596 



?', = , ± , ±2— ±/.— (lOV) 



4.T 4jr 4.-T 



//' <^/At'>' values of p ivere admitted for a uniformly rotatini/ dipole, 

 It löoidd he possible that by reversal of the described adiahatic process 

 sinusoidal vibrations were obtained, ivith an amount of energy ivhich 

 ivoidd come in collision with Planck's assumptions (3) and (2). 



If we have K dipoles, and if with given total energy, we wish 

 to calculate the "most probable" distribution of tlie dipoles o^er the 

 possible motions (10), it is still to be fixed bv deflnition to what 

 regions in the (r/,p)-plane the same probability must be assigned. 

 By the ."adiabatic influencing" every separate ellipse of Planck's in 

 the {(j, /^)-plane passes linally into a definite pair of straight lines of 

 the length of 2.t, wdiich lie s§'ni metrically on either side of the 

 (/-axis. If in the statistic treatment of dipoles vibrating sinusoidally 

 Avith Planck we consider all the separate ellipses as regions of 

 equal probability, we are naturally led to treat the just-mentioned 

 })airs of lines for the uniformly rotating dipoles as regions of equal 

 probability ^) (Hypothesis A). However natural this may be, yet it 

 is a new hypothesis. Is this hypothesis inevitable? 



Seemingly the following course is open. Let us start from iV 

 dipoles vibrating sinusoidally (frequency rj, which are distributed 

 over Planck's ellipses in the most probable manner. Apply the 

 above-described "adiabatic intkiencing" to all the poles at the same 

 time. Then an entirely definite distribution of the xV-dipoles over 

 the different modes of motion is obtained finally (10). This distri- 

 bution (distribution B) is, however, another than follows as the most 

 "probable" from the hypothesis A (distribution A). Is distribution B 

 to be taken as the distribution wiiich corresponds with the state of 

 equilibruim, and is therefore the distribution A and the hypothesis 

 A. to be rejected? The remarks nuide in the following § try to 

 demonstrate that the distribution B cannot be considered as a distri- 

 bution of equilibrium. 



§ 4. In case of adiabatic compression black radiation is trans- 



1) In my monograph: "Bemerk, betreffs der specif. Warme zweiatomiger Gase", 

 Verh. d. deutsch. phys. Ges. 15 (1913) p. 453, 1 have erroneously put: 



'h ^* 



'^^i = :^ SO p^ — ....±n~- 



This, however, has no further influence on the derivations given tliere than that 

 the numerical value of the moment of inertia ]j of the hydrogen molecule calculated 

 finally must be divided by four. 



~) P. Ehrenfest, Bemerk, betreffs der specif. Warme zweiatomiger Gase. Verh. 

 d. deutschen phys. Ges. 15 (1913) p. 458. 



