252 



gray: linear expansivity 



itudinal temperature gradient in a tubular air bath of considerable 

 length. It i.s a plug formed of two thick blocks of a good heat con- 

 ductor united by a thin shell and separated by a considerable layer of 

 a poor conductor. Fig. 3 represents one in each end of a tube the cir- 

 cumference of which is heated by a fluid circulating spirally around it 

 and returning to the inlet end by linear flow through an outer concen- 

 tric passage. In such a return-flow tube, whose ends were plugged for 

 10 cm. essentially as indicated, though not quite so well, an air column 

 4 cm in diameter and more than a meter long has been repeatedly heated 

 by circulating oil to over 100°C. so uniformly thruout its entire 

 length that right against the plugs the temperature was only a few 

 hundredths of a degree lower than in the center, while several centi- 

 meters farther inward the drop was only a few thousandths. Yet with 



Fig. 3. Oil-heated return-flow tube showing double plugs to reduce tempera- 

 ture gradient. 



the ends closed by equally long plugs of such heat insulators as cotton- 

 wool and cork there was a marked lowering of the temperature in the 

 immediate vicinity within, even when the interior was only slightly 

 warmer than the room; and in many of the experiments made this drop 

 was apparent for a long distance within the tube. If the inside tube 

 (of brass 1 mm. thick) were extended 4 cm. beyond the heating jacket, 

 relatively enormous end effects were produced in the air column, espe- 

 cially if the projections were not well protected by lagging. But un- 

 hindered radiation and conduction through the sides of the tube to 

 the surrounding air was found to disturb the distribution of temperature 

 within only slightly. In nearly all cases where the tube was not sur- 

 rounded by an insulating jacket, there was evidence of a small progress- 

 ive drop in temperature in the direction away from the inlet and exit 

 end. Doubtless this could have been made imperceptible by substi- 

 tuting a poorly conducting tube for the thin brass one separating the 

 oppositely directed currents ol heating Huid. 



