176 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1907. 
the are, by varying the energy expended in it, expand and con- 
tract the column of incandescent vapor forming the true arc, and 
therefore the layers of air next to the are, and hence send out air 
waves which are heard as sound. Frequencies up to 10,000 or so are 
possible, although many physicists, such as Bannti, Corbino, and also 
Maisel, contend that much higher frequencies can be obtained. In 
1903 Mr. Poulsen introduced a further improvement. He found that 
by inclosing the arc in a vessel containing hydrogen or coal gas, and 
forming the are between a cold metal terminal, which is the positive, 
and a large carbon terminal, which is the negative, the arc being also 
traversed by a strong magnetic field, much higher oscillation fre- 
quencies could be obtained than with the double carbon arc in air. 
(See fig. 11.) 
He also found it is an advantage to rotate the carbon terminal. 
When this arc is shunted by an appropriate small condenser in series 
Fig. 11.—Poulsen’s apparatus for creating undamped oscillations. 
with an inductance, we can obtain in this last circuit electric oscilla- 
tions having a frequency of a million or more depending on the 
capacity and inductance used. If a suitably tuned antenna is con- 
nected to one terminal of this condenser, and one arc terminal to 
the earth, as shown in the diagram, we are able to radiate from the 
antenna undamped trains of electric waves. 
I have before me an apparatus of this kind with which much work 
has been done in my laboratory during the last few months. It con- 
sists of a water-jacketed brass cylinder with marble ends, through 
which project at one end a thick carbon rod, kept in rotation by a 
motor, and at the other a water-cooled brass tube with copper beak 
at the end. An electric are is formed with 400-500 volts between 
these terminals taking 6-10 amperes. 
