378 BELL SYSTEM TECHNICAL JOURNAL 



devices as having great promise, and Arnold's initial research efforts were 

 concentrated in this field, using a mercury arc. The suggestion to use the 

 mercury arc as an amplifier was old, having been advanced in this country by 

 Peter Cooper Hewitt, the inventor of the mercury vapor lamp, but it had 

 never proved to be feasible in a practical way, because of its variable ampli- 

 fication, inefficiency, noise, and distortion. Arnold contributed new fea- 

 tures based in part on novel phenomena discovered in his experimental work. 

 These substantially reduced or eliminated the defects above listed. Pat- 

 ents were issued to him in due course, and rights were also obtained under 

 the Hewitt patents. 



The basic element of the Arnold amplifier was a stream of ionized mole- 

 cules of mercury vapor flowing vertically from the positive electrode at the 

 upper end of an evacuated tube to the negative electrode, a pool of mercury 

 at the lower end, the energy for maintaining the arc being furnished by a 

 direct current power source which included in its circuit a stabilizing choke 

 coil and a regulating rheostat. Within the tube there were two auxiliary 

 side electrodes (cathodes) symmetrically disposed with respect to the axis 

 of the tube and closely spaced thereto. There also was a starting electrode 

 located within an associated condensing chamber. The ionized mercury 

 vapor stream was vibrated transversely between the two auxiliary side 

 cathodes, by virtue of the electromagnetic action of the input telephone 

 current flowing through coils which were mounted on the pole pieces of an 

 external electromagnet and so disposed that the axis of the magnetic field 

 was perpendicular to the axis of the ionic stream. The output circuit 

 included a transformer which had a split primary winding, with its two 

 main terminals connected to the two side cathodes respectively, and its 

 mid-point connected to the negative terminal of the d-c power source 

 previously mentioned. When there was no input telephone current in the 

 receiver coils the arc stream flowed steadily, and no current was induced 

 in the secondary windings of the output transformer because the equal 

 currents from the two side cathodes flowed in opposite directions in the two 

 halves of the primary winding and inductively annulled each other's effect 

 on the core. On the other hand, when a telephone current flowed through 

 the magnet coils, the arc stream was magnetically deflected first to one side 

 cathode and then to the other, depending upon the magnetic polarity. 

 This caused changes in the magnitudes of the currents flowing in the indi- 

 vidual halves of the primary winding of the output transformer; as the cur- 

 rent in one half-winding increased that in the other half-winding necessarily 

 decreased, and vice versa. The resultant induced current in the secondary 

 winding of the output coil had similar frequency components to the incoming 

 telephone current and much greater energy, which was supplied by the cur- 

 rent from the external battery, or other power source. 



