150 BELL SYSTEM TECHNICAL JOURNAL 



current that is used to control the fork, it is important that this 

 fluctuation be reduced to a small amount. This may be done by 

 using a large capacity C or a large resistance Ri. However, the 

 effect of increasing the capacity or resistance is to increase the time 

 re(|uired for the control to change, when compensating for changes in 

 the fork frequency. Accordingly the values chosen must be a com- 

 promise. If we assume that the control is capable of giving a max- 

 imum change in frequency of .1%, and we allow a fluctuation in this 

 control of 5% each half second, this will cause a fluctuation in fre- 

 quency of 5%, of .1%, or .005%. However, the inertia of the fork 

 prevents it from following such a rapid fluctuation in damping cur- 

 rent and hence the actual change in fork frequency is very much 

 smaller than just indicated. 



The fact that non-cumulalivc fluctuatitjns in the control as great 

 as 5% have only a negligible effect on the fork frequency is an im- 

 portant point. Such fluctuations are likely to arise through hunt- 

 ing in the synchronous motor, irregularities in the time of operation 

 of the relay, etc., and since their effects average one another out, 

 there is no danger of their being transferred to the fork. 



The ratio of the charging resistance to the discharge or grid leak 

 resistance is not a governing factor, except that the charging resistance 

 must be less than the discharge resistance. The phase position of 

 the fork to the clock under normal conditions is also go\erned by the 

 relative values of these resistances. For the present circuit ri has a 

 resistance one-half that of ro, and these resistances and the condenser 

 are of such values that it takes approximately 15 minutes for the fork 

 to come into the correct phase relation with the clock when started 

 under the most unfavorable conditions. 



While this method of control will hold the fork frequency for an 

 indefinite period in synchronism with the clock, it is possible that the 

 phase relation of the clock to the fork may change. This change 

 may be periodic, that is, it may take the form of an oscillation about 

 the mean phase ])osition, or there may be a gradual change due to 

 changes in the various constants of the control occurring over com- 



r, 

 parati\el\' lung inter\-als. l-"or instance, an\- change m the ratio — . 



such as might occur with temper.it iiro, will change the phase relation 

 between the fork and the clock. 



Chronograph records show that there are no phase changes greater 

 than one cycle f)f the fork over periods as large as 8 hours. To de- 

 termine the possibility of hunting, that is, of oscillation of the fork 

 frequency around its mean value, the phase relation was actually dis- 



