PHYSICS: MILLIKAN AND SOUDER 
21 
be found to vanish in the way in which it did in the experiments reported by 
these authors. 
This article formed the subject of some correspondence between Pro- 
fessor Hallwachs and one of us, and quite recently Wiedmann^ working 
in Hallwachs' laboratory has followed out the suggestion above made and 
obtained by cleansing his surfaces by successive distillation instead of 
by cutting, as we had done previously, results altogether similar to ours, 
though of less pronounced character and extending over a much smaller 
range of wave-lengths. Wiedmann's actual observation is that potas- 
sium, whose selective resonance band, according to Pohl and Pringsheim, 
is at 436 ^jl^jl, shows a relatively stronger photo-sensitiveness for line 365 fx^x 
than for either 406 mm or 436 mm, when it has been freed from gas by 
repeated distillation. But when distillations are less numerous and 
freedom from gas less complete the photo-sensitiveness of this line is 
some four times greater than that of the other lines. This means merely 
in terms of our diagrams that lines like 2535 A. and 2804 A. which have 
wave lengths shorter than that corresponding to the selective point are 
relatively strong with respect to those whose wave-lengths are at or 
above that of the selective point when gas has had little chance to act, 
but that they become relatively weak after the gas had had a chance 
to act. 
These results from Professor Hallwachs' laboratory constitute then 
very interesting confirmation of both the results and the conclusions 
which we published one and one-half years ago, namely, that when 
tested with sufficiently short wave-lengths, freeing a surface from gas does 
not at all eliminate its photo-electric sensibility, though, as we pointed 
out then, it does diminish the photo-sensitiveness on the long wave- 
length side, i. e., at or above the region of the wave-lengths at which 
Pohl and Pringsheim's selective effect reaches its maximum. 
Within the past year we have made experiments which show that it is 
very difficult if not impossible to set up any definite criterion for the se- 
lective effect since the so-called normal and selective effects merge by 
insensible gradations into one another. At present we prefer to regard 
the selective effect as nothing but the normal effect taking place in the 
neighborhood of an absorption band, i. e., in the neighborhood at which 
oscillations of one particular frequency predominate. Our reasons for 
this point of view are as follows: 
(1) We have shown that the energy of emission is always governed by 
the normal equation, ^MV^ = PDe = hv — P, whether the impressed fre- 
quency coincides with the maximum of the selective effect or not. 
