262 LIQUID HYDKOGEN. 



the quill tube is attached, then, on repeating- the expenment, the same 

 results follow, only the volume of the liquid air formed agrees with 

 the total quantity present in the vessel. This suggests that any air 

 left in the closed vessel must have a very small pressure. This Is con- 

 firmed by attaching a mercurial gauge to any vessel containing air, 

 when it will be seen the vacuum produced by hydrogen cooling is 

 equal to that of a Torricellian vacuum (fig. 2, PI. I). To reach such a 

 high exhaustion the solid oxygen and nitrogen at the boiling point of 

 hydrogen must be practically nonvolatile or have an exceedingly 

 small vapor pressure. If the ordinary air contains free hydrogen, 

 helium, etc., which are noncondensable in this way of working, then 

 the vacuum would not be so high as with pure oxygen or nitrogen. 

 This method may be used to separate the incondensable gases from 

 the air. Such air vacua when examined spectroscopically show the 

 lines of hydrogen, helium, and neon. We may now employ this proc- 

 ess to produce high vacua, and test their exhaustion by the character 

 of the electric discharge. Vacuum tul)es which have been prepared in 

 this way show extraordinary resistance to the passage of the electric 

 discharge; they also show the marked phosphorescence of the glass 

 characteristic of Crookes tubes (Figs. F and G, PI. II). It is, how- 

 ever, the rapidity with which such high exhaustions can be attained 

 that is so interesting. You will observe that this large Geissler tube, 

 previously exhausted to some 3 inches pressure, will, when the end 

 part is immersed in liquid hydrogen, pass through all the Avell-known 

 changes in the phases of striation — the glow on the poles, the phosphor- 

 escence of the glass — in the space of a fraction of a minute. From 

 this it follows that theoretically we need not exhaust the air out of our 

 double- walled vessel when liquid hydrogen has to be stored or collected. 

 This makes a striking contrast to the behavoir of liquid air under 

 similar circumstances. The rapid exhaustion caused by the solidifica- 

 tion of the air on the surface of a double-walled unexhausted test 

 tube, when liquid hydrogen is placed in it, may be shown in another 

 way. Leave a little mercury in the vessel containing air, just as if it 

 had been left from making a mercurial vacuum. Now, we know mer- 

 cury, in such a vacuum, can easily be made to distill at the ordinary 

 temperature when we cool a part of the vessel with liciuid air, so that 

 we should expect the mercury in the unexhausted test tube to distill on 

 to the surface cooled with the liquid hydrogen. This actually takes 

 place. A rough comparison of the relative temperatures of boiling 

 hydrogen and oxygen may be made by placing two nearly identical 

 hydrogen-gas thermometers operating at constant pressure side by side 

 and cooling each with one of the liquids (PI. III). It will be seen that 

 the contraction in the thermometer cooled with lic^uid hydrogen ele- 

 vates the li(piid some six times higher than that of the corresponding 

 liciuid column of the thermometer placed in the licjuid oxygen. A 

 constant-volume hydrogen thermometer constructed as shown in PL IV 



