180 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1909. 
phone, but when rectified by the valve they become equivalent to an 
intermittent unidirectional current, and can then affect the telephone 
or a galvanometer, or any instrument for detecting a direct current. 
On the other hand, we may take advantage, as I have more recently 
shown, of the nonlinear form of the characteristic curve. In other 
words, of the fact that the conductivity of the ionized gas is a func- 
it 
Fic. 18. Fie. 19. 
Connections for oscillation valve used as radiotelegraphic detector. 
tion of the voltage applied to it, and in this second method the valve 
and receiving circuits are arranged as shown in figure 20. In this 
case, we have to apply to the ionized gas a unidirectional electromotive 
force which corresponds to a point of inflexion on the character- 
istic curve, and then to add to this 
voltage the alternating voltage of the 
oscillations set up by the incident elec- 
tric waves in the receiving circuit. The 
result is to cause a change in the aver- 
age value of the current through the 
telephone, and therefore to produce a 
sound in it, long or short, according to 
the number of trains of waves falling 
on the antenna. This last method, 
then, requires the application in the 
Fic. 20.—Connections for oscilla- telephone circuit of an accurately ad- 
Sea eapern as a radiotele’ Justed steady electromotive force, not 
any electromotive force, but just that 
value which corresponds to a point on the characteristic curve at 
which there is a sudden change of curvature. 
At this point we may notice a broad generalization which has 
already been made by H. Brandes, viz, that any materials such as the 
erystals mentioned, or ionized gases, which do not obey Ohm’s law as 
regards the independence of conductivity on impressed voltage, can 
be used as radiotelegraphic receivers. It is necessary to be able to 
test the relative sensibility of detectors to know whether any new 
