1178 
sion and absorption, so that a certain wavelength may be represented 
in the spectrum to its normal amount, even if the emissive power 
of the walls be imperceptibly small for that wavelength, provided 
the absorption have a corresponding, small value. The small value of 
the emissive power has no influence on the final partition. It only 
occasions that radiation of other energy-partition will only very 
slowly be transformed into the normal partition. 
So we shall assume that the centra of radiation are vibrators 
whose equations of motion are for the present unknown. These 
equations cannot have rigorously the form (1), but they need differ 
only very little from it. 
§ 3. The independent variables. The ensemble. 
We will imagine an ensemble each system of which consists of a 
parallelopipedic space inclosed in perfectly reflecting walls and 
containing one vibrator, whose centre has a fixed position in that 
space. We will assume that the motion of that vibrator is determined 
by one coordinate. 
The choice of the independent variables requires a certain circum- 
spection. The aether namely represents an infinite number of degrees 
of freedom, each of which can therefore possess an infinitely small 
amount of energy. The vibrator on the other hand possesses a finite 
amount of energy. It seems, however, difficult to deal with an ensemble 
in which one variable possesses on an average infinite times as 
much energy as the other variables. Therefore I will choose the 
variables as follows: If a monochromatic ray of light passes a vibrator 
the latter will be set into vibration. After a certain time this vibration 
will have become stationary. Now [ will determine by one coordinate 
the amplitude of the ray and the stationary vibration of the vibrator 
caused by it. 
Besides this I will assume that the vibrator has a “proper” coor- 
dinate. Now if this proper coordinate, and also its time derivative 
are zero, this does not mean that the vibrator stands still in its 
position of equilibrium. It does mean that the motion of the vibrator 
consists exclusively of the stationary vibration, which it assumes 
through the influence of the radiation to which it is subjected. If 
the proper coordinate is not zero, then the vibrator has a motion 
which does not agree with the absorbed vibration. So it is possible 
to assume, that in a radiation field which is in equilibrium (i.e. in 
which the energy partition is that of the normal spectrum) the proper 
coordinate of the vibrator has always an infinitely small amount of 
energy (in the same way as the separate coordinates which determine 
