80 



THE VOYAGE OF H.M.S. CHALLENGER. 



Fig. 17.— Tufa of Tristan da C'unha. 



Section of a polysynthetic crystal of 

 augite with straight outline corre- 

 sponding with Pec of one individual. 



outlines of these reentrant angles are replaced by a straight face, which restores these 

 broken lines, as is to be seen in fig. 17, showing a polysynthetic augite from this tufa. 



In the upper, most clearly developed, part of fig. 17, we 

 ought, considering the size of the polysynthetic lamellae, to 

 recognise the successive traces of the angles formed by the 

 juxtaposition of the twinned individuals; but we find only 

 one straight line whose direction corresponds to Poo of one 

 of the individuals. We often observe intercrystallisations 

 of augite and plagioclase ; sometimes the two minerals, 

 embedded the one in the other, have their vertical axes 

 parallel. The crystals of plagioclase, augite, olivine, and 

 magnetite are often of somewhat large dimensions. Those 

 of augite and plagioclase are corroded, and show the effects of the action of the base 

 which surrounds them. In this matrix we find the same minerals, but of much smaller 

 dimensions ; the small plagioclastic crystals sometimes assume the shape of rhombic 

 tables, often observed for the bytownite of recent eruptive rocks. 



As we have just seen, the superposed rocks that form the horizontal beds all belong 

 to the felspathic basalts, with vitreous matrix. Among the specimens which we have 

 examined, and which, according to Mr. Buchanan's notes, are to be regarded as 

 lavas, we find some that show certain peculiarities of structure. They are more 

 scoriaceous, but their mineralogical composition is the same. Among the scoriaceous 

 rocks there are some of dark-greyish colour, having their vesicles studded with zeolites ; 

 they contain crystals of augite measuring a centimetre. Under the microscope large 

 lamellar sections of plagioclase are seen, often twinned on the Carlsbad type ; two 

 simple twinned individuals give very different extinctions, 35° for one individual and 

 24° for the other, so that very probably we are dealing with a section parallel at once 

 to both P and x. In the thin sections the augite is dark green, with a yellowish tint 

 produced by incipient alteration ; apatite sometimes occurs as an inclusion in the 

 augite ; the preparations also show olivine, magnetite somewhat rarely, and scales of 

 hematite. These various minerals stand in a ground-mass in which are gathered very 

 minute microliths of plagioclase, augite, and magnetite, with almost no intervening 

 matrix. 



Other specimens of lava exhibit transition towards the pyroxenic andesites. These 

 rocks are compact, like the basaltic lavas described above ; their microscopic appearance 

 is identical, only we find no olivine in the preparation ; the constituent minerals are 

 plagioclase, augite, and magnetite, with the addition of biotite in small brownish 

 lamella?. These small crystals are all set in a matrix formed of faintly-coloured glass. 



Hornblende is rare in the lavas of Tristan, only one rock having been found to 

 yield it. This rock closely resembles the andesitic lavas in microscopical characters ; 



