618 BELL SYSTEM TECHNICAL JOURNAL 



frequency range to 20,000 cycles involved so little difficulty that it 

 was decided to provide this very liberal frequency range. 



In the particular television system which has been described the 

 very low frequencies (below about 10 cycles) are suppressed. It 

 was, therefore, not necessary that the line transmit these very low 

 frequencies. The frequency range which the line should transmit was 

 accordingly set as 10 cycles to 20,000 cycles. 



Attenuation. Referring to still picture transmission, it has been 

 found that variations of attenuation with frequency of several trans- 

 mission units do not appreciably impair the quality of the picture. 

 Since no great difficulty was anticipated in meeting closer limits, 

 however, it was decided to set the limits for the variation of attenuation 

 with frequency at ± 2 T U within the frequency range of 10 to 20,000 

 cycles. 



Phase Characteristics. A characteristic of wire lines, whose im- 

 portance has been increasingly realized in recent years, is their phase 

 characteristic. In speech transmission, transients due to unequal 

 velocity of the different frequency components have been found to be 

 an important consideration on some types of lines. In picture trans- 

 mission and television, also, it is important that this phase distortion 

 be controlled, as otherwise the image might be blurred due to the 

 arrival of the various frequency components at different times. 

 The type of transient which has been found to impair the quality of 

 pictures is the type which is relatively rapid and the aim has been to 

 make the phase characteristics such that those transients would be 

 small. 



The requirement with respect to phase for distortionless trans- 

 mission is that (S/o) be a constant where jS is the phase change in radians 

 for the entire circuit, and co is equal to 2x times the frequency. |8/co is 

 known as the "phase delay" or the steady-state time of transmission. 

 d^/doj is the time required for the transmission of the envelope of a 

 wave whose components center closely about the frequency cij/27r and 

 it will be referred to as the "envelope delay." Since it is more con- 

 venient to measure the envelope delay, the requirements were set up 

 in terms of this quantity. When /S/co is constant, it is evident that 

 d^/do) is also constant. While the converse of this is not in general 

 true, the conditions as actually encountered were such as to permit 

 its use as a measure of the small variations involved. 



The envelope delay characteristics of a number of circuits, which 

 have been found to give varying degrees of transient on still pictures, 

 have been measured. Also data were available from tests of picture 

 transmission through filters and other networks whose delay charac- 



