,02 DISCOVERY REPORTS 



is characteristic of the whole series, inckiding the dacite in which the nak ratio is only 62-6. Another 

 general feature is the low F' jM' ratio (less than 0-5) in the prevalent basic lavas. This is in agreement 

 with the Recent basalt lavas of the South Shetlands, but is in strong contrast with the equivalent 

 lavas of the Andes, in which the F' jM' ratio fluctuates round about unity. It is a remarkable and 

 perhaps significant fact that West Indian or Antillean lavas agree best with those of the South Sandwich 

 Islands in this respect (cols. D and F, Table 8). 



The dacite of Thule Island (Table 8, col. i), while of sodic type, does not compare very well with 

 the analogous santorinites of Deception Island (this Memoir, p. 58). Comparing the tiak ratios 

 (Table 8, cols, i. A) it is seen to be much less alkalic than the Deception Island rock, and that entails 

 a much larger amount of free silica (O). It compares rather closely, however, with an Andean dacite 

 from Colombia (Table 8, col. B). The hypersthene-andesite of Thule Island (Table 8, col. 2) compares 

 fairly closely with the Bridgeman Island basalt (Table 8, col. C), but no Andean lava of like silica 

 percentage could be found with even an approximately similar F'l'M' ratio. Backstrom's ' olivine-free 

 basalt' from Saunders Island (Table 8, col. 3) finds its closest analogue in an olivine-basalt from Mont- 

 serrat (Table 8, col. D). The olivine-basalt of Cook Island (Table 8, col. 4) is only slightly undersaturated 

 (^= — i-o), notwithstanding its comparatively large content of olivine. This illustrates its affinity 

 with the more basic types of andesite. It compares well with the olivine-basalt lava of the Penguin 

 Island volcano (Table 8, col. E), with the exception that it is slightly less undersaturated and somewhat 

 more potassic than that rock. Again, the closest analogue of this rock is a labradorite-basalt lava from 

 St Vincent in the West Indies (Table 8, col. F). 



It would appear, therefore, that the predominant basic lavas of the South Sandwich Islands show 

 closer affinities with the comparable rocks of the Antilles than with those of the Andes. This may, 

 in turn, be regarded as evidence in favour of the view that the South Sandwich Islands do not lie 

 on the main line of the Scotia Arc, but form an easternmost ridge parallel to and in echelon with it. 

 On this view the main line of the Scotia Arc may curve southward from the eastern end of South 

 Georgia and join up with the South Orkneys. The most recent chart of the Scotia Sea^ shows South 

 Georgia trending to the south-east away from the line connecting it with the South Sandwich Arc, 

 and pointing towards a marked northerly projection of the 3000 m. depth-contour which, in turn, 

 leads towards the South Orkney Islands. 



Of the basement on which the volcanoes of the South Sandwich Islands stand we possess only very 

 exiguous and doubtful scraps of information, namely, a comparatively large piece of dolomitic 

 mudstone dredged off Zavodovski Island (p. 94), and fragments of epidote-biotite-gneiss and vein 

 quartz taken from a piece of floating ice near Bristol Island (p. 99). Any future geological exploration 

 of the islands should therefore include search for exposures of this foundation, and examination of 

 coarse fragmental igneous deposits for non-volcanic material which may be presumed to have been 

 derived from the basement. The latter line of research is much more likely to be fruitful than the 

 former, except perhaps on Bristol Island. 



Acknowledgements. The author's thanks are due to the Discovery Committee for defraying the cost 

 of the new rock analyses published in this work, to Prof. W. J. McCallien, D.Sc, for re-drawing 

 Fig. 8, p. 60, to J. M. Wordie, M.A., for his valuable introductory Foreword, and to Dr N. A. 

 Mackintosh for his editorial vigilance during the progress of this Memoir towards publication. 



1 H. F. P. Herdman, 'Report on Soundings taken during the Discovery Investigations, 1926-32', Discovery Reports, 

 VI, pi. xlv(i932). 



