148 BF.IJ. SVSTF.if TF.CIIXICAL JOURNAL 



dark space is positive from the edge of the negative glow ahiiost but 

 not quite to the cathode; there is a thin region just above the cathode 

 where there is more negative charge than positive. This is splendid 

 material for the theorist, and it is deplorable that the method cannot 

 be applied excciit when the rathode-fall is anomalous and exceedingly 

 large. 



When a narrow straight iiole is pierced in the catliode, the positive 

 ions making for it shoot clear through, and can be manipulated in 

 a chamber provided behind the cathode. In particular the ratios of 

 their charges to their masses can be measured, and thence their masses 

 can be inferred. This is Thomson's "positive-ray analysis," which 

 .Aston developed into the most generally available of all methods for 

 analyzing elements into their isotopes. If the density of the gas is so 

 far reduced that the Oookes dark space extends to the anode, the 

 electrons can be stutlied in the same way and their charge-mass ratio 

 determined. Hence the mass of the electron can be deduced, and 

 its dependence upon the speed of the electron ascertained, yielding 

 precious evidence in su.pporl of the special or restricted theory of rela- 

 tivity. These are among the simple phenomena which I mentioned 

 at the beginning of this article, in which the properties of the ultimate 

 atoms of electricity and matter are revealed. 



The positive column, which is the lirilliant, colorful and CDiispicuous 

 part of the glow, resembles in some ways the positive ccjlumn of the 

 mercury arc. In it the ]X)tential-gradient decreases with increasing 

 current, and the characteristic of the glow is negative (Fig. 6). 

 Often the jjositive column subdivides itself into a regular procession of 

 cloudlets or slriations, all just alike and equally spaced (Fig. 7). The 

 potenlial-flitTerence f)etween two consecutive striations has the same 

 value all along the procession, and everyone feels instinctively that it 

 ought to be the i(»nizing-p(ttential or the resonance-potential of the 

 gas; but this is evidently too simi)le an interpretation for the general 

 case, although striations at potential-intervals of 4.9 volts have been 

 realized in mercurv' va|)or. deneralK', if not always, the striations 

 appear when the gas is contaminatetl with a small admixture of 

 another. In this fact the key to the jiroblem of their origin probably 

 lies. 



The (ilo'iC in a dense jias (as dense as the atmosphere, or more so) 

 is visible only when the surface of either or both electrodes is curved, 

 with a radius of curvature smaller than the minimum distance between 

 the two. In these circumstance.s the field strength varies very greatly 

 from one point to another of the inters|)ace, at least before the space- 

 charges Ix-gin to distort the field, and presumably afterwards as well; 



