254 gray: linear expansivity 



uniformity was practically perfect for 12 cm. in front of the radi- 

 ating diaphragm at 1244?9, drops of 8°, 10°, and 11° took place 

 within the second 4 cm. of this same region when the diaphragm 

 was at 621°, 1041°, and 1308°, respectively. Day and Sosman 5 

 made five separate attempts to wind a tubular furnace with two 

 small side openings so that it would produce a uniform tempera- 

 ture along a platinum-iridium bar 25 cm. long, the linear expan- 

 sivity of which they were determining. The deviations that 

 they report within the central 24 cm. of their two most successful 

 furnaces amount to 5° at 300°, increasing steadily to about 20° 

 at 700°, and reaching about 50° at 1000? While the thick- 

 walled iron tube used in these latter trials produced markedly 

 better results than the porcelain tube used in the earlier ones; 

 still it is quite clear that even the conductivity of the iron was 

 very far from adequate to secure uniformity, expecially at tem- 

 peratures much removed from the one at which a particular 

 distribution of the heating coils gave the best results. 



The double plug with small temperature gradient across an 

 insulating layer, which was described above, indicated a way of 

 reducing such difficulties as have just been enumerated. Accord- 

 ingly, an electrically heated furnace was constructed for deter- 

 mining the expansivity of bars, the elongations being measured 

 by the suspended-wire method outlined in the first section of 

 this paper. Fig. 4 represents the essential features. 



Since this furnace was designed primarily for convenience of manipu- 

 lation and for securing merely moderate uniformity of temperature, 

 only a partial use was made of the principles enunciated on page 251. 

 The furnace tube, of iron fairly uniformly wound for its entire length 

 with constantan ribbon, had its central portion completely filled with 

 an iron block pierced by two longitudinal cavities, of square cross- 

 section, symmetrically situated above and below the geometrical axis of 

 the tube. The lower cavity was almost completely filled by the bar 

 to be measured, while the upper one contained a similar dummy bar, 

 the temperature of which was determined by a thermo-element placed 

 within a hole following the axis of the bar. While the conductivity of 

 the large metal masses probably contributed considerably toward the 

 production of temperature uniformity within this furnace, and while 



5 Described in a paper by A. L. Day and J. K. Clement, Am. Jour. Science 

 26: 425. 1908. 



