344 
MR. T. ROYDS ON THE CONSTITUTION OF THE ELECTRIC SPARK. 
vapour, or somewhere in the middle of the spark gap, does not appear at first sight 
from these considerations. The photographs show, however, that generally the 
vapour at the electrode becomes luminous first, but occasionally the luminosity is 
propagated also from a point within the gap both towards the centre of the gap and 
towards the electrode. The average velocity with which the luminosity is propagated 
through the vapour of bismuth has been found to he 800 met./sec. for X 4260 and 
2200 met./sec. for X 4561. In the case of mercury the velocity of the vapour is larger 
than the velocity of propagation of luminosity through the vapour. 
The luminosity which produces those portions of the streamers not forming part of 
the envelope is probably propagated in a manner similar to that of the luminosity 
of the air. In the moving photographs the air lines are not inclined sufficiently to be 
measurable. But this does not mean that the air molecules are moving with an 
exceedingly high velocity, and we should not expect a Doppler effect commensurate 
with such a velocity, for the air rendered luminous in the middle of the gap probably 
has not come from the electrodes. So in the atmosphere of metallic vapour in which 
streamers may be produced by the oscillations of the discharge we should expect to 
find streamers which are not due to a velocity of particles, and which will probably be 
steep. f 
8. Duration of the Luminosity .—The air lines are slightly broadened in the 
moving photographs. The amount of broadening shows that in 6'9x HU 7 second the 
intensity lias become insufficient to affect the photographic film. Schuster and 
Hemsalech found that the air line X 3995 was thinner near the electrodes than in the 
centre of the spark. Now the air lines are generally fainter at the electrodes, and 
the apparent broadening of X 3995 is seen to be due, at any rate in part, to two faint 
air lines near it, generally distinguishable in the moving photograph, but which at the 
electrodes are too faint to appear. 
Stationary photographs measure the total amount of light falling on the plate 
whilst the spark is passing. Whether a line appears or not in the moving photograph 
depends not only on the intensity of the light, but also on the length of time for 
which the luminosity endures. Thus a line of short duration may be visible on the 
photograph when a line equally strong on the stationary, but of longer duration, may 
disappear. From their relative intensities in the moving photographs it is found that 
lines in the spark which also occur in the arc are of greater duration than the spark 
lines, and that the longer the line the greater is its duration, the short lines appearing 
also much sharper and finer than the long lines. It may be noted that Schenck, who 
measured the interval between the times when a line first appeared and when it last 
appeared, found this interval greater for the arc than for the spark lines. 
9. Summary of Conclusions. 
I. The velocity of the metal vapour is given by inclination of an envelope formed 
by the meeting of the first few streamers. 
