146 MK. J. A. DOUGLAS ON CHANGES OP PHYSICAL [May I907, 



in a specific- gravity flask. Thus the percentage-increase in volume 

 of the rock on becoming glass could be calculated. 



The objections which maybe brought forward against this method 

 are the following : — 



(1) In dealing with a cubic decimetre of the rock, the whole may not have 

 become completely fused. This possibility would be greatly increased if the 

 rock were not reduced to the finest powder, as appears to have been tbe case 

 since it was at first slowly heated to prevent the projection of chips. 



The following description seems to suggest incomplete fusion: — 'Tantot le 

 verre est parfaitement homogene, tantot au contraire, ainsi que de Saussure 

 l'avait deja remarque, on y observe de petits squelettes blancs formes de quartz 

 et quelquefois de feldspatb, qui ne se dissolvent pas dans la masse.' (Op. cit. 

 p. 1386.) 



(2) Kesearches on the synthesis of minerals have shown that very rapid 

 cooling is necessary to obtain a perfectly-homogeneous glass. This, I imagine, 

 is only possible where the bulk of the glass to be cooled is very minute. 



(3) The whole of the glass may not have been of the same composition. 



Dr. J. Morozewicz has described how, when melting a granite of specific 

 gravity 2 - 716, he obtained a glass, the specific gravity of which varied from 

 2*238 in the upper layers to 2*484 in the lower ; the silica being more abundant 

 at the surface. (Tscherm. Min. & Petr. Mitth. vol. xviii, 1898-99, pp. 1-90, 

 105-240.) 



(4) The molten glass had an action on the crucible. 



To use Delesse's own words : — e II faut cependant faire exception pour les 

 roches riches en mica, ainsi que pour les roches volcaniques qui corrodent 

 fortement les creusets et peuvent les percer.' (Op. cit. p. 1382.) 



Dr. Morozewicz has also experienced the same difficulty when dealing with 

 molten solutions in clay-crucibles. (Neues Jahrb. vol. ii, 1893, p. 42.) 



Bischof, moreover, has described how, after adding 30 per cent, of clay-slate 

 to a fused lava, he still obtained a perfectly-clear and homogeneous glass 

 (' Chem. & Phys. Geologie ' Suppl. 1871, p. 98). Delesse's glasses, then, may 

 have attacked the crucibles to a greater extent than he imagined. 



(5) There appears to exist no means of testing the accuracy of the method 

 employed in taking the specific gravity of the glasses, which, although crushed, 

 were full of bubbles. 



In the present research I have tried to eliminate, as far as 

 possible, these sources of error by finely powdering the rock in an 

 agate mortar, and melting only a very small quantity ; by using 

 platinum which is not attacked by the molten glass ; by the instan- 

 taneous cooling possible with electrical heating ; and by taking the 

 specific gravities of the glasses in a diffusion-column of heavy fluids: 

 to obtain good results with these necessitates the glasses being free 

 from bubbles and perfectly homogeneous. 



The apparatus 1 used in melting the rocks was very simple in 

 construction. It consisted of a short length of platinum- ribbon, 

 held by two horizontal brass-clamps which were 2 centimetres 

 apart ; these were supported by vertical pillars on a bed of slate, 

 2\ cms. thick, which acted as a complete insulator preventing any 

 leakage of current. The ribbon was 7 millimetres broad, and a 

 length of about 3 centimetres was used for each experiment; it 



1 The apparatus is similar to that suggested by Prof. Joly ' On the Volume- 

 Change of Pocks & Minerals attending Fusion ' Trans. Poy. Dublin Soc. ser 2, 

 vol. vi (1897-98) p. 283. 



