ME, T. ROYDS ON THE CONSTITUTION OF THE ELECTRIC SPARK. 
345 
2. All lines of the same spectrum do not indicate the same velocity. They may be 
divided into two classes :— 
(i.) Long lines, indicating the normal velocity ; 
(ii.) Short lines, indicating a greater velocity. 
3. The velocity is greatest near the electrodes, and falls away to a nearly uniform 
velocity. 
4. The velocity of lines of type I. is probably due to diffusion ; that of type II. 
and of the mercury lines is too large to be explained in this way. 
5. The envelope indicating the velocity of the vapour is occasionally sinuous, 
indicating that the particles are in these cases charged. 
6. The velocity varies little with the spark voltage or with the self-induction ol the 
circuit. 
7. The velocity is apparently dependent on the amount of matter vaporised, and is, 
generally speaking, less for a metal of higher boiling-point. 
8. The luminosities of the streamers of the short or “spark” lines are of shorter 
duration than those of the long or “ arc ” lines. 
9. The streamers start at both electrodes, but are stronger when the electrode is 
negative, except the first streamer from the initially positive electrode. The streamers 
from the positive electrode are but rarely seen in the lines of longer duration.* 
10. The period of the circuit is practically independent of the material of the 
electrodes. 
My best thanks are due to Professor Schuster for his interest throughout the 
research and for many valuable suggestions ; also to Professor Rutherford and the 
staff of the laboratory for numerous suggestions. I have been indebted to Mr. C. 
Riley, B.Sc., for his careful assistance in making the experiments and measuring the 
photographs. 
- [*April 3 .—In the paper as originally presented this sentence ran as follows :—“ 1'he streamers from the 
positive electrode are only seen in the lines of short duration.”] 
2 Y 
VOL. CCVIII.-A. 
