and Rock Densities at High Temperatures. 15 



6. Quartz. 



A considerable amount of data already exists on the expan- 

 sion of quartz. Fizeau* and Benoitf have determined its linear 

 expansion, both parallel and perpendicular to the optic axis, 

 with great accuracy below 100°. Reimerdes^: has extended 

 the parallel coefficient to 220°, and Randall§ to 500°. 

 Le Chatelier| made approximate measurements on both axes 

 up to 1060°, at a comparatively small number of tempera- 

 tures. 



The quartz which we used was all from Minas Geraes, 

 Brazil, and was in the form of clean transparent blocks weigh- 

 ing from 35 to 85 grams. 



Measurements were made up to 1602° under metallic tin, in 

 the inverted crucible apparatus described on page 7. Check 

 measurements were also made in another form of apparatus 

 designed for lower temperatures and for solid blocks only. 



This apparatus consisted of a cage similar in form to the 

 graphite cage, but made of monel metal ^[ (a nickel-copper 

 alloy). The graphite float-crucible was replaced by a four- 

 pronged metal clasp which held the block of quartz. The 

 crucible containing the metal was of steel. The furnace was 

 wound with platinum wire. 



Measurements were made with this apparatus, using the 

 molten eutectic of tin and lead, the density of which had been 

 determined by means of a graphite block. 



The individual determinations are too numerous to be repro- 

 duced here. Table II, giving the results of one of the series, 

 will suffice to show the order of magnitude of the quantities 

 measured, and the precision. 



The first column gives the temperature, and the second the 

 total displacement, which is equal to the sum of the weights of 

 the sample, the graphite crucible and cage, the pan, and the 

 added weights. The third column gives the density of liquid 

 tin at the temperature in question, and the fourth column the 

 resulting volume of tin displaced. Subtracting the volume of 

 the graphite crucible in column 5 gives the volume of the 

 sample, and dividing this by the mass of the sample gives the 

 specific volume in the last column. For convenience this is 

 multiplied by 1000, so that the figures represent the " kilogram- 

 volume" or volume of one kilogram in cubic centimeters. 



*Pogg. Ann., cxxviii, 564-589, 1866. 

 fTrav. Mem Bur. Int., vol. vi, 1888. 

 % Inaug. Diss. Jena, 1896. 

 §H. M. Eandall, Phys. Rev., xx, 10-37, 1905. 

 H H. LeChatelier, Compt. rend., cviii, 1046-1049, 1889. 

 ^[ Steel, which was used at first, is too magnetic, and a small error is 

 introduced on account of the force exerted by the furnace current. 



