A THEORY OF SCANNING 465 



one-dimensional steady state Fourier components. Such a method of 

 treatment naturally gives no information in regard to the reproduction 

 or distortion of the detail in the original image across the direction of 

 scanning, nor, as will appear below, does it give any detailed informa- 

 tion in regard to the fine-structure distribution of energy over the 

 frequency range occupied by the signal. 



The need of a more detailed theoretical treatment originally arose in 

 connection with studies of the reproduction of detail in telephoto- 

 graphic systems, especially in comparisons of distortion occurring 

 along the direction of scanning with that across this direction. Later, 

 this same need was strikingly shown by the discovery that a television 

 signal leaves certain parts of the frequency range relatively empty of 

 current components. Certain considerations indicated that a large 

 part of the energy of a signal might be located in bands at multiples of 

 the frequency of line scanning. Actual frequency analyses more than 

 confirmed this suspicion. The energy was found to be so closely con- 

 fined to such bands as to leave the regions between relatively empty of 

 signal energy. 



Such bands and intervening empty regions are illustrated by the 

 examples of current-frequency curves in Fig. 1. These curves were 

 taken with the various subjects as indicated, and the television current 

 was generated by an apparatus scanning a field of view in 50 lines at a 

 rate of about 940 lines per second. The energy is grouped in bands at 

 multiples of 940 cycles and the regions between are substantially de- 

 void of current components. In addition to the bands shown by the 

 curves, it is known that similar bands occur up to about 18,000 cycles 

 and that there is also a band of energy extending up from about 20 

 cycles. 



Certain of the relatively empty frequency regions were also investi- 

 gated by including a narrow band elimination filter in a television 

 circuit. The filter eliminated a band about 250 cycles wide and was 

 variable so that the band of elimination could be shifted along the 

 frequency scale at will. By shifting the region of elimination along in 

 this manner it was found that a band about 500 or 600 cycles wide 

 could be removed from a television channel between any two of the 

 current components without producing any detectable effect on the 

 reproduced image. 



At a later date a 1500-cycle current suitable for synchronization was 

 introduced into a relatively empty frequency region, transmitted over 

 the same channel with a television current, and filtered out — all with- 

 out visibly affecting the image. 



These results indicated quite clearly the need of a more complete 



