1908] on the Garners of Positive Eledricitij. 193 



escence due to the canal-rajs ; this might be explained by supposing 

 that it takes time to liberate sufficient hydrogen to produce appreciable 

 Canalstrahlen. I have made many experiments on this lag and these 

 show that it has no special connexion with hydrogen, but is due to an 

 alteration in the pressure of the gas produced by the discharge. It is 

 well known that the Canalstrahlen are only well developed when the 

 pressure in the tube is between certain limits. It is only when the 

 initial pressure is near to, but outside, one of these limits that the lag 

 occurs, and then the alteration in pressure which occurs when the dis- 

 charge passes may accumulate until the pressure is brought within 

 the required limits. That this, and not the introduction of hydrogeni 

 rather than any other gas, is the explanation of the lag is I think; 

 proved in the following experiments. If the presence of hydrogeni 

 were all that is wanted for the Canalstrahlen, then the lag should not 

 occur when the tube is tilled with hydrogen : we find that the lag 

 occurs when the tube is filled with hydrogen, as well as when great pre- 

 cautions have been taken to remove this gas from the tube. Again, 

 in a tube from which hydrogen has been removed and the lag is well 

 developed, the admission of a small quantity of dry air will remove 

 the lag just as effectively as the admission of hydrogen. When once 

 the lag has been got rid of it is necessary to give the tube a long rest 

 from the discharge before it returns. The fact that the lag may be 

 destroyed by admitting a small quantity of gas shows that it is [due to 

 the alteration in pressure and not to a change produced by the 

 discharge in the surface of the electrode. This can also be proved 

 in the following way : two discharge-tubes A and B are connected 

 together and with the pump, and the pressure is adjusted so that both 

 A and B show the lag ; then if the discharge is sent through A until 

 the lag disappears from that tube, it will be found to have simul- 

 taneously disappeared from B, though no discharge has been running 

 through this tube. 



It is somewhat remarkable that we do not, when the tube is filled 

 with oxygen, get any trace in the Canalstrahlen of particles having 

 masses comparable with those of the ions in oxygen. For though 

 such ions would not be formed in very intense electric fields, there are 

 places in the discharge-tube where the electric field is weak, as, for 

 example, outside the cathode dark space ; we might expect positive 

 ions to be formed in these regions, and then dragged by the electric 

 field up to and through a perforated cathode mingling with the 

 Canalstrahlen. The reason that we get no evidence of these oxygen 

 ions in the Canalstrahlen is, I think, as follows : Let A be a positive 

 ion, B a corpuscle, and let the relative velocity of A and B at the 

 instant under consideration be at right angles to A B and equal to Y. 



Then it is easy to show that A and B will not part company if —^ 



is less than -— = , where m is the mass and e the charge of the cor- 

 Vol. XIX. (No. 102) o 



