Discharge in Rarefied Groses. 237 



iely coincide with the direction of the axis of the cylin- 



.: portion of the tnbe follows from the previously-mentioned 

 fact, that the surface extends towards the positive side a little 

 beyond the section of the negative branch of the tnbe by the 

 positive branch. Observations made with a tnbe to be de- 

 scribed further on. show that the deviation of the rays from 

 the direction of the axis is equal in every direction: whence it 

 follows that the rays of the positive light are not parallel to 

 each other, but form a conical bundle of small angle whose 

 section is circular. 



"We therefore regard the phosphorescent surface in the 

 bends as the luminous sections oi the column of light which 

 traverses the negative branch of the bend by the wall of the 

 tube : and we have arrived at the following general view of 

 the positive light in highly exhausted cylindrical tubes : — 

 1 ie positive light in high, are/led ise-s :: isists of reciili- 

 raus which are propagated/ : the m jatioe to the positive 

 The rays fo\ : : ical bundle of small angle whose axis 



is the axis of the cylindrical portion of the tube; where this 

 bundle cuts the wall of the tale, the parts of the rays in imme- 

 diate contact with the wall excite phosphe •eseent light in it. 



The analogy which these properties show exists between the 

 rays of the positive light and the kathode-rays is obvious. I 

 have inquired whether this analogy loes not go so far as to 

 amount to a negation of the characteristic property hitherto 

 always ascribed to the positive light, according to which it 

 invariably takes the shortest course from the boundary turned 

 towards the kathode to the anode. Although the katkode-light 

 radiates its rays without reference to the position of the anode, 

 Eh . : the positive light is essentially conditioned by Hie 

 position of the anode. 



Tubes of the form oi rig. 10 (PL IV. ) -ervei for the trial 

 whether the behaviour of the positive light corresponded in fact 

 with 'he usual assumption. Th • wh lie vessel, with the exception 

 of the ellipsoidal bulbs A and B. is made out of the same piece 

 of glass tubing. Care was taken that there was no change in 

 section at the pom:- a. 3. and y. where the tubes joined. 



In a particular case the tube was 1 centim. wide, the dis- 

 tance- d the points a and ,3 and j3 and 7 was centims., 

 and the " blind alley " prolongations of the tnbe X, X, X had 

 each a length of 2 eentims. from the rectangular junction. 



If the tube is highly exhausted and B be made the kathode, 

 the discharge presents the following appearance. 



The positive light (lilac at this pressure, and blue at the 

 greatest exhaustion possible) extends from B through the 

 branch 1 01 the tube till its rectilinear rays impinge on the 



