346 MR. T. ROYDS ON THE CONSTITUTION OF THE ELECTRIC SPARK. 
ADDENDUM. 
[Note added April 3.—The drawing reproduced on p. 338 supra has been prepared 
at the suggestion of one of the referees, and shows, with all the clearness of the 
original negative, the first edge of each streamer of X 4387 in fig. 5 (Plate 29), which 
contains few accidental irregularities. 
The referee suggests that the positive streamers are curved and the negative ones 
straight, this explanation bringing the phenomena of the spark more into line with 
those of the arc. 
The presence of positive streamers in the long lines is exceptional, but in all cases 
the edge of what I have considered as one negative streamer forms a perfectly 
continuous line throughout its length. In several photographs instances occur where 
the positive streamer runs side by side with the negative (e.g., in X 4387 of fig. 5, the 
5th + and 6th — at the upper electrode), the bend in the negative streamer taking 
place before the positive has met it. 
If we are dealing with one of the negative streamers which fuse together into an 
envelope, i.e., in which the streamer has again raised to incandescence the vapour 
produced at the commencement of the spark, the streamer ceases to be a pulse of 
luminosity on overtaking this vapour which is leading the way to the middle of the 
spark gap and must become inclined, since the vapour is travelling slower than the 
pulse of luminosity. It therefore seems to me to be not the simplest explanation to 
ascribe this inclination to a positive streamer previously almost invisible. 
In those later pulses of luminosity which do not overtake the diffusing vapour it 
would be difficult to explain why the pulse from the negative should stop dead in the 
middle of the vapour at a distance shorter than that to which previous pulses have 
extended, and how the positive pulse, previously faint or invisible, should at this 
instant burst into intense luminescence. 
As regards the sinuosity of the envelope, the points of discontinuity are where the 
streamer meets the negative streamer and not the positive, as the referee’s suggestion 
requires, e.g., at the points in the drawing marked A and B, but not at C. My 
present opinion is that the cause of the sinuosity is the alternation in the temperature 
of the metallic vapour. 
The crossing point of the two envelopes in fig. 5 is very slightly nearer the upper 
electrode (2 millims. in an enlargement in which the width of the spectrum is 
68 millims.), but this is by no means a rule ; in general the inclination of the envelopes 
is the same at the two electrodes. 
Photographs have already been taken with larger self-induction, but they do not 
give any information on the point, for the streamers die out before the following ones 
overtake them. 
