chemist, and though he was unable to take an actual part in experimental work he remained very closely associated with the investigation in hand from start to finish. The late Dr. John Shields who worked with him for some time said of him: "What Mr. Mond does not know of practical inorganic chemistry is not worth knowing." I persuaded Ramsay to write to Mr. Mond, and a cheque came back by return of post. If any condition was attached, it was probably the usual one made by Mr. Mond, that his name should not be mentioned.

A "Whitehead torpedo compressor" of the type then used on board ship was purchased, and also a suitable electric motor, and the whole plant was delivered at the College just before Easter, 1899.

We had made arrangements to erect it in a disused lavatory at the end of the lower corridor, the only space available in our over-crowded department. Baly and I proposed to do the work ourselves, and I showed Ramsay a sketch of the lay-out. He was not, however, mechanically minded, and I may now be permitted to tell one story against him. When I told him that I proposed to fix both the compressor and motor to a pitch-pine frame bolted to the concrete floor, he said that this was totally unnecessary. Professor Ostwald, of Leipzig, had purchased a similar plant, which he had seen when he last went to Germany. The motor and compressor were fixed to a frame, but it was certainly not bolted to the floor. This was actually the case; for we heard later that when the plant was started up, the compressor, a very ill-balanced machine, set up such a rattle that the plant on the frame began literally to chase the operator round the room, and only ceased from the pursuit when the leads to the motor broke away. Ramsay went away for the week, so Baly and I, taking no chances, fixed the frame to the concrete floor with six good lewis bolts.

I shall describe the liquid air plant in a later chapter; it probably saved our lives, and it certainly made life at College more nearly worth living. Also we were at last able to organize the collection of liquid air residues, and the residues from about 30 litres of liquid air were soon available. The gas obtained by the evaporation of the liquid air was, of course, mainly oxygen, and this was got rid of by burning with hydrogen. The gases were led by separate pipes into an iron tube, which was heated in a gas flame. The other end of the iron tube was connected to a condenser, a receiver for the water formed, the gas leaving the receiver entering a hard glass tube half filled with metallic copper, and half with copper oxide, and heated in a small tube furnace. This showed whether the gas leaving the apparatus contained excess of oxygen or excess of hydrogen, and, of course, removed such excess, leaving only nitrogen and inactive gas. The nitrogen was removed by passing the gas over lime-magnesium mixture heated in a hard glass tube, this mixture being

a much more rapid absorbent than magnesium alone. Finally the gas passed over hot copper oxide.

Once having obtained a sufficient quantity of liquid air residue it was but six weeks' work to separate and purify krypton and xenon. However the quantities of gas which we obtained were very small so that we were obliged to make a weighing globe of only 7 c.cs. capacity, and this often contained the gas under reduced pressure. We could thus work with 5 c.cs. of krypton or 3 c.cs. of xenon, weighing 0.015 to 0.012 gram of gas on a very sensitive balance, obtaining a degree of accuracy of o 5 per cent.

I have already described the general method of proceeding in fractionating the gases. When the gas consisted mainly of argon, it liquefied only under a pressure greater than atmospheric pressure. However, as we found later, krypton and xenon solidify at liquid air temperature, the former having a vapour pressure of 160 mms., the latter being practically non-volatile. Operating by the method described there was a tendency for the gas to condense in the stem of the distillation bulb and not in the bulb itself. In condensing the gas it was therefore usual to allow the bottom of the bulb only to touch the liquid on which the gas was condensing, and when the whole of the gas had condensed in the bulb to raise the vacuum vessel and measure it completely in the liquid air. On account of the low vapour pressure of the krypton and xenon, all but first fractions were allowed to evaporate into the pump, and finally when no more gas was taken away by the pump, the liquid air was removed and the small trace of gas was collected in the xenon storage tube.

