98 



Prof. J. J. Thomson. 



[Mar. 9, 



ratns had to be readjusted. When everything was fnsed together, 

 and there were no flexible joints, this had to be done by letting the 

 delivery tubes F, G dip into separate basins filled with mercury, and 

 then to raise or lower the level of the mercury in one or other of 

 these basins until the rates of flow of the gases into the two eudio- 

 meter tubes were approximately equal. 



It is not, however, necessary to have two vessels if the narrow 

 portions of the exit tabes are connected with the main portions by 

 flexible rubber joints turned under the surface of the mercury, as in 

 this case it is very easy to raise or lower the end of one or other of 

 the tubes without interfering with the rest of the apparatus. 



These flexible connexions do not, as I have found by direct experi- 

 ment, introduce any source of error, and add greatly to the longevity 

 of the tube. When everything is fused up and the connexions are 

 rigid, the shocks due to the explosion of the mixed gases are exceed- 

 ingly liable to break the exit tubes from off the main tubes, while 

 they are comparatively harmless when there is a flexible connexion 

 between the piece of the exit tube immediately under the collecting 

 tube and the rest of the apparatus. 



When the exit tubes had been adjusted so that the rates of flow 

 through the two tubes were the same, the mixed gases were emptied 

 out of the collecting tubes, which were refilled with mercury; the 

 water voltameter was placed in series with the steam tube, and the 

 coil again set in action. 



The steam which came up the collecting tubes condensed into hot 

 water which soon displaced the mercury ; the mixed gases collected 

 over this hot water, and were exploded at short intervals of time by 

 sparks from a small Wimshurst machine. The gases did not dis- 

 appear entirely when the sparks passed; a small fraction of the 

 volume remained over after each explosion, and {he volume which 

 remained was greater in one tube than in the other. 



The residual gas which had the largest volume was found on 

 analysis to be hydrogen ; the other was oxygen. Thus, by comparing 

 the volumes of the residual gases in the two tubes it could readily be 

 ascertained next to which electrodes the excesses of hydrogen and 

 oxygen were appearing. 



There are other differences in the behaviour of the gases in the two 

 tubes which, though less obvious than the difference in volume, are 

 quite as characteristic. One of these is the difference in the ease 

 with which explosions take place ; the gases explode much more 

 readily on the side at which the oxygen is in excess than on the 

 other. Another very characteristic difference is that when sparks 

 pass in rapid succession through the tube in which the oxygen is in 

 excess bright spangles often appear floating about in the gas, due, I 

 imagine, to the ignition of small pieces of platinum torn from the 



