836 THE BELL SYSTEM TECHNICAL JOURNAL, JULY 1953 



Having established that a minimum of information should be col- 

 lected, we return to the time variation of the system, which in theory 

 can make the information obsolete before it can be used. However, 

 without yet bringing in the practical fact that its most rapid rate is 

 relatively slow, we should introduce the fact that the ways in which the 

 system can vary at its most rapid rate are quite restricted as compared 

 with the manner in which it can vary at slower rates, and we note that 

 even these are relatively limited. (This probably applies to most trans- 

 mission systems, not just to the L3 coaxial.) 



Thus while the information regarding the more rapid, but simple 

 changes must be collected more frequently it consists of a small amount 

 of information per sample; whereas the information regarding the slow 

 (but more complex) changes need not be collected very often, but it 

 represents a relatively large amount of information per sample. Since 

 the total rate of information collection is proportional to the information 

 per sample times the rate of sampling, the minimum information col- 

 lection principle would say that the rate of sampling should also be held 

 to a minimum. 



This demonstrates the value of association of particular types of sys- 

 tem change with the rate and amount of their variation, because one may 

 thus eliminate from the more rapid sampling the collection of informa- 

 tion about changes that occur at slow rates. Furthermore, one may es- 

 tablish the sampling rates for the various system effects at the lowest 

 possible value. (There is also a very practical value in knowing the rates 

 and amounts of the various deviations, because system misalignment 

 requirements force the equalization to be suitably distributed along the 

 line.) 



In the form of a rule, this is: >. 



Rule II 



To the greatest practicable extent the overall system behavior should be 

 separated into individual effects each having its own time rate of occurrence 

 and corrections should be made for each effect at the minimum tolerable 

 rate for each. 



It is of interest to note that, if we couple the logic of Rule II with the 

 fact that the fastest changes (due to changes in the temperature of the 

 repeater huts) take hours to become appreciable, we see that the con- 

 tinuous collection of information from continuous pilots (and the con- 

 tinuous correction by pilot-controlled regulators) is in principle un- 

 necessary and inefficient — except, of course, for its other function of 

 giving alarms under trouble conditions. 



