April i;, 1902J 



NATURE 



577 



to the liquefaction of that gas ; a conclusion which is confirmed 

 by this research. There seemed no doubt,' however, that after 

 due modification the Hampson-Linde method would be ap- 

 plicable to the liquefaction of hydrogen, the necessary condition 

 being that the gas should be cooled to below the temperature at 

 which the Joule-Kelvin effect changes sign before entering the 

 regenerator coil of the apparatus. Below this temperature the 

 gas would become cooled on free expansion ; as in the case of 

 air in the ordinary Hampson apparatus, the cooling would be 

 progressive and, were the heat insulation sufficiently perfect, 

 would result in the partial liquefaction of the gas. 



As is well known, Dewar and Travers (Phil, Ma^., 1901) 

 have both succeeded in applying this principle to the production 

 of liquid hydrogen. Both investigators employed as a pre- 

 liminary refrigerant liquid air boiling under reduced pressure, 

 and by this means cooled the hydrogen to - 200° C. The 

 question now arises as to whether this is really necessary, or 

 whether it would not be possible to liquefy hydrogen by allowing 

 it to undergo free expansion after cooling it to a temperature 

 less difficult to attain. The following investigation was under- 

 taken in the hope of throwing some light on the subject and 

 of determining the temperature at which the sign of the Joule- 

 Kelvin effect changes. 



Witkowski {Roz. pr. Akad. krak., W.M.P. xxxv. 247; id. 

 189S) has applied two methods to the calculation of this tem- 

 perature. The first method, based on the theory of correspond- 

 ing stales, gave -46° C. as the temperature ; the second method, 

 in which Rose-Innes' formula was applied to the extra- 

 polation of Joule and Lord Kelvin's experimental values at 

 temperatures between 0° and 100° C, gave -79°'3, a result 

 which is nearer to that obtained by me experimentally, viz. 

 -So°-5. Rose-Innes' formula (Phil. /l/oj-. [5], xlv, 228), 



is purely empirical, but employing 64'i and O'JJI as values 

 for the constants, it reproduces the experimental results with 

 remarkable accuracy. 



Descriplioii of the Apparatus. 



The hydrogen employed in these researches was obtained by 

 the action of pure dilute sulphuric acid on commercial zinc ; the 

 gas was purified by passage through solutions of caustic soda and 

 potassium permanganate, anil finally through a tower contain- 

 ing pumije-stone soaked in mercuric chloride solution. The gas 

 was collected in a zinc gasometer capable of holding 1200 litres, 

 and from this was taken directly into the Whitehead torpedo- 

 compressor, and compressed at 180 atmospheres into a steel 

 cylinder of 13 litres capacity ; in this cylinder was suspended 

 a wire net containing sticks of caustic potash. During these 

 operations every care was taken to remove all traces of air from 

 the apparatus. 



The expansion apparatus is shown in the accompanying 

 figure. The steel cylinder, which contains the hydrogen under 

 pressure, communicates by means of a copper tube with a 

 manometer and with the tube a of the apparatus. The tube a 

 is bent into a spiral h^ which terminates in a valve c. The 

 valve consists of a jet enclosed within a steel tube, perforated at 

 the sides for the escape of the gas, which serves as a support for 

 it. The valve spindle d passes through a gland packed with 

 asbestos at 0, and is screwed below to fit the opening through 

 which it passes, so as to allow of the valve being opened or 

 closed. The valve is enclosed in a thin metal box /;/(, which 

 is lined with chamois leather ; the gas can escape by a tube/^, 

 through the vertical portion of which an electric resistance 

 thermometer is introduced. This instrument has previously 

 been employed in the determination of the critical and boiling 

 points of hydrogen (Phil. Mug. [5], xxxix. 199; xl. 202) ; the 

 electrical connections are made by means of binding screws at 



/ and g- 



So as to be able to surround the coil b and the metal box 

 enclosing the jet and thermometerwith different refrigerants, the 

 metal cap nn is cemented to the top of a thick-walled glass 

 vessel //. This in turn contains a thin glass vessel mm, insu- 

 lated from the former at the top and bottom. 



The refrigerating substances, liquid air, liquid ethylene, and 



1 Kammerlingh Onnes (Leyden CommitnicationSy xxiii. 1896, ]6) pointed 

 out the possibility of liquefying hydrogen by means of LinUe's appar.itus, 

 .and stated the conditions, based on the theory of corresponding states, 

 under which lii[uefaction can occur. 



solid carbonic acid and ether, can be introduced into tnm through 

 a T-tube passing through the cover of the apparatus, and by con- 

 necting one branch /■ with an exhaust pump the substance can 

 be evaporated under reduced pressure and the temperature 

 varied considerably. The temperature of the coil and jet can 

 be roughly determined from the readings of a mercury mano- 

 meter attached to the apparatus. 



Method of Experiment. 



The first experimentsweremadeatabout — igo'C, the temper- 

 ature of liquid air. The initial pressure on the hydrogen was 

 about 170 atmospheres, and on opening the valve for about five 

 seconds a considerable cooling look place, the beatii of the 

 galvanometer attached to the electric thermometer moving 

 200 mm. over the scale. Using as a refrigetant liquid ethylene 

 boiling at - 003° C, and with an initial pressure on the hydrogen 

 of 150 atmospheres, cooling 

 took place when the valve 

 was opened, but to a less 

 extent, the ray from the 

 galvanometer moving only 

 30 mm. This was taken to 

 indicate that the tempera- 

 ture of inversion lay con- 

 siderably above —100° C, 

 and a series of experiments 

 was next made about the 

 temperature of solid car- 

 bonic acid and ether. 



In this series twenty-five 

 experiments were made, 

 starting in each case with 

 a temperature of -78° C. 

 and an initial pressure of 

 between 117 and no atmo- 

 spheres. To regulate the 

 expansion, a small steel 

 cylinder of 0'6 litre capacity 

 was introduced between the 

 main cylinder and the tube 

 a ; by filling the small 

 cylinder from the main 

 cylinder and closing the 

 communication between the 

 two before each expansion, 

 the quantity of gas which 

 escaped from the valve, and 

 consequently the temperature 

 change, was always the 

 same. It appeared from 

 these experiments that at 



- 78° hydrogen becomes 

 warmer on free expansion, 

 but only very slightly so, 

 the ray from the galvano- 

 meter moving 3 mm. over 

 the scale. p 



By reducing the pressure 

 on the carbonic acid by 

 careful pumping, the temperature was slowly reduced. At 



- 83°, decided cooling took place on expansion, the galvano- 

 meter beam moving 5 mm. over the scale. By numerous trials 

 it was found that the temperature at which the Joule-Kelvin 

 effect became zero appeared to be at about - So' '5 C. 



Concli/stoiis. 

 From these experiments it appears that the temperature of 

 the inversion of the Joule-Kelvin effect for hydrogen lies about 



- So^'S C, a number which agrees very closely with that arrived 

 at by Witkowski from the Rose-Innes equation (- 79''3). This 

 agreement makes it worth while considering the further applica- 

 tion of this equation to other substances, particularly to air, and 

 the application of the results to the calculation of the critical 

 temperature of hydrogen on the basis of the theory of corre- 

 sponding states. We have the following data : — ■ 



Temperature of 



inversion of Critical . 



Joule-Kelvin effect. temperature. 



Air ... ... ... 633° Abs. 133° Abs. 



Hydrogen ... ... I92°'S Abs. — 



NO. 1694, VOL. 65J 



