ON A VTi:W OP THE CONSTITUTION OP A LUMINOUS, &C. 64<0 



account oue liiie of the spectrum, tlie estimated value of K — 1 ought 

 to be too siuall, iiah^ed very mucli too small if the gas is oiie having 

 a gfeat uuinber of liues iu its spectrum. Let us suppose tliat the gas 

 is at atmospheric pressure and that ail the molécules of Hélium aregiviug 

 out the yellow Hue A = 5,9 X 10 ""^ approximately. Then N will be 

 the number of molécules iu a cubic ceutimetre of auy gas at atmosphe- 

 ric pressure aud température, aud takiug the smallest possible value of 

 e : viz the charge ou the hydrogeu iou in the electrolysis of solutions, 

 if (' is measured in Electromagnetic units, since one unit libérâtes 1/23 

 cubic centimètres of hydrogen, N'e = 0,4. Substituting thèse values we 

 find that if we only take into account the yellow liue then A — 1 from 

 équation (2) would be 1,1 X 10~'^; the actual value is 8,5 X 10~' 

 (this is deduced from Lord Kayleigh's détermination of the refractive 

 index of Hélium). Thus the calculated value is about 13 tiines the real 

 value and as in the calculatiou we only took account of one out of 

 several of the Hues given by Hélium, the calculated value ought to hâve 

 been much too small. I find that the same thiug is true for ail the gases 

 I bave tried, the value of K — 1 given by équation (2) beiug enorm- 

 ously too large if we, as in the preceding calculatiou, assume that every 

 molécule of the gas eau give out every liue in the spectrum. This 

 suggests that in a luminous gas the spectral lines are not given out by 

 every molécule of the gas, but by only a comparatively small number 

 of Systems formed im some way or other out of the molécules, perhaps 

 aggregations of greater complexity than the molécules. It may be that 

 the Systems which give out oue liue are not those which give out the 

 otliers. ïhis would atford au explanation of the extraordinary variations 

 which sometimes take place in the relative brightuess of the liues given 

 out by a gas wheu the circumstances are clianged. If this view is correct 

 then the number of lines in the sjjectrum of the gas is not necessarily 

 connected with the number of degrees of freedom of the average molé- 

 cule, a resuit which is of some importance in connection with the 

 theory of the ratio of the spécifie beats. 



Cambridge, Nov. 1900, 



41* 



