Till: ( oNVniCTION OF TIM! KI.KITKIC M'AIIK. 



201 



The figure is enlarged in the ratio of about 4'2. 



19, Plate 10. Disc spinning with linear velocity at centre of spectrum, 

 v = 99 met. /sec. The air lines are seen to run straight across the spectrum, but the 

 metallic lines are curved ;m<l broadened. It follows that the metallic vapours remain 

 luminous longer than the air lines, and that the metallic particles are projected with a 



Fig. n. 



measurable speed from the electrodes. The broadening of the nitrogen doublet at 5004, 

 marked N, in the photographs, only amounts to about '04 millim., which limits the 



luminosity as far as that line is concerned to ^r: n = 4 X 10~ 7 . The air line at 3995, 



y t uuvj 



and marked N..,, is drawn out rather more, but is thinner near the poles than in the 

 centre of the spark ; its appearance, which in some of the photographs taken is even 

 more marked than here, is illustrated in fig. 11. We must conclude that the air 

 remains luminous in the centre of the spark rather longer than near the poles. This 

 fact, which is very apparent for the luminosity of the metallic vapours, will l>e 

 referred to again further on (p. 210). 



The displacements on this photograph being the first that were actually measured, 

 the best methods of drawing the reference lines (see p. 198) had not been adopted, 

 and these lines were therefore at unequal intervals. In the tabular arrangement of 

 our results we give in the first columns the coordinates of the curved lines which 

 have actually been measured ; x denoting the horizontal displacement at a distance y 

 from the edge of the spectrum. If y t , y 2 ; x t , x t are coordinates of two points which 

 are near together, and v is the linear velocity in metres per second, the molecular 



speed is V = Kr ;/ ' (p. 197), at a distance h = K *LJ from the pole, where K is 

 s* ~~ x i 



the factor correcting the optical reduction, which is 1'16 in all our photographs. 

 The displacements being difficult to measure, a small error, in x l and x 2 near the pole, 

 may produce very large differences in the result. For the comparison of different 



metals with each other we therefore calculate also the quantity V = Kr -' which is 



the average speed tetween the pole and a point at a distance h' = Ky from the pole ; 

 VOL. CXCIH. A. 2 D 



