120 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1911. 
An important innovation from a practical point of view was the 
adoption at Clifden and Glace Bay of air condensers, composed of 
insulated metallic plates suspended in air at ordinary pressure. In 
this manner we greatly reduce the loss of energy which would take 
place in consequence of dielectric hysteresis were a glass or solid 
dielectric employed. A very considerable economy in working also 
results from the absence of dielectric breakages, for, should the 
potential be so raised as to even produce a discharge from plate to 
plate across the condenser, this does not permanently affect the value 
of the dielectric, as air is self-healing and one of the few commodities 
which can be replaced at a minimum of cost. 
Various arrangements have been tried and tested for obtaining 
continuous or very prolonged trains 
of waves, but it has been my expe- 
rience that, when utilizing the best re- 
ceivers at present available, itis neither 
economical nor efficient to attempt to 
make the waves toocontinuous. Much 
better results are obtained when 
groups of waves (fig. 5) are emitted et 
regular intervals in such manner that 
their cumulative effect produces a 
clear musical note in the receiver, 
which is tuned not only to the period- 
icity of the electric waves transmitted . 
but also to their group frequency. 
In this manner the receiver may be 
doubly tuned, with the result that a 
far greater selectivity can be obtained 
than by the employment of wave tun- 
ing alone. 
In fact, it is quite easy to pick up simultaneously different messages 
transmitted om the same wave length, but syntonized to different 
group frequencies. 
As far as wave tuning goes, very good results—almost as good es 
are obtainable by means of continuous oscillations—can be achieved 
with groups of waves, the decrement of which is in each group 0.03 
or 0.04, which means that about 30 or 40 useful oscillations are 
rudiated before their amplitude has become too small to perceptibly 
affect the receiver. 
The condenser circuit at Clifden has a decrement of from 0.015 
%o 0.03 for fairly long waves. 
% This persistency of the oscillations has been obtained by the 
@mployment of the system shown in figure 6, which I first described 
in a patent taken out in September, 1907. This method eliminates 
FIG. 4. 