After a number of operations such as have been described, we had obtained three fractions of gas, which we will call A, B, and C, of which the densities were, A = B = 47.55; C = 56.04;

32.07;

and from these we were trying to separate samples of pure gases of which, according to the periodic law and the theory of gases, the densities should be

Each of the fractions was divided into a lighter and a heavier fraction, and these were combined to form three fractions, K, L, and I as shown. Fractions K and L. were already fairly pure krypton, but there was a big rise in density between L and I, the latter containing a good deal of xenon. Fractions K and L were then fractionated separately as below:

The refractivities of these two fractions were also determined and the following values were obtained:

M 1.450
N I.449

The conclusion from these figures is stated in the paper in the Philosophical Transactions of the Royal Society in which the work is described in the words: "We considered therefore that since we had succeeded in separating by fractional distillation from a lighter and from a heavier impurity respectively two samples of gas which agreed in two distinct physical properties, we were justified in assuming that we had isolated a definite chemical substance."

No actual determinations of the ratio of the specific heats were made with fractions M and N, but with fractions K and L the value 1.673, and 1.659 had been obtained, from which it was clear that the gas was monatomic.

The collection of the xenon had been a slow and tedious process, single bubbles of gas being obtained at the end of each process of fractionating liquid air residues, and this material consisting mainly of krypton, for although this gas had a very much higher vapour pressure than xenon at liquid air temperature, the krypton appeared to disolve in the solidified xenon, from which it was very difficult to separate it. The gas rich in xenon was repeatedly condensed and the bulb containing the solid allowed to remain in communication with the pump, and exhausted as far as possible. Finally, rather less than 3 c.cs. of gas was obtained, giving a density 64.01. Before weighing this gas again it became contaminated with a trace of air, and its density fell to 62.9. However, after further treatment with liquid air the density rose to 63.69.

We did not repeat the determination of the density. The last determination had been made at the end of July, and we opened the following session. with an attack upon the problem of separating the helium-xenon-argon

mixture. The densities of krypton and xenon had been determined with sufficient accuracy to fix the positions of these gases in the periodic table, thus:

Nine years later Professor R. B. Moore, working in Ramsay's laboratory with the residue, said to be from 120 tons of liquid air, produced in the process of making liquid oxygen, obtained large quantities of krypton and xenon, and found for the atomic weights the values 82.9 and 130.2.

neon

I

N September, 1898, Baly's spectroscopic examination of the volatile gas had shown that it contained helium as well as xenon and argon, and as

our early attempts to separate the constituents of this mixture were quite unsuccessful, we turned our attention to the solution of the easier problem, the separation of argon, krypton, and xenon. This work was completed before the end of the summer session of the year 1899, and in the autumn of that year we returned to the study of the lighter gases, and carried out the experiments which I shall describe in the next chapter. These led to the conclusion that it would only be possible to obtain pure neon by condensation and fractionation of the mixture by means of liquid hydrogen, just as the argonkrypton-xenon mixture had been condensed and fractionated by means of liquid air.

...

The liquefaction of hydrogen was an incident in the research on the rare gases, and was at the time regarded only as a means to an end. When we had finally decided that it was only by means of liquid hydrogen that we could separate the neon and complete the research, Ramsay left the details to me. He seemed to take it for granted that I should succeed, and in the introduction to the paper in which I described the apparatus he wrote: " .. Dr. Travers undertook to design and make an apparatus which would produce liquid hydrogen, and the following account of his experiments shows that his hopes have been justified." He would take no credit for the work, but the encouragement which he gave me was most helpful. He knew that no slight effort was required to carry out the undertaking. My friends in the laboratory were somewhat sceptical as to the possibility of success, but were convinced that I would blow myself up. I must admit that the experience was most nerve-racking.

The history of the liquefaction of hydrogen down to the time of which I am writing was briefly as follows. Cailletet and Pictet, working quite independently in the year 1877, had shown that the permanent gases could be made to condense in the form of mist or spray by compression into a glass tube, cooling to a low temperature, and subsequent sudden expansion. It was generally admitted that all of the gases then known could be liquefied by this