October 10, 1902.] 



SCIUNCE. 



571 



mental values. We may safely infer, 

 therefore, that if hereafter a gas be dis- 

 covered in small quantity even four times 

 more volatile than liquid hydrogen, yet by 

 a study of its isothermals at low tempera- 

 ture we shall succeed in finding its most 

 important liquid constants, although the 

 isolation of the real liquid may for the 

 time be impossible. It is perhaps not too 

 much to say that as a prolific source of 

 knowledge in the department dealing with 

 the continuity of state in matter, it would 

 be necessary to go back to Carnot's cycle 

 to find a proposition of greater importance 

 than the theory of van der Waals and his 

 development of the law of corresponding 

 states. 



It will be apparent from what has just 

 been said that, thanks to the labors of 

 Andrews, van der Waals, and others, 

 theory had again far outrun experiment. 

 We could calculate the constants and pre- 

 dict some of the simple physical character- 

 istics of liquid oxygen, hydrogen, or nitro- 

 gen with a high degree of confidence long 

 before any one of the three had been ob- 

 tained in the static liquid condition per- 

 mitting of the experimental verification of 

 the theory. This was the more tantaliz- 

 ing, because, with whatever confidence the 

 chemist may anticipate the substantial cor- 

 roboration of his theory, he also anticipates 

 with almost equal conviction that as he 

 approaches more and more nearly to the 

 zero of absolute temperature, he will en- 

 counter phenomena compelling modifica- 

 tion, revision, and refinement of formulas 

 which fairly covered the facts previously 

 known. Just as nearly seventy years ago 

 chemists were waiting for some means of 

 getting a temperature of 100 degrees below 

 melting ice, so ten years ago they were 

 casting aboiit for the means of going 100 

 degrees lower still. The difficulty, it need 

 hardly be said, increases in a geometrical 

 rather than in an arithmetical ratio. Its 



magnitude may be estimated from the fact 

 that to produce liquid air in the atmos- 

 phere of an ordinary laboratory is a feat 

 analogous to the production of liquid water 

 starting from steam at a white heat, and 

 working with all the implements and sur- 

 roundings at the same high temperature. 

 The problem was not so much how to pro- 

 duce intense cold as how to save it when 

 produced from being immediately levelled 

 up by the relatively superheated surround- 

 ings. Ordinary non-conducting packings 

 were inadmissible because they are both 

 ciunbrous and opaque, while in working 

 near the limits of our resources it is essen- 

 tial that the product should be visible and 

 readily handled. It was while puzzling 

 over this mechanical and manipulative 

 difficulty in 1892 that it occurred to me 

 that the principle of an arrangement used 

 nearly twenty years before in some calori- 

 metric experiments, which was based upon 

 the work of Dulong and Petit on radia- 

 tion, might be employed with advantage as 

 well to protect cold substances from heat 

 as hot ones from rapid cooling. I there- 

 fore tried the effect of keeping liquefied 

 gases in vessels having a double wall, the 

 annular space between being very highly 

 exhausted. Experiments showed that 

 liquid air evaporated at only one fifth of 

 the rate prevailing when it was placed in 

 a similar unexhausted vessel, owing to the 

 convective transference of heat by the gas 

 particles being enoi-mously reduced by the 

 high vacuum. But, in addition, these ves- 

 sels lend themselves to an arrangement by 

 which radiant heat can also be cut off. It 

 was found that when the inner walls were 

 coated with a bright deposit of silver the 

 influx of heat was diminished to one sixth 

 the amount entering without the metallic 

 coating. The total effect of the high 

 vacuum and the silvering is to reduce the 

 ingoing heat to about three per cent. The 

 efficiency of such vessels depends upon get- 



