TRANSACTIONS OF SECTION A. 637 



8. Further Experiments on the Constitution of the Electric Sj>ark. 



By T. RoYDS, M.Sc, 



The results obtained by photographing on a rapidly moving photographic film 

 the spectrum of the ordinary oscillatory spark with small self-induction have 

 already been communicated to the Royal Society. The method has now been 

 applied to study the spark when the self-induction of the circuit is gradually 

 increased. On account of the feebleness of the light in these cases it has beeu 

 found necessary to remove the prisms, photographing directly the image of the 

 slit, in order to avoid the loss of light due to dispersion by the prism. It is still 

 possible under these conditions to differentiate by their character the streamers 

 through the metallic vapour from those through the air. 



Condensers, whose capacity amounted to about one-third of a microfarad, 

 were charged from a large Wimshurst machine, and the spark of 8 mm. length 

 passed between metal electrodes. Self-inductances ranging in value up to 0026 

 henries could be inserted in the circuit. The velocity of the photographic film 

 was about 100 metres per sec. 



The commencement of the spark is marked by a sudden and almost instan- 

 taneous luminosity of the air, showing on the photographs as a narrow vertical 

 line. The duration of tbe luminosity is not appreciably affected even when the 

 period of the circuit reaches a value of about ^ x 10"* sees. This initial air di.s- 

 charge does not start qu: .e simultaneously at all parts of the spark length. It is 

 followed by another type of streamer through the air in which the duration is 

 comparable with the interval between the oscillations ; in some cases the oscillation 

 is divided into more than one air-streamer. 



After the initial air-discharge has passed, the light from the spark is chiefly 

 due to the metallic vapour which is produced. The moment of vaporisation of the 

 metal is simultaneous with the passage of the initial air-discharge. The first 

 streamer in the metallic vapour often reaches to the centre of the spark. The 

 slope of this streamer gives the velocity with which the vapour first produced is 

 moving along the line joining the two electrodes. The luminosity of tbe metallic 

 vapour afterwards consists of streamers starting from the electrodes. At the instan- 

 taneously positive electrode the luminosity of the streamers is of longer duration 

 than at the negative ; the streamers are not so numerous nor so distinct. There is 

 not, however, the marked difference in the intensity of the streamers at the instan- 

 taneously positive and negative electrodes which was noticed in a previous paper 

 on the spark with small self-induction. It is again found, however, that at the 

 commencement of the spark, vapour is produced at both the positive and negative 

 electrodes. Dark spaces at the electrodes separate the oscillations of the circuit 

 and serve to measure the period of the oscillations. 



The Velocity of the Metallic Vapour.— The measurements of the photographs 

 show that the velocity of the metallic vapour first produced is smaller in the spark 

 with self-induction than in the ordinary spark. In the case of mercury, for 

 example, the velocity becomes 620 metres per second ; the velocities found in the 

 ordinary spark are 1,150 and 940 metres per sec. for two different types of spectral 

 lines. This is more probably due to a reduced temperature rather than to the 

 supposition that the quantity of metal vaporised is not sufficient in the former case 

 to produce the difference of pressure required to make the velocity of diflusion 

 attain its maximum value. The introduction of self-induction brings out the ' arc ' 

 lines of the metal. The evidence now obtained that the arc lines are due to, or 

 accompanied by, a lower temperature confirms Lockyer's supposition which forms 

 the basis of his researches on star temperatures. 



When the self-inductance of the spark is increased, it is seen that during a 

 single oscillation several streamers start from the electrodes ; they are more dis- 

 tinct with magnesium than with lead, bismuth, or mercury. The streamers are 

 too numerous and close together to make useful the supposition that they are due 

 to harmonic overtones of the fundamental period of the circuit. The streamers 

 are found to recur at definite intervals after the commencement of each oscillation 



