of Magnesium, Copper, and Manganese. 299 



state that graphite plates vary considerably with regard to 

 the nature and quantity of impurities contained. Thus it 

 often happens that of two plates, although cut from the same 

 stratum, one will be quite free from iron, whereas the other 

 will show a highly developed spectrum of this element. 

 It is therefore essential, when comparing photographs of 

 fringe emissions as given by different plates, to choose only 

 such records as spectroscopically contain similar amounts of 

 impurities. 



What complications may arise through the presence of 

 certain impurities is further illustrated by the fact that in 

 their absence the fringe emission is but slightly affected by 

 external magnetic fields. Thus, in the case of graphite 

 plates free from spectroscopic traces of iron, but containing 

 small amounts of Ca, Sr, and Ba, a magnetic field of from 

 1500 to 2000 C.G.S. units is required in order to appreciably 

 displace the fringe emission. Much greater displacements 

 are obtained in the presence of large amounts of iron 

 vapour, derived from boiling iron metal on the top of the 

 plate. The results of these new observations help to explain 

 the relatively great displacements which I obtained for the 

 fringe emission with feeble magnetic fields. The consider- 

 able amount of luminous vapours driven out by the carbor- 

 undum and passing through the protected space beneath 

 the plate formed a highly conducting medium through 

 which the thermelectronic current was able to diffuse. 

 Further, the swarms of electrified particles resulting pro- 

 bably from the breaking up of a potassium compound, and 

 the true nature of which had then not yet been recognized, 

 were largely instrumental in bringing about the drawn-out 

 appearance of the red fringe emission. The positively 

 charged particles are especially affected by magnetic forces, 

 and on being drawn away from the plate the space through 

 which they pass becomes strongly ionized. On this view 

 the great displacements of the fringe emission would be 

 caused by the diffusion downwards of the thermelectronic 

 current, which, helped by the external magnetic field, 

 would tend to follow the track of the electrified particles. 



As a result of this new development of the subject, the 

 explanation which I have given respecting the sharp outline 

 of the red fringe and its spectrum emission becomes 

 untenable, except perhaps in the case when free electrified 

 particles are present ; for in view of the small effect which 

 magnetic forces have upon the fringe emission when only 

 small amounts of ionized vapour are present, such as in the 

 case with the unprotected plate, it seems hardly possible 

 that the relatively feeble magnetic field due to the heating 



X2 



