and Rock Densities at High Temperatures. 29 



Above the rather accidental temperature at which the glass 

 crystallizes, the crystal aggregate again begins to melt. The 

 melting does not occur at a single temperature, but is spread 

 over a range of temperature, in accord with the fact that this 

 is a complex system of eight or more components. The 

 remelting begins at about 1150° and is practically complete 

 somewhat below 1300°. The rock flows readily at a tempera- 

 ture of 1225°, at which temperature all is fused excepting 

 about a third of the feldspar. A little feldspar still remains 

 undissolved at 1250°, but only traces remain at 1300.°* 



A considerable source of uncertainty begins to be apparent 

 at 1100°, which increases as the temperature rises. The reduc- 

 ing action of graphite upon the iron oxides present causes the 

 formation of carbon monoxide between the graphite crucible 

 and the molten rock. The carbon monoxide escapes through 

 the porous graphite, but being constantly renewed, it tends to 

 make the volume of the liquid diabase appear slightly too 

 great. We have found no other material than graphite which 

 can be used in this form of apparatus, so that this gas-produc- 

 ing reaction is a serious limitation upon the method when 

 applied to iron-bearing silicates. Calcium and magnesium 

 silicates seem to be acted upon by graphite very little or not 

 at all at these temperatures. 



On the cooling curves a sudden contraction was again 

 observed just below 1050°, caused by the crystallization of the 

 liquid diabase. The resulting volume was approximately that 

 attained when the block of glass crystallized with rising tem- 

 perature. Below this temperature the contraction was normal 

 as far as it was observed. At room temperature after the 

 measurements the blocks were found to be tine-grained, strong, 

 and very tough. They appeared free from bubbles or vacuoles, 

 except at the surface. Examination under the microscope 

 showed less than 5 per cent of glass present. The blocks were 

 seen to consist of clear feldspar crystals of variable index, the 

 majority corresponding to a basic labradorite ; also masses of 

 dark brown feathery crystals too fine to be identified. At 20° 

 the gross kilogram-volume of the blocks was 349*6 and 348'4, 

 mean 349 - ; this is 3*8 per cent greater than the volume of 

 the original rock, while the volume of the clear glass was 7'7 

 per cent greater than that of the rock. 



Barus" 1 curves. — The foregoing considerations plainly show 

 what must have occurred in the experiments of Barus (see 

 page 5). Although by rapid cooling in the open air he 

 obtained in his preliminary tests a clear diabase glass, in his 

 fusion tube there was time for the liquid to crystallize at 



*For details of fusion tests, see Sosman and Merwin, J. Wash. Acad. Sci., 

 iii, 389-395, 1913. 



