198 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1909. 
is one that we should expect to hold for a collection of a large number 
of machines of any type, provided that we could not directly affect 
the individual machines, but could only observe the average effects 
produced by an enormous number of them. On this view, the second 
law, as well as the first, should be incapable of saying that the ma- 
chines were of any particular type; so that investigations founded 
on thermodynamics, though the expressions they lead to may sug- 
gest—can not, I think, be regarded as proving—the unit structure 
of light energy. 
Tt would seem as if in the application of thermodynamics to radia- 
tion some additional assumption has been implicitly introduced, for . 
these applications lead to definite relations between the energy of the 
hight of any particular wave length and the temperature of the 
luminous body. 
Now, a possible way of accounting for the hght emitted by hot 
- bodies is to suppose that it arises from the collisions of corpuscles 
with the molecules of the hot body, but it is only for one particular 
law of force between the corpuscles and the molecules that the distri- 
bution of energy would be the same as that deduced by the second 
law of thermodynamics, so that in this case, as in the other, the results 
obtained by the application of thermodynamics to radiation would 
require us to suppose that the second law of thermodynamics is only 
true for radiation when the radiation is produced by mechanism of a 
special type. 
Quite apart, however, from considerations of thermodynamics, we 
should expect that the ight from a luminous source should in many 
cases consist of parcels, possessing, at any rate to begin with, a definite 
amount of energy. Consider, for example, the case of a gas like 
sodium vapor, emitting hght of a definite wave length; we may 
imagine that this light, consisting of electrical waves, is emitted by 
systems resembling Leyden jars. The energy originally possessed by 
such a system will be the electrostatic energy of the charged jar. 
When the vibrations are started this energy will be radiated away 
into space, the radiation forming a complex system, containing, if the 
jar has no electrical resistance, the energy stored up in the jar. 
The amount of this energy will depend on the size of the jar and 
the quantity of electricity with which it is charged. With regard to 
the charge, we must remember that we are dealing with systems 
formed out of single molecules, so that the charge will only consist 
of one or two natural units of electricity, or, at all events, some small 
multiple of that unit, while for geometrically similar Leyden jars the 
energy for a given charge will be proportional to the frequency of 
the vibration; thus the energy in the bundle of radiation will be pro- 
portional to the frequency of the vibration, 
