I also showed that nitrogen, boiling in vacuo at 4 millim., gives the lowest possible temperature, reaching -225°, but that it can be used as a cooling agent only as far as -213°, for at lower temperatures, under pressures below 60 millim. Hg, it solidifies to a snowy opaque mass, which is a bad conductor of heat. It follows that liquid oxygen and air are the best means to obtain the lowest temperatures, for neither of them solidifies at all, even at the lowest pressures, and neither possesses transparency. I had already used these cooling agents in my former attempts to liquefy hydrogen; I then exhausted all possible means of obtaining the lowest temperatures without obtaining the desired results; for the temperature of -220°, i. ., the lowest which can be produced by means of liquefied air, proved to be above the critical temperature of hydrogen. On repeating my former experiments I had no hope of obtaining a lower temperature by means of any cooling agent, but I hoped that the expansion of hydrogen would be more efficacious, on account of the larger scale on which the experiment was made.

The quantity of the frigorific medium, viz., of liquid air or oxygen, did not exceed 2-3 cub. centim. under atmospheric pressure, and became considerably less by the use of the vacuum; accordingly, the glass tube which contained hydrogen was only 2 millim. in internal diameter. The phenomenon of liquefaction, or rather of sudden ebullition of hydrogen which appeared in the tube during the expansion, lasted only a fraction of a second, and required a relatively sudden but not complete expansion. The hydrogen, cooled by its expansion below its critical temperature, was at once heated in so narrow a tube to the temperature of the surrounding frigorific medium.

In the subjoined diagram (fig. 2) a represents the lower end of the steel cylinder, serving to liquefy oxygen or air; this cylinder is enclosed in a glass vessel (double, if oxygen be employed; triple, if liquefied air) which serves to receive liquid ethylene. The cylinder a is a component part of the apparatus represented in fig. 1, and is therefore marked with the same letter; but it was increased for these experiments to 200 cub. centim. in capacity. The glass tube f is destined for the liquefaction of hydrogen; the external diameter of this lower and wider part is 11 millim., the internal diameter is 7 millim. Within it I place a short glass tube, with very thin walls 6 millim. in diameter; it serves to isolate from the warmer walls of the larger tube the hydrogen which is considerably cooled by its expansion. The tube f is placed in a larger glass tube e, with thin walls, measuring 30 millim. in

internal diameter, and serving to receive the liquid oxygen under atmospheric pressure. The tube e is surrounded by two

others, of greater and greater diameter, and hermetically closed above with a brass plate g. The whole apparatus is hermetically placed in a larger glass cylinder, with calcium chloride at the bottom, serving to dry the enclosed air. The top of the tube ƒ is connected with the manometer b and the iron bottle c, containing absolutely pure and dry hydrogen, under a pressure of 150 atm.; the cock I serves to let the hydrogen out of the tube f, thus producing the expansion; through the upper end of the tube f, which is closed with the screw m, a thermoelectric junction may be introduced, if it be required to measure the temperature of the hydrogen at

the moment of expansion. I performed the experiment as follows:

By opening the cock d I let the liquid oxygen, contained in the steel cylinder a, into the vessel e; a part of the oxygen, which thus returned into the gaseous state, escaped with violence through the tube i; another was cooled down to its boiling-point (-181°4 C.) and collected as a bluish liquid in the tube e, to a height of 6-7 centim., so that the wider part of the tube f was plunged in liquid oxygen. I afterwards closed the cock d and joined the tube i to the pump, by slowly opening a cock, which is not represented in the figure; the mercury manometer h indicated the pressure of the liquid oxygen in the vessel e. Liquid oxygen behaves very quietly in the vessel e under atmospheric pressure, boiling quickly but uniformly on being pumped: if we do not reach very low pressures, the oxygen, after cooling according to the lowered pressure, boils again quietly. But if the pressure falls below 10 millim. (or less), the oxygen boils irregularly, the liquid is thrown up, and shortly disappears. To avoid this I introduced into the vessel e a capillary glass tube o, the lower end of which reaches to the very bottom of the vessel e, the upper end is connected with the iron cylinder k, containing dry hydrogen under a pressure of several atmospheres. During the pumping of the oxygen the cock of the cylinder k is little by little opened, and a slow stream of hydrogen is let through the liquid oxygen; by this means the liquid oxygen is constantly and easily removed, and it boils quietly, even if the pressure falls to 4 millim.

When the manometer h indicated a pressure of 10–4 millim. I introduced hydrogen into the tube f, by slowly opening the cylinder c, until the pressure became 140 atm., as indicated by the manometer b. When the hydrogen in the tube ƒ has come down to the temperature of the cooling agent, I little by little produced expansion, by opening the screw-cock 1. The phenomenon of hydrogen ebullition, which was then observed, was much more marked and much longer than during my former investigations in the same direction (3, 4). But even then I could not perceive any meniscus of liquid hydrogen.

