REPORT ON THE PETROLOGY OF OCEANIC ISLANDS. 99 



"stone rivers:" — "The progress of knowledge will probably some day give a simple 

 explanation of this phenomenon, as it already has of the so long thought inexplicable 

 transport of the erratic boulders which are strewed over the plains of Europe." ' 



The specimens collected by Thomson show lithological characters of some interest. 

 One of these blocks is in the form of a quadratic prism, measuring about 40 centi- 

 metres by 10 ; the fracture is regular and polyhedric ; the edges hardly show a trace 

 of weathering, but the surface is covered by a less coherent layer of slight thickness. 

 Beneath this thin altered surface the rock remains remarkably fresh. To the naked 

 eye it appears to possess a granitoid structure with grains of medium size ; with the 

 lens a plagioclastic felspar can be seen, associated with a black mineral of the 

 amphibolic or pyroxenic group. This rock belongs to the type occurring in the eruptive 

 masses often embedded or injected amongst palaeozoic strata, such as those of the 

 Falkland Islands. Microscopic examination shows that the fragment in question must 

 be classed as a diabase, and it also reveals that the rock possesses peculiarities of some 

 interest, and of a kind to which the attention of lithologists is specially directed. This 

 diabase is composed of plagioclase, augite, hornblende, biotite, and magnetite. Of all 

 these minerals, that which at present plays the most important part is unquestionably 

 hornblende ; but this constituent is of secondary origin, and can only take a subordinate 

 place in classifying the rock lithologically. The sections of felspar are remarkable on 

 account of the very great number of fine plagioclastic striae which they present. In 

 exceptional cases only the Carlsbad twin is apparent, but in others the section 

 shows, at the same time, lamella? twinned according to the albite and pericline 

 laws. These plagioclase sections do not present definite crystallograjjhic outlines, 

 but microscopic examination shows that they are generally elongated following on the 

 edge PjM. It is somewhat rare to find a section parallel to M which would suffice to 

 determine the sign and the angle of extinction. This was possible only in one case : 

 a section presenting two cleavages, parallel to P and to T, crossing at an angle of 

 more than G0°, gave a negative extinction of about 30°. This observation shows that 

 the plagioclase in question approaches closely to a mixture analogous to that of 

 bytownite. These sections of plagioclase are remarkably clear, and the phenomena of 

 chromatic polarisation are sharp and brilliant ; the decomposition, so often found in 

 the felspars of granitoid rocks, has, as yet, only affected the plagioclase lightly. This 

 mineral has been subjected to mechanical deformation ; some of the lamellae are 

 laminated, showing an undulating extinction ; they are strained, curved, and split up 

 into numerous slices. 



The augite of this rock presents some noteworthy features. Like the felspar it has 

 no definite crystallographic outline. In the sections perpendicular to the axis c a net- 



1 Darwin, Journal of Researches, 1879, pp. 198, 199. 



