July 25, 1901] 



NA TURE 



6^0 



Wiobleu-ski and Olszewski in 1SS3 succeeded for the first 

 time in cooling oxygen or air to such a low temperature 

 that under moderate pressure it condensed and remained 

 as a visible liquid, of which, when it was allowed to boil 

 at ordinary pressure, a portion remained liquid at a 

 lower temperature. This cascaded vaporisation method, 

 the direct descendant of Faraday's system, was subse- 

 quently used by Dewar in improved apparatus on a 

 larger scale, and was the only means of obtaining con- 

 siderable quantities of liquid air down to 1895. Attempts 

 were made by others than Pictet to apply it to the lique- 

 faction of hydrogen. But the critical temperature of 

 hydrogen, above which no pressure can liquefy it, is so 

 low that even air boiling into a vacuum at, say, - 2 10 C , 

 or solid nitrogen at - 225^ C, is not cold enough to 

 cool it below its critical point. If an intermediate gas 

 could have been found, with a critical point high enough 

 to admit of its being condensed under high pressure at 

 the lowest temperature of liquid air, and boiling under 

 reduced pressure at a temperature below the critical 

 point of hydrogen, the problem would have been solved. 

 Nature having provided no such gas, Dewar tried to 

 make one by mi.xing nitrogen and hydrogen, in the hope 

 that, after the manner of oxygen and nitrogen in air, 

 they would liquefy together. Olszewski made a similar 

 attempt with a mixture of oxygen and hydrogen ; but no 

 one succeeded in liquefying hydrogen by the employ- 

 ment of vaporisation cooling, though intensified by 

 cascading in four stages. 



Meantime, another method of obtaining a cooling 

 effect had been employed. Thomson and Rankine had 

 shown theoretically in 1S52, and Giffard practically in 

 1S73, that if compressed gas be allowed to expand in a 

 cylinder doing work against a piston, the work done 

 externally is represented by a corresponding diminution 

 of the heat-energy of the gas. Similarly, if an iron 

 vessel containing highly compressed gas have the valve 

 opened, the contained gas is forcibly driven out against 

 the resistance due to the generation of a very high velocity. 

 The work of driving it out against this resistance is at 

 any moment being done by the expansion of that which 

 remains inside the vessel, and this remaining gas is 

 cooler in virtue of the work so done. This is the cooling 

 of a gas by work-expansion. In 1S77 Cailletet made use of 

 this method to give the first definite practical proof that it 

 was possible to liquefy oxygen, then known as a per- 

 manent gas. The vessel in which he had it enclosed 

 under high pressure was a strong glass tube of small 

 bore, which was surrounded by liquid sulphur dioxide or 

 nitrous oxide to give the compressed oxygen a pre- 

 liminary cooling. The opening of a water-valve then 

 allowed the gas to do the work of driving some water 

 forcibly through, and the gas, after this work-expansion, 

 was so much colder that part of it was condensed into a 

 visible, though evanescent, mist or vapour of oxygen. 

 In 1S84 Wrdblewski and Olszewski applied the same 

 method to hydrogen, using for preliminary cooling the 

 lowest temperature of liquid air under reduced pressure, 

 and obtained a similar result. Thus a combination of 

 cooling by work-expansion with preliminary cooling by 

 cascaded vaporisation succeeded in practically proving 

 that hydrogen could be liquefied, though it was not 

 possible by any such combination to keep, examine and 

 work with liquid hydrogen. 



But there is a third method of cooling a gas — that of 

 free expansion. In the case of the iron vessel containing 

 compressed gas, that gas which is at any moment 

 expanding from the valve is found to be colder imme- 

 diately after expansion than it was immediately before, 

 though it has, in the act of expansion, done no tangible 

 external work such as it does when expanding within 

 the vessel or behind a piston. It has, however, displaced 

 atmospheric air, given itself considerable residual mo- 

 mentum, and overcome the forces of intermolecular and 



NO. 1656, VOL. 64] 



intramolecular attraction. In virtue of this work done, 

 it has undergone some cooling, the cooling of free expan- 

 sion. This cooling from free expansion is much less for a 

 givenchangeof pressurethan that from work-expansion, or 

 that from vaporisation ; so much so that Thomson and Joule 

 had proposed no use for it, and great practical authorities, 

 Siemens and Coleman, had declared that nothing could be 

 accomplished by it — a judgment apparently confirmed by 

 the abortive result of Piazzi Smyth's persevering efforts 

 to utilise it. It is obvious, however, that if there were a 

 method of refrigeration in which the cooling could be 

 continually intensified by accumulation, this method 

 would have a great advantage and would lead ultimately 

 to lower temperatures than other methods which had the 

 benefit of greater initial cooling. This proved to be the 

 case with the method of free expansion. In 1S94 Hamp- 

 son proposed to intensify continually the cooling on this 

 method by accumulating in the compressed gas to be 

 expanded all, or nearly all, of the refrigeration produced 

 by the free expansion of previous portions. This was to 

 be done by letting the compressed gas expand through a 

 nozzle or valve from one end of a long tube and making 

 all the gas, when expanded, immediately return over the 

 tube which it had previously traversed as compressed 

 gas towards the expansion-valve. In the course of this 

 return it cools the succeeding portions of compressed gas 

 which are flowing past it inside the tube, so that as they 

 pass to the expansion point they contain all the cooling 

 which has been previously effected. Thus the com- 

 pressed gas is continually expanding from a lower 

 temperature than before, and is consequently reaching, 

 with the added expansion-cooling, a lower temperature 

 than had been reached by previous portions of expanded 

 gas. This intensification goes on until the cooling is 

 great enough to liquefy a small portion of the expanding 

 gas. The losses in this system are due to imperfect 

 interchange of temperature between the compressed and 

 the expanded gas and to the penetration of external 

 heat, so that its performance depends on the effi- 

 ciency and compactness of the interchanger or counter- 

 current accumulator. This method of obtaining in- 

 tense refrigeration involves the combination of free 

 expansion of gas (not liquid) with intensification 

 by counter-current interchange. Hampson constructed 

 and worked his apparatus in 1896, and in its 

 present form it begins liquefying air in less than ten 

 minutes without employing auxiliary refrigerants. A 

 process involving substantially the same combination 

 was invented at or near the same time by Linde, who, 

 in 1S95, succeeded in liquefying air with it in fifteen 

 hours. His form of apparatus has since liquefied air in 

 two hours, but requires auxiliary refrigeration in the form 

 of ice and salt or a subsidiary ammonia machine. 



The special advantage of the Hampson or Linde 

 method for the liquefaction of hydrogen is that it can 

 take gas at an initial temperature from two to three times 

 as high as its critical temperature, and cool it progres- 

 sively to the point at which it condenses continuously, 

 without the assistance of any substance boiling below its 

 critical temperature — a condition which had been the 

 stumbling-block of the methods employed by Wrdblewski, 

 Olszewski and Dewar. With such an appliance avail- 

 able it would seem that the last difficulty in the way of 

 liquefying hydrogen had been removed. Joule and 

 Thomson, however, had observed that hydrogen, on free 

 expansion, instead of being cooled, is actually heated a 

 little. But they had also observed facts which showed 

 how this difficulty could be overcome. The amount of 

 cooling on free expansion varies with the expansion- 

 temperature and with the nature of the gas. Firstly, as 

 to temperature. The lower the initial temperature the 

 greater the cooling for a given degree of expansion, the 

 variation being inversely proportional to the square of 

 the temperature on the absolute scale. Thus, for every 



