GEOLOGICAL OBSERVATIONS IN FIJI. 141 



flexible, impervious skin and the slight disturbance produced. Tt 

 would seem probable that the retention of the volcanic gases under 

 such circumstances would allow hypersthene and hornblende to- 

 develop. 



Hypersthene and hornblende should, therefore, characterize inter- 

 mittent and submarine \olcanoes, whereas the monoclinic pyroxenes 

 should be found near volcanoes having open vents. The hornblende, 

 hypersthene, and augite andesites are closely related to basalts and 

 are considered to be pure dilTerentiates of a primary basaltic magma. 

 Because of insufficient time it was impossible to make a thorough, 

 study of the distribution of lavas about individual volcanic vents.. 

 Guppy (1903, pp. ()2-G5) observed that augite andesites were inti- 

 mately mingled with olivine basalts in the sub-Recent crater of Mount 

 Seatura, western Viti Levu. The writer observed that the ground- 

 mass of olivine basalts in which zoned feldspars occurred usually 

 contained a feldspar more acid than the phenocrysts and that the- 

 ferromagnesian mineral present in such a ground-mass was usually 

 augite. The recent experiments of Bowen (1915, p. 47) practically 

 establish the fact that an augite andesite may differentiate by gravity 

 from an olivine basalt. 



In conclusion it may be said that the trend of the volcanic succes- 

 sion in Fiji has been from acid to basic types; yet, while this is true, 

 hornblende andesites have persisted in Kandavu from the first. 

 Another striking fact is the persistency of volcanic vents. Few 

 instances are known in which the later, basaltic eruptions have built 

 up independent craters. On the other hand, a number of examples- 

 have been given which show that later basaltic cones were formed oiu 

 or near earlier andesitic volcanoes. 



