688 BELL SYSTEM TECHNICAL JOURNAL 



greater by 2.13 and by 2.95 equivalent volts. The accuracy is such 

 that one may attempt to compare them with the energy-values of the 

 excited states of the helium atom, but here there is a disappointment — 

 the agreement is not so good as one would like. MacMillen is disposed 

 to think that the value 21.50 should be corrected to 21.12 and the two 

 others shifted equally, whereupon the first would agree accurately 

 and the two others passably with the energy-values of known excited 

 states. But these three do not comprise the two lowest of the excited 

 states, the 19.77 and 20.55-volt levels which I mentioned above; it 

 would then be necessary to assume that these are much less likely to be 

 attained than three of those above them, which seems surprising but 

 not impossible. At any rate it is evident that the time has arrived 

 for experimenting in ways which permit of locating the shifted peaks 

 with the highest accuracy possible. 



Continuing with Harnwell's data: neon yielded a set of curves very 

 like those of helium, except that the mean value of the energy-loss, 

 deduced from the separation of the pair of peaks corresponding to the 

 pairs in Fig. 9 amounts to some 18 equivalent volts; this is predictable. 

 Molecular hydrogen displayed an energy-loss of 12.3 equivalent volts; 

 but the most striking feature of the curves is the prominence of the 

 peak formed by electrons which have lost that amount of energy— 

 indeed, with 75-volt electrons, even those which go through nearly 

 undeflected include a larger proportion of such, than of corpuscles 

 which have retained their capital intact. Harnwell made measure- 

 ments on nitrogen also, and on a mixture of molecular with atomic 

 hydrogen, this being supplied from a discharge-tube in operation; and 

 I reproduce as Fig. 11 his graph showing the distribution-in-angle 

 of electrons scattered without loss of energy, the hollow circles relating 

 to molecular hydrogen, the dots to the mixture. ^^ The corresponding 

 curves for electrons scattered ivith loss of energy lie very close together, 

 and the one marked C in Fig. 11 represents them both. MacMillen 

 traced similar but not exactly concordant curves for hydrogen, and 

 others for helium and for argon. In his data also, electrons scattered 

 through small angles predominate more and more, the greater their 

 initial speed ; and it is rather surprising to find how large a proportion of 

 the corpuscles which have lost energy to helium atoms continue never- 

 theless with little or no deflection. 



Somewhat earlier, Jones and Whiddington had studied the distribu- 

 tion-of-energy of electrons after passage through hydrogen, confining 



18 The continuous curves marked A and B are graphs of predicted distribution- 

 curves deduced (for atomic hydrogen) from the assumptions of wave-mechanics by 

 Born. The ordinates of all the curves, experimental and theoretical alike, have 

 been adjusted so that all five intersect at 5°. 



