* Langmuir, , J. Amer. Chem. Soc. 34 (1912), 1310CrossRefGoogle Scholar; Langmuir, , J. Amer. Chem. Soc. 37 (1915), 417CrossRefGoogle Scholar. He has also (General Electric Rev. 29 (1926), 153Google Scholar—references to earlier work are given here) deduced values for the heat of dissociation from some measurements (Langmuir, and Mackay, , J. Amer. Chem. Soc. 36 (1914), 1708)CrossRefGoogle Scholar of the heat loss from tungsten filaments at high temperatures in the gas.

† To show that these differences between the present and the earlier results were not due to the tungsten, Langmuir's experiments were repeated, that is the hydrogen atoms were trapped on glass cooled in liquid air. A marked fatigue effect was observed, and it was found that, when the glass surface had taken up an amount of hydrogen corresponding roughly to a monolayer, no more gas disappeared. This would account for the variable results obtained by Langmuir. The fastest rates of disappearance when a fresh glass surface was used were of the same order as the fastest rates observed by Langmuir for a given filament temperature.

* Johnson, , J. Franklin Inst. 207 (1929), 629CrossRefGoogle Scholar; 210 (1930), 135.

† The oxide layer produced in this way was deep royal blue in colour. In some check experiments (see later) tungsten oxide was used and deposited in the same way. It was also blue. Both oxides turned yellow when the bulb was heated in air to about 300° C. This suggests that the oxides initially deposited are not the trioxides, which are yellow, but lower oxides or mixtures of oxides; for example the colour of W₂O₅ is stated to be deep royal blue.

‡ All taps were lubricated with Apiezon grease L.

§ International Critical Tables, 5 (1929), 245.Google Scholar

∥ Since it seemed possible that water might be produced by the interaction with hydrogen atoms, preliminary experiments were carried out in which most of the bulb was at room temperature but a small part at the bottom was cooled in liquid air so as to condense any vapour. After a large amount (0.03 c.c. at normal temperature and pressure) of hydrogen had been made to combine with the oxide the liquid air was removed and it was found that no appreciable amount of water vapour was evolved. In the actual experiments therefore no liquid air was used since it would have complicated the calculation of the number of molecules disappearing from the measured pressure decrease and the volume of the system. It may be mentioned that with the whole bulb immersed in liquid air the hydrogen atoms were not trapped quite so efficiently as when it was at a high temperature. This suggests that there is a small energy of activation for the trapping process.

* Langmuir, , MacLane, and Blodgett, , Phys. Rev. 35 (1930), 478CrossRefGoogle Scholar. In the present experiments the energy carried from the filament by the hydrogen did not in any case exceed two per cent, of that lost by radiation and the temperature distribution in the filament was assumed to be the same as if it were in vacuo.