I have remarked in these experiments, that with a slow expansion the phenomenon of sudden ebullition always appears under the same pressure, no matter how great the initial pressure may be, provided that value be not too low. So that by expansions made, beginning with the pressures of 80, 90, 100, 110, 120, 130, 140 atm., the phenomenon described constantly appeared at 20 atm.; but if the initial pressure was 70, 60, and 50 atm., the ebullition appeared at a lower and lower

pressure, viz., approximately at 18, 16, and 14 atm. I repeated the same experiment a good many times, and always obtained the same results. These experiments bring me to the conclusion, that the 20 atm. at which the ebullition of hydrogen always appears represents its critical pressure. If hydrogen, cooled by means of liquid oxygen, boiling in vacuo, to the temperature -211° C., which, we may suppose, is several degrees above the critical temperature of hydrogen, is submitted to a slow expansion from a high pressure, its temperature is lowered to the critical temperature, hitherto unknown. If the initial pressure is high enough-in my experiments it was above 80 atm.-then, by means of a slow expansion, the temperature of hydrogen sinks to its critical value, before its critical pressure is reached, and then liquid hydrogen will appear the moment we lower the pressure to its critical value. But if the initial pressure is too low, a slow expansion cools the hydrogen to the critical temperature only after the critical pressure has been passed: the lower the initial pressure is the greater is the expansion needed to cool the hydrogen below its condensing temperature. We may thus explain the changing pressures, corresponding to the phenomenon of ebullition or instantaneous liquefaction in the case of expansion from an insufficient initial temperature. And if the initial pressure is still lower, the instantaneous liquefaction will not appear at all.

To ascertain the truth of this statement I performed two series of analogous experiments with gases, the critical pressures and temperatures of which are accurately known, viz., with oxygen and ethylene. The critical temperature of oxygen is, according to my former researches, 118°.8 C., its critical pressure is 50-8 atm. In the same apparatus which I used for the experiments with hydrogen I cooled oxygen by means of ethylene boiling under atmospheric pressure (-102°.5), then to a temperature 16.3 degrees below the critical temperature of oxygen, and subjected it to a slow expansion, beginning with different initial pressures, from 40 atm. up to 100 atm. The ebullition of oxygen always appeared at a pressure of about 51 atm., provided the initial pressure was not lower than 80 atm.: at the same time there also appeared a meniscus of liquid oxygen. As the initial pressure became lower and lower, so did the ebullition pressure too.

The critical temperature of ethylene according to Dewar is 10°.1, the critical pressure 51 atm.; my own minations of the same quantities yielded results agreeing with the above-cited, viz., 10° C. and 51.7 atm.

I made similar experiments with ethylene, using the apparatus of Cailletet; one series at a temperature of 17° C., another at 27°; then at temperatures, which were first 7°, then 17° higher than the critical temperature of ethylene. During the first series of experiments, the ebullition of ethylene, and at the same time the meniscus, appeared constantly in consequence of a slow expansion at a pressure of about 51 atm., if the initial pressure was 70, 80, 90, 100, and 110 atm. During the second series of experiments the ebullition appeared at the same pressure, if the initial pressure was 100, 110, 120, and 130 atm. In proportion as the initial pressure was loweredin the first series below 70 atm., in the second below 100 atm.-the ebullition pressure was lowered too. I must, however, mention that in the apparatus of Cailletet, in which I made the experiments with ethylene, the conditions of ebullition of any gas by expansion are much less advantageous than in the apparatus described above.

Hence it follows that the determination of critical pressures by means of expansion is possible, even if the gases have a temperature which is several or many degrees higher than their critical temperature. This dynamical method of determination of critical pressure is really of no advantage if applied to the other gases, for these pressures may be more easily and precisely determined by the vanishing of the meniscus ; but with hydrogen it is the sole possible way to determine not only its critical pressure, but also its critical temperature.

On repeating these experiments in November 1891 I used liquid air, boiling under a pressure of 4-10 millim., as a cooling agent, and obtained the same results, with the only difference that the ebullition of hydrogen on expansion appeared still more distinctly and persisted somewhat longer.

The reason for which it has not been hitherto possible to liquefy hydrogen in a static state, is that there exists no gas having a density between those of hydrogen and of nitrogen, and which might be for instance 7-10 (H=1). Such a gas could be liquefied by means of liquid oxygen or air as cooling agent, and be afterwards used as a frigorific menstruum in the liquefaction of hydrogen.

The subjoined figure 3, taken from the original, represents my apparatus for liquefying large quantities of oxygen and air, connected with the apparatus serving to determine the critical pressure of hydrogen. The following brief description will help to understand the figure :

(a) The steel cylinder, 200 cub. centim. in capacity, for the liquefaction of oxygen or air.

(b) The glass vessel with triple walls, serving to receive liquid ethylene under diminished pressure.