390 Prof. W. (I. Duffield : A Comparison of the 



displacements are nearly, if not quite, the same, whether 

 produced in the spark or the arc discharge. 



6. The conclusion reached in 3 and 5 is interesting, because 

 it shows that the spark and arc changes of wave-length under 

 pressure are approximately equal for those lines which re- 

 main bright, and also for the emission centres of lines whose 

 vibrations are subject to absorption by the outer layers of 

 the arc or spark. It appears to be the condition of the 

 vapour as regards its pressure or density, rather than the 

 means whereby it is excited, which in the main determines 

 the displacements of the lines. Any effect due to length of 

 arc, self-induction or capacity (which have been stated to 

 influence the displacement of lines) appears to be superposed 

 upon this general effect, and is probably of a smaller order 

 of magnitude. 



7. We note the following additional points of interest in 

 the comparison of the arc and spark under pressure : — 



(a) The arc readings at 10 atmospheres were regarded as 



unduly high ; the agreement between bright spark 

 and arc lines which are only slightly broadened, 

 and also between those which are greatly broadened 

 towards the red, points to this being a real pres- 

 sure effect and not a spurious one due to the 

 personal equation of the observer. 



(b) The rate of displacement of arc lines decreased as 



the pressure was increased from 10 to 100 atmo- 

 spheres or more. Bilham's values for the spark 

 are relatively higher at 5 than at 10 atmospheres,, 

 which points to the same conclusion. 



(c) The continuous background is much more noticeable 



in the spark than in the arc spectrum at the same 

 pressure. In the former, especially in the regions 

 of small wave-length, the spectrum looks like a 

 continuous spectrum crossed by a number of 

 absorption lines. 



((/) Though the continuous background makes the re- 

 versals more prominent in the spark, they are not 

 more numerous nor appreciably broader than those 

 in the arc under pressure. There is remarkable 

 agreement between Bilham's photographs and my 

 own in this respect. 



(e) Certain lines in the spark spectrum are much more 

 broadened by pressure than in the arc spectrum, 

 e. g. 4401-70. 



( /') There is satisfactory agreement between the four iron 

 lines appearing as impurities in the nickel arc and 

 spark under 10 atmospheres pressure. (Table III.) 



