MR. T. ROYDS ON THE CONSTITUTION OF THE ELECTRIC SPARK. 
337 
alternately at the top and bottom of the spectrum. However, in the case of the 
intense short lines of the spectrum there generally appear, in addition to these, fainter 
streamers, which are simultaneous with the strong ones at the opposite electrode ; the 
faint streamers rarely appear in the two lead lines shown and long lines in general. 
By comparing the positions of the commencements of the streamers relative to the air 
lines with the positions of the spectral lines in the stationary photograph we can 
measure the time from the passage of the air discharge when the streamers start, and 
find the sign of the potential of the electrode from which they start if we know the 
polarity of the electrodes before the spark passes. It is found that the first streamer 
starts at both electrodes practically simultaneously with the air lines. The intense 
streamers come from the electrodes when negative, and those coming from the 
electrodes when the potential is positive are faint or absent, except for the first 
streamer from the initially positive electrode, which is always quite strong. 
Schuster and Hemsalech discovered that all lines of the same spectrum did not 
indicate the same velocity. The measurements show that there are, as a rule, two 
velocities. The long spark lines form one class having the smaller velocity, the short 
lines comprising the second class with the greater velocity. Only in the case of 
mercury is the line with greater velocity a long one. 
Let us examine a typical case of a line of the first class, e.g. XX 4387, 4245 of the 
lead spectrum, and X 4260 of the bismuth spectrum shown in Plate 29, fig. 6. Consider 
only the streamers from one electrode. The first streamer does not reach more than 
about 1 millim. (in the actual spark) towards the middle of the gap, the second 
reaches farther and the third still farther, and so on until the streamers from the 
other electrode are met, beyond which they do not as a rule pass. The first streamer 
is almost vertical at its commencement and becomes more inclined as it extends 
farther out. The first streamer is followed by another intense streamer from the same 
electrode when the potential becomes negative (for when the electrode is positive the 
streamer is either faint or quite invisible), i.e., after a half period if the electrode 
considered is the initially positive one (the upper electrode in the photographs), or 
after a whole period if the electrode is the initially negative one. This second 
streamer begins at the electrode and is practically vertical throughout its length until 
it overtakes the first streamer, when it bends over more or less suddenly into a line 
continuous with the first streamer. The third streamer behaves similarly, bending 
over when it overtakes the second. In the majority of cases of lead and bismuth 
photographs the third streamer reaches to the middle of the gap. It is thus seen that 
the first streamer together with the latter portions of the successive streamers form 
a kind of envelope reaching from the electrode to the middle of the gap. The fourth 
and following streamers are similar to their predecessors. A careful examination 
shows that the streamers after the first have near their bases for a length of about 
4 millim. (in the actual spark) a large inclination which would indicate a velocity of 
about 170 met./sec. and have in this short length become less intense. The streamer 
VOL. CCVIII.—A. 2 X 
