LOAD RATING THEORY 627 



handled separately, at least for negative feedback amplifiers, it being 

 understood that the results are combined in the final design. As 

 previously stated, only the load capacity problem will be considered 

 in detail here but many of the methods used have been applied suc- 

 cessfully to the interchannel modulation problem. 



The Load on a Single Channel 



The total load applied to a multi-channel amplifier varies rapidly 

 between widely separated limits. A complete knowledge of the 

 variations in the load applied to a single channel is necessary first; 

 these variations arise from several recognizable causes which may be 

 discussed separately. 



Number of Active Channels 



First of all, a single channel at a given instant may or may not be 

 carrying speech; if not, it contributes nothing to the multi-channel 

 load. A channel will be called "active" whenever continuous speech 

 is being introduced into it; i.e., a channel is active during the time it is 

 actually carrying speech power, and also during the short pauses that 

 occur between words and syllables of ordinary connected speech. A 

 channel is said to be "busy" when it is not available to the operator 

 for completing a new call. Busy time is by no means all active time, 

 for a busy channel is inactive during much of the time the connection 

 is being completed, during pauses in the conversation, and finally 

 during the time the other party is talking. The fraction of time 

 during the busiest hour that a channel may be busy depends on the 

 size of the group of circuits of which it is a member and on the methods 

 of traffic operation. Measurements on circuits in large groups, made 

 by Mr. M. S. Burgess, indicate that the largest fraction of the busiest 

 hour that a channel may be active is about \. For channels in small 

 circuit groups, this figure may become considerably smaller but it is 

 unlikely that any probable increase in group size or improvement in 

 operating practices will increase it appreciably. This figure, which 

 will be represented by r, may accordingly be taken as a conservative 

 estimate of the limiting probability that a channel will be active in 

 the busiest hour. 



The number of channels that are active at a given instant in an 

 iV-channel system may be anything from zero to N. Inasmuch as the 

 channels are independent, it is possible to write down at once the 

 probability that exactly n of them are simultaneously active. This 

 probability is 



