194 BELL SYSTEM TECHNICAL JOURNAL 



affecting the record at these peak points. At the same time these 

 observations were made, the wave shape of the signal rectified from 

 the antenna current at the transmitter was recorded by an oscillo- 

 graph. These oscillograms showed the signal to be free from dis- 

 tortion at the transmitter. 



The weaving of these shadowy traces together with their width 

 gives a record of the change in phase and amplitude of the irregu- 

 larities in the wave shape of the signal. Although the wave shape 

 of the signal is continually changing, it persists in substantially the 

 same form for a great many cycles. Thus the record shows that, 

 in the transmission of this simple tone modulated signal from the 

 transmitting to the receiving antenna, it has been so modified that 

 entirely new frequencies appear at the receiver. This receiver was 

 shown by local tests to be free of any appreciable distortion within 

 itself. While these new frequencies look like harmonics of the mod- 

 ulating tone in the snap-shot record it is obvious from the slow record 

 that they are not true harmonics but that they differ from the har- 

 monics by a very small amount and are incommensurable with the 

 modulating tone since they undergo progressive but irregular phase 

 changes with reference to it. 



These records represent in a nutshell the signal distortion problem 

 as it first presented itself to us. Our work then consisted in raveling 

 out the complicated relations so that their nature could be ascer- 

 tained and a theory of the causes established. In this paper, in the 

 interest of clarity of presentation we have departed considerably 

 from the actual order of the experimental work but at this point 

 perhaps the actual order is best to follow for a moment. 



With such a weird-looking distortion to analyze, and if possible 

 eliminate, our first thought was as to whether the terminal apparatus 

 might not involve unrecognized peculiarities which would be a con- 

 tributing cause. Local tests and daytime tests of the receiving 

 system absolved it from doubt and attention was focussed on the 

 transmitting apparatus. 



It was suspected that present day radio telephone transmitters lea\e 

 something to be desired in regard to what we may call, for lack of a 

 better term, their dynamic frequency stability. A very large per- 

 centage of the transmitters in use throughout the world today pro- 

 duce amplitude modulation of the carrier by the action of modu- 

 lating tulies directly upon an oscillating tube circuit. It is to be ex- 

 pected that the c>rlic changes in circuit conditions occurring at the 

 modulating frequency will have some cyclic effect on the absolute 

 frequency of the carrier and that this effect will be in the nature of a 



