822 Professor Dewar [April 1, 



transparency, and that a non-conducting body became inefifective at 

 low temperatures in shielding a vessel from the influx of heat. 

 Experiments, about to be detailed, however, prove that such is not 

 the case, the transference of heat observed by Pictet appearing to 

 be due not so much to the materials themselves as to the air con- 

 tained in their interstices. Good exhaustion in the ordinary vacuum 

 vessels used in low temperature work reduces the influx of heat to 

 one-fifth of what is conveyed when the annular space of such double- 

 walled vessels is filled with air. If the interior walls are silvered, 

 or excess of mercury is allowed to remain, the influx of heat is 

 diminished to one-sixth of the amount entering without the metallic 

 coating. The total efi'ect due to the high vacuum and silvering is 

 to reduce the ingoing heat to one-thirtieth of the original amount, 

 i.e. roughly, to 3 J per cent. 



By filling the annular space between the walls of several similar 

 vacuum vessels with various substances, and exhausting them all 

 to the same low pressure, large differences in the thermal isolation 

 were observed. The rate of evaporation of equal volumes of liquid 

 air contained in the respective vessels measures the rate of influx of 

 heat. Moreover, it appears that what might bo called under the 

 circumstances the thermal transparency of some materials diminished 

 at very low temperatures instead of increasing, as had been asserted. 

 Thus, of two vacuum tubes (one simply exhausted, and the other having 

 powdered carbon in the vacuum space), the latter, at low temperature, 

 was the most efficient preserver of liquid air, showing that tbe carbon 

 diminished the radiation and gas convection. But when the vacuum 

 was destroyed and air admitted into the space, the liquid air in the 

 carbon tube boiled oft' much more vigorously than that in the simple 

 tube, indicating that at ordinary temperatures carbon allowed more 

 heat to pass than did air. 



In conducting these experiments, generally sets of three double- 

 walled glass tubes, as nearly identical in size and shape as possible, 

 were mounted on a common stem, and two out of the three filled with 

 different kinds of powders, while the third is left empty as a standard 

 for comparison (Fig. 3). In this way each set had the same vacuum, 

 and as intercommunication between the tubes after sealing off from 

 the pump was left free, any deterioration in the vacuum on keeping 

 affected all three vacuum tubes to the same extent. 



The preparation of such tubes entails enormous labour, because 

 it takes days of exhaustion with a mercurial pump to extract the 

 occluded gases, even at as high a temperature as the glass would 

 stand. Before beginning the experiment, the vacuum tubes of each 

 triple set were filled with liquid air, and allowed to stand half an 

 hour in order to get the heat conduction in the porous mass into a 

 steady state. The tubes after this treatment were filled to the same 

 height, and the relative times required to distil off the same volume 

 of liquid air from each observed — the outer surface of the vacuum 

 tubes being maintained at a steady temperature by immersion in a large 

 vessel of water. Neither the tubes nor the shape of the vacuum space 



