10 -Day, etc- — Determination of Mineral 



absorbing agents to remove traces of carbon dioxide, hydrogen 

 sulphide and water vapor. 



3. Preliminary Measurements. 



The first datum necessary is the density or volume curve of 

 the metal up to the highest temperature to be used. As an 

 absolute basis for determining this volume, we must have the 

 measured linear or volumetric expansion of some solid sub- 

 stance up to the same temperature. The only one available 

 seemed to be amorphous silica (quartz glass), whose expansion 

 has been determined by several investigators, and is, further- 

 more, so small as to be almost negligible. We sought to 

 employ this known datum by making two series of measure- 

 ments : first, the displacement of metal by a solid graphite 

 float ; second, the displacement of metal by a similar float in 

 which as much as possible of the graphite was replaced by 

 some silica glass disks, cut from a block of glass made by Day 

 and Shepherd in this laboratory. A series of simultaneous 

 equations was thus obtained, containing the density of silver 

 and the expansion coefficient of graphite as unknown 

 quantities. 



Although this series gave the density of silver within - 5 

 per cent, the results were not entirely satisfactory, first because 

 the silica glass crystallized to cristobalite ai" about 1200°, cut- 

 ting off the measurements at that temperature ; and second, 

 because too much uncertainty was left in the expansion coef- 

 ficient of the graphite, which was required to be used in later 

 work as a correction. Therefore, direct determinations were 

 made of the expansion of the graphite used, and these were 

 taken as the basis for the volume determinations on the metals. 



4. Expansion of Graphite. 



The data on the expansion coefficient of Acheson artificial 

 graphite, together with a comparison and discussion of results, 

 have already been published elsewhere.* The measurements 

 were made on bars about 700 mm long, in two types of electrically 

 heated furnaces. The mean coefficient of expansion from 0° 

 is expressed by the formula 



1073 = 0-55 + 0-0016 t. 



The actual determinations are plotted in fig. 3. 



To obtain an accuracy of 0*1 per cent in the specific volume 

 of tin, the expansion coefficient of graphite needs to be known 

 with an accuracy of only 7'5 per cent at 1500°, and of 40 per 



* A. L. Day and R. B. Sosman, Jour. Ind. Eiig. Chem., iv, 490-493, 1912 ; 

 Jour. Washington Acad. Sci., ii, 284-289, 1912. 



