Curves of the Oscillating Spark. 719 



spark-gap, e the ionic charge, Y the velocity of the negative 

 ions, Fthe field, and the small fraction of the current carried 

 by the positive ions is neglected. N is unknown at the 

 temperature of the spark, but we can approximately take 

 NV as uninfluenced by changes of density and temperature, 

 and assume for it the same value as for the negative ions in 

 flames. At 1800° C, N would be 2"75x 10 19 x 273/2073; and 

 according to the measurements of Gold*, V in flames at 

 1800° lies between 8000 and 13,000 (say 10,500) cm. per sec. 

 per volt per cm. Also e = l'bl x 10~ 19 coul., and for F we 

 may put the rate of increase of the arc p.d. in the spark 

 with the spark-length. This has been mea-ured by Roschansky 

 for numerous electrodes, and averages 10' 5 volts per cm. 

 With these values we get 



C = 63 ; 000 amp./sq. cm. 



as an estimate of the maximum current density which the air 

 is capable of carrying with the field which exists in the spark. 

 This is of the same order as the experimental number, and 

 no more is of course to be expected. 



2. Spectra of Core and Sheath. — The three spectra in 

 fig. 10, PI XX., are those of the image of a horizontal spark 

 between copper electrodes thrown on the vertical slit of a 

 spectroscope. They refer to sparks with a capacity of "017 mf ., 

 and inductances of 1*3, 40, and 170 mcrh. respectively. An 

 examination of (a) shows that some of the lines (such as 

 that marked «, the hydrogen line 6563) are distinctly longer 

 than other lines (e. g. /3, the nitrogen line 5003), even 1 hough 

 the latter may be the stronger of the two. The short lines 

 represent the spectrum of the core of the spark, the longer 

 those which exist in the sheath as well (the very long lines 

 are due to Cu vapour from the electrodes). The line of 

 demarcation is really more distinct than appears in the 

 photographs, since the exposure required about 20 sparks, 

 and their images cannot all be made to fall on exactly the 

 same place on the si t. As the inductance is diminished the 

 core lines diminish in length (fig. l>) as the core diminishes 

 in diameter ; and they finally disappear altogether, leaving 

 the spectrum shown in fig. c, which is quite different from 

 the core-line spectrum. An account of the last spectrum, 

 which corresponds to such inductance that there is no 

 differentiation of the spark into core and sheath, is given by 

 Hemsalechj, who has called it the "second air-line " spectrum. 



That there would be a difference in the spectra of the core 



* Proc. R. S. lxxix. p. 43 (1907). 



t Heuisalecb, Comptes Bendus, clii. p. 1007 (1911). 



