found in III it is 

 probable that the abundant garnet in this rock is not a schorlomite or 

 iiwaarite rich in Ti0 2 , but a more normal andradite. This is confirmed 

 by an analysis of garnet from Magnet Cove by Stromeyer,J which only 

 showed about 3 per cent of Ti0 2 . The more basic iolites are, however, 

 richer in titaniferous melanite or schorlomite. 



The prevailing ferromagnesian mineral is pyroxene, which varies from 

 segirine and segirine-augite in the syenites, through diopside, to a titanif- 

 erous augite in the most basic rock. Hornblende seems to be quite 



* The analysis is rather unsatisfactory, inasmuch as there is an abnormally large amount of CaO, 

 giving an excess above that needed to form augite. A new analysis of this type is highly desir- 

 able, but any probable change in the results would not affect the general discussion. 



f Derby : Am. Jpur. Sci., vol. xli, 1891, p. 314. 



J Quoted in Hintze. Mineralogie, vol. ii, no. Ixiv, p. 91. 



