af 
may Af (lg) des 
us Os TINGE ht = apt io 
7: 
When we reverse the relation found for RT, omitting log 2 by 
(I—6) 
because p? = 
0 
A 
the side of the so much larger term /og Gan 1), where B—A = 
= L, is small compared with A, and putting also 6 =1 (which is 
fulfilled for large values of ¢: f), we get: 
2A 
A 
eRT sl 
E=A+ (9) 
As we already remarked in our first paper, it is indeed exceed- 
ingly remarkable that (with the exception of a few numerical factors) 
exactly the same relation between F and 7’ appears here as was 
derived by Pranck on the ground of the hypothesis of “quanta” 
drawn up by him. For this it was only required to take into account 
the time averages in the ordinary dynamic ,relations, which gives 
rise especially at low temperatures to a considerable difference between 
ui” (the time average of the value of w,’, which has greatly increased 
under the influence of the attractive forces) and w,’, both being very 
slight and approaching to 0. 
From (9) follows with 7/,A: R=a: 
ep 
(ele + DE ae (Ei) oss gw (9) 
; e Een 1 ge (e it —1)? 
which exponentially approaches to 0 (via to aR a 5) when 7’ 
approaches to 0. 
There is, however, one great difference with Puanck’s formula. 
Apart from this that in Pranck’s work the well-known quantity 
Nhr appears instead of ?/,A4 = Nf (/—o)*, so that hy would have 
to be hy=f(/—«c)?), our formula (g) is only valid for very low 
_ temperatures, and (/) only for very high temperatures. This is of 
course only owing to this that (c,) only ensues from the general 
formula (c), when y is supposed to be small, whereas with large 
values of ¢ the relation (c,) results from it. Accordingly our (g) may, 
1) Cf. what has been said concerning /—c under b) of Note 2. 
