862 Mr. Gr. A. Hemsalech and the Comte de Grainont on 



explanation of the emission of spark lines in. the arc must 

 account for the following facts : — 



(1) The emission of spark lines in the first phase ; 



(2) their symmetrical broadening, especially during 



the first phase ; 



(3) their abrupt cessation before the end of the second 



phase. 



(4) The emission as a sharp narrow line of X4481 by a 



stable arc in nitrogen. 



Let us first consider the case of the water arc at 80 volts. 

 With this low voltage no spark will leap across the 

 electrodes through the water however small the distance 

 between them be made, for, as is well known, the dielectric 

 strength of a medium increases as the pole distance de- 

 creases. Thus, at very small distances the electric field 

 acting between the poles can be made very great. If, for 

 example, the distance between the electrodes is 0*05 mm. — 

 namely, equal to about one-half the thickness of the water 

 film at the end of the first phase, — the acting electric 

 field would be 16,000 -^- s . But the electric fields set up in 



' cm . r 



this way maintain their high values only so long as no 

 discharge pasees between the electrodes, and with the 

 establishment of an arc the intensity of the acting field 

 would immediately descend to a much smaller value. Let 

 us now follow the course of events when an arc is struck by 

 bringing the electrodes into contact. Whilst the electrodes 

 are touching, the intensity of the electric field between them 

 is of course zero. But at the instant of their separation 

 a very powerful electric field is suddenly brought into 

 operation which, up to a pole distance of 0'05 mm., lasts only 

 for about 0*00005 second on account of the rapid upward 

 motion of the top electrode, as explained in § 4. Now thi& 

 short space of time is about the same as the duration of the 

 second phase of a powerful condenser discharge with an initial 



electric field of about 20,000 — (pole ends 2'5 mm. radius 



' ran M 



cm. 



and 10 mm. spark gap), so that it is quite conceivable that 

 also the average drop of potential along the path of the 

 current in the two cases fare and spark) will have approxi- 

 mately equal values. We should therefore expect the 

 characters of the spectra emitted by these two sources to 

 be similar. This inference is completely borne out by the 

 results of our observations. For not only are the spark 

 lines predominant during the first phase of the water arc 

 but they are also symmetrically widened, which is precisely 



