DAY AND SOSMAN: EXPANSION COEFFICIENT OF GRAPHITE 285 



fragments of aluminum (658°), silver (960°), copper (1083°) and the 

 mineral diopside (1391°) lying upon the bar. Intermediate temperatures 

 were read by means of a Holborn-Kurlbaum optical pyrometer, which 

 thus served simply as a device for interpolating between the melting 

 points. It was not necessary to obtain great accuracy in the tempera- 

 ture measurement, as the expansion is small. 



Below 700° the expansion is so small that we found it necessary, in 

 order to obtain accurate measurements of the coefficient, to use the 

 method which we had previously employed for the platinum alloys: 

 namely, to heat the bar in a narrow tubular resistance furnace, and read 

 the expansion on a series of very fine lines 0.2 mm. apart, drawn with a 

 dividing engine. These lines were drawn upon the polished ends of small 

 plugs of silver or platinum set into the graphite at points 500 mm. apart. 

 Up to the red heat of 700° the bar could be protected fairly well from 

 oxidation, and contamination of the thermoelement could be prevented 

 by enclosing it in Jena glass capillaries. 



Table I contains the data. In the first two columns are given the 

 date and the temperature at the time of the measurement. In the third 

 column is the initial distance at 0°, between two reference lines on the 

 silver or platinum markers on the graphite bar, in mm. ; this distance was 

 frequently changed slightly by resetting the silver or platinum markers. 

 The fourth column contains the expansion from 0°, in mm., and the fifth. 



the mean linear expansion on coefficient j— from 0° to 2°. 



i t 



The coefficient is seen to be very small, though increasing rapidly with 

 rising temperature. The total expansion up to 1000° is only about one- 

 fifth of that of platinum. Two samples were used, but no systematic 

 difference exceeding the error of measurement was observed. 



The greatest uncertainty arises from changes in length of the bar after 

 heating. In every case where the change was measurable, the bar was 

 longer than before heating, as if this somewhat porous material had 

 failed, after expanding, to entirely pull itself together again. The 

 largest change noted was 0.046 mm., which is 0.009 per cent on the total 

 length, and therefore quite negligible when the total length or volume 

 is being considered. 



Within the limits of error, the results may be expressed by the 

 formula 10 H /3 = 0.55 + 0.0016 t in which /3 is the mean linear coeffi- 

 cient — from 0°. The "true coefficient" a, or rate of expansion, 



to" 



( 



1 dl\ ,,,,,, 



— • — I at any temperature t would be therefore; 



10 6 a = 0.55 + 0.0032 t. 



If this equation continues to hold below 0°, the true coeffi- 

 cient would become zero at about — 170°; in other words this 



