T/fK EFFECT OF GASKS 147 of the experiments fire in complete disagreement with the results of experiments made with well-glowed-out carbon filaments by Deininger1 in 1908, and of the more recent ex- periments of the writer awl of Langmuir. In the last two cases the precautions described on p, 14 were taken in pre- paring the bulbs, and although it is not claimed that every trace of gas was got out of this very difficult substance, the conditions were much better in this respect than in the ex- periments of Pring and of Pring and Parker. The same claim can almost certainly be made for Deininger's work, In fact, it is quite impossible to attain good vacuum conditions with the large quantities of hot material and the other arrangements used by Pring, Taking all the facts into consideration they appear to the writer to afford no support to the contention that the emission from carbon has anything whatever to do with chemical action. It may be that such a chemical effect exists, but its existence is not demonstrable, or even rendered probable, by the evidence which has been submitted. Thus there is no case In which it has been established with certainty that chemical action is the direct and immediate cause of an emission of thctwns* The majority of chemical actions between solids and gases certainly do not give rise to electrical effects of this kind to any appreciable extent (see Chap, IX), The only case in which the evidence renders the occurrence of electron emission as a chemical effect probable is that of the alkali metals. The experiments of Haber and Just do, on the whole, indicate a balance of probability in favour of a direct chemical effect in this case, although, in the judgment of the writer, they cannot be held to establish it with certainty* The advocates of the chemical point of view have, held that the emissions usually observed are due to actions between the hot metal and minute traces of residual gas and not to chemical actions on any considerable scale. In support of this it may be urged that, an the cflbct is a purely .superficial one, a small quantity of giw will exert as large an effect m a 141 Ann,