152 BELL SYSTEM TECHNICAL JOURNAL 



more than 15 minutes, this gives an accuracy as high as one part 

 in 50,000. 



For small time intervals, an entirely different method for measur- 

 ing the constancy of the fork must be used. Two methods are avail- 

 able. We may either compare the high harmonics of the fork directly 

 with some high frequency which can be held extremely constant over 

 short periods of time, and observe fluctuations in the relati\"e values 

 of these frequencies, or we may compare the fundamental frequenc\- 

 of the fork with a high frequency by some means which will enable 

 us to measure the divergence from an exact integral multiple rela- 

 tion in terms of the higher frequency. 



To explain in more detail, we may pick out the one hundredth 

 harmonic of the fork by means of a filter and amplifier and compare 

 it with a 5000-cycle frequency obtained from a constant frequency 

 oscillator by some method of detection which will allow us to count 

 the difTerence in cycles. By this means we may observe variations 

 in the relative rate of the fork and the oscillator to an accuracy of 

 about one-tenth of a cycle over a period of a few seconds, and this 

 gives us a comparison to an accuracy of 1 part in 50,000. The prin- 

 cipal objection to this method is the difficulty involved in separating 

 the higher harmonics of any alternating current wave obtained from 

 the fork. For instance, the separation of the hundredth harmonic 

 from those immediately above and below it would require a circuit 

 so selective that it would probably be very difficult to construct and 

 cumbersome to operate. 



If we had means to determine when some high fr^iuency such as 

 5000 cycles was an exact multiple of the 50 cycles and to measure the 

 difTerence in terms of the 5000-cycle wave, we would be able to obtain 

 the same results, and avoid the above difficulty. 



A device which will allow us to do this is the low voltage cathode 

 ray tube developed by Johnson'. The two frequencies to be compared 

 are connected to the two pairs of plates of the tube and the combina- 

 tion of the two deflections causes the luminous spot to trace out a 

 path which repeats itself indefinitely if one frequency is an exact 

 integral multiple of the other, and a stationary figure is produced. 

 In this way any frequency which is a multiple of the fundamental 

 50 cycles may be accurately determined. As the method of com- 

 parison is an electrostatic one practically no power is used. 



I'or the t>pi- of lube used, a deflection of about 1 centimeter is 

 obtained for a potential difTerence of 10 volts between plates, and 



' J. B. Johnson, Bell System Technical Journal Nov. l')22, "\ Low Voltage Cathode 

 Ray Oscillograph." 



