490 



SCIENCE. 



I N. S. Vol. VIII. No. 198, 



gas becoming more prominent as the pres- 

 sure was reduced. I do not doubt that this 

 was due to the interference of the heavier 

 and slower moving molecules of the ordi- 

 nary gases always mixed with the new gas. 

 In last year's paper I described the same 

 effect found in a mixture of carbon dioxide 

 and hydrogen. When the new gas is ob- 

 tained in a state of purity I expect to find 

 its curve of heat conductivity similar to 

 those of hydrogen and helium, but with 

 immensely increased ordinates. 



Believing that the new gas is very much 

 lighter than air or hydrogen, and may 

 therefore be separated from the atmosphere 

 by successive diffusions, I have spent sev- 

 eral months in experiments with this end 

 in view. Quite recently my efforts have 

 been crowned with most promising success. 

 The difficulty has been to find a suitable 

 porous septum ; one free from holes, and 

 sufficiently fine-grained to prevent more 

 than a very slow passage of air, while a 

 considerable surface is exposed to the at- 

 mosphere on one side, and to a fairly good 

 vacuum on the other. Many substances 

 were tried. India-rubber gave encouraging 

 results, but was capricious and very slow. 

 The best results have been obtained with 

 porous porcelain having its superficial pores 

 nearly closed by suitable treatment. A tube 

 of this kind, closed at one end, and exposing 

 rather more than five square inches of sur- 

 face to the air, was connected with the ap- 

 paratus, and the whole kept exhausted to a 

 pressure of 1.3 mm. About 19 cc. of gas 

 was diffused per hour. After 36 hours the 

 diffusion tube was shut off, the exhaustion 

 continued, and data for the curve vi n ob- 

 tained. It was found impracticable to carry 

 the exhaustion below six millionths, because 

 of the presence of moisture, which inter- 

 fered with the action of the pump, no 

 desiccating agent being used. As both 

 phosphorus pentoxide and soda-lime absorb 

 the new gas they cannot be used for desic- 



cation. If the exhaustion had been carried 

 as far as in the curve o p it is probable that 

 a higher conductivity than in that curve 

 would have been shown ; i.e., a conductiv- 

 ity more than 42 times that of hydrogen, or 

 something like a hundred times that of air. 

 Here we have the heat conductivity of air 

 at very low pressure, increased something 

 like a hundred-fold by one diffusion. I have 

 not yet tried a secondary diffusion, but feel 

 confident that successive diffusions of air 

 will afford a practicable means of obtaining 

 the new gas in a state of approximate 

 purity. 



Now let us discover, if possible, the mean- 

 ing of the extraordinary heat conductivity 

 of the new gas. I have tabulated the molec- 

 ular weight, density, specific heat, mean 

 molecular velocity and heat conductivity 

 of hydrogen, helium, carbon monoxide, air 

 and carbon dioxide, which are the known 

 gases represented in the curve chart. For 

 easy comparison, I have taken not only the 

 density, but the specific heat, mean molec- 

 ular velocity and heat conductivity of 

 hydrogen as unity. The similarity of 

 values in the fifth and sixth columns, for 

 molecular velocity and heat conductivity, 

 is striking. On the chart the curves for 

 carbon monoxide, air and carbon dioxide 

 are evidently too high to correspond well 

 with the relative values of molecular veloc- 

 ity in the fifth column ; but these curves 

 represent only the last 50 millionths of the 

 complete curves. By following these back 

 to 500 millionths, which is still a rather 

 small pressure, and taking their values at 

 intervals of 10 millionths from no pressure 

 upward, we obtain as a mean of all these 

 values for each gas the quantities shown in 

 the sixth column. These agree fairly well 

 with the molecular velocities. 



I offer the suggestion that the relatively 

 high conductivity of the last three gases at 

 the low pressure shown in the chart is due 

 to the dissociation of some of their mole- 



