LOADED TELEGRAPH CABLES 397 



recorded on the strip containing the record of their transmitted train. 

 Station 1 measures on its oscillogram the time elapsing between its 

 transmitted train and its received train, and at Station 2 the time 

 elapsing between the received train and the transmitted train is 

 measured. After making suitable corrections, which will be described 

 later, the difference between the interval measured at Station 1 and 

 that measured at Station 2 will be equal to twice the time of trans- 

 mission of the train of reversals over the cable. 



A typical record such as would be obtained at Station 1 is shown 

 in Fig. 4. It will be observed that in the record of received voltage 

 the first few cycles are somewhat distorted because of the fact that 

 the steady state has not yet been reached. Because of this fact the 

 time of arrival or departure of a train is referred to a later cycle in 

 the series, say the fifth. The times of departure and arrival of the 

 various zeros following this cycle are measured, and the average of 

 the values so obtained is defined as the time of arrival or departure 

 of the train. In this way the possible errors due to interference or to 

 distortion in the sent record due to improper functioning of the 

 transmitter are eliminated. 



It will be observed that, mainly on account of the presence of the 

 condenser C, the voltage reversals impressed on the cable are not 

 flat-topped and the zero phase of the fundamental component which 

 we are measuring occurs somewhat ahead of the point in the trans- 

 mitted voltage which we have used as the zero of reference in measuring 

 the oscillograms. Since we are interested in the time elapsing between 

 zero phase of the fundamental frequency in the transmitted voltage 

 and the zero phase of the corresponding cycle in the received train, 

 it is necessary to compute this interval, either by graphical analysis 

 of the oscillogram or by computation from the constants of the circuit, 

 and add the corresponding time to the time which has been measured. 



Although the mechanical arrangement by which the timing lines are 

 obtained on the oscillogram is adjusted as accurately as possible so 

 that the interval between lines is very nearly one hundredth of a 

 second, the very slight variations which occur in such a system are 

 apt to introduce considerable error into the measurement of time of 

 propagation. This is due to the fact that the time of propagation is 

 obtained from the difference of two intervals each of which may be 

 as much as ten times the time of propagation. An error in either 

 interval will therefore result in a tenfold error in the final result. 

 To guard against this condition a record is taken during the experiment 

 of a periodic voltage obtained from a standard oscillator or fork, and 

 the peaks of this oscillation serve as a check on the timing lines. As 



