DR. CAMPBELL'S MEMORANDA OF 1907 AND 1912 571 



sending end impedance) be added, it still seems to me probable that 

 the improved performance of the circuit will prove in this extra 

 equipment. 



2. The Use of Repeaters of Small Amplification at Periodic Intervals 



along the Line I 



Theoretically, a given total amplification can be secured with a 

 larger singing margin if it is distributed among a number of properly 

 spaced points along the line rather than concentrated at a single point. 

 For example, four equally spaced repeaters each giving an amplification 

 of five miles might be substituted for a single repeater giving twenty 

 miles. If the circuit suggested by the above sketch were employed 

 this would mean a total of eight repeater elements of which four would 

 be used, one after another, as one-way repeaters in each direction. 

 This raises the old question as to whether equally good quality can be 

 obtained when several repeaters are used in securing a given amplifica- 

 tion. This point seems worth further direct experimental investi- 

 gation; one step in the right direction has probably been made by 

 raising the natural period of the diaphragm. 



3. The Use of a Compensating Device Such as an Artificial Line to 

 Reduce the Amplification at the Resonant Frequencies to the Level 



of the A mplification at Other Telephonic Frequencies 



In the sketch, equalizing artificial lines are shown at AA ; obviously 

 the same result may be secured by introducing them at any of a number 

 of other points in the circuit. In this way the singing margin can be 

 increased and the quality be somewhat improved, without materially 

 reducing the telephonic amplification. But on general principles it 

 would seem desirable to carry the equalization as far as possible in the 

 repeater itself. The variability of the repeater sets a limit to what 

 may be accomplished by any compensating device which reduces the 

 total amplification by an invariable amount at each frequency and 

 thereby increases the percentage variation. If it became necessary 

 merely to eliminate certain frequencies lying outside of the range 

 required for telephony, the use of an artificial selecting circuit would 

 seem to present no difficulty. 



The variation as well as the average amplification obtained from 

 repeaters should be investigated. When these data have been ob- 

 tained for the best type of repeater it will be possible to determine 

 whether any material benefits can be derived by the introduction of 

 compensating circuits. 



