1896.] and, on the zigzag path of Lightning. 219 



Position of the electrodes, Angle of the short tubes and Occluded 

 Gases. When the short tubes containing the electrodes were fixed 

 obliquely, it was found that the spark passed along the limb 

 which allowed the negative current the more direct course into 

 the main tube. 



This took place with all gases, including Oxygen and Hydro- 

 chloric acid, which had no effect in the preceding experiments. 



When however the knobs were pushed forward, the old form 

 of reversal was obtained, Oxygen giving poor results and Hydro- 

 chloric acid none at all. 



Since reversals were obtained with electrodes that occluded 

 gases and not with Aluminium, it was of interest to try by 

 experiment whether such gases produced any part of the effect. 

 Three platinum wires with knobs at the ends were fixed to an 

 ebonite rod. The distance between the first and second was 

 exactly the same as that between the second and third. The first 

 and third were joined together, the whole placed under a receiver 

 and exhausted. Afterwards Hydrogen was liberated electro- 

 lytically upon No. 1 and Oxygen upon No. 3, but on replacing 

 and exhausting no difference could be detected. 



Charging of the tube. Supposing that it was possible that the 

 glass tube might become charged and so influence the spark, I 

 tried the effect of increasing such charge by placing strips of 

 metal under one limb of the tube and found that the discharge 

 was drawn down the side on which the metal was placed. 



By using two pieces of metal, or two small saddles of lead 

 joined by a wire, and placing one at C (Fig. 1) and the other at D, 

 or one at E and the other at G, I found that the spark could be 

 made to pass down either side and to reverse when the current 

 was changed. The direction being from the negative electrode 

 past the nearest saddle. Thus if the negative were on the right 

 hand and the saddles at E and G, the spark passed along the limb 

 A and on reversing along the limb F, but if the saddles were at D 

 and G it passed through F and reversed through A. 



We could thus produce the reversals as we wished and part of 

 the original phenomenon was explained. To determine how much, 

 two other pieces of apparatus were tried, but as both methods 

 gave the same results I will describe one only. 



A jar 6 inches in diameter and 8 inches in height was placed 

 upon a metal plate and made tight with soft wax. The two lower 

 electrodes A and B were in metallic communication with the 

 plate, the upper ones C and D were connected together outside of 

 the jar. D was fused into a glass tube which could be moved, 

 backward or forward, in order to make the distances exactly equal 

 (3 inohps). 



Revorsals were observed between the pressures 45 mm. and 



17—2 



