STATISTICS OF T i;hl-;\ ISK ).\ SIGNALS 703 



eon'olation was measured for two typical motion-picture films, by means 

 of the apparatus descrilxMl in tiie section on autocorrelation.* The i-esults 

 were 0.80 and 0.8(), after coi'i'ectioii for tlu> l-inc t)aiid\vi(lth limitation. 

 This means that "[)re\ious-frame" pixMliction can ivmoxe only slifi;htly 

 mor(> than one bit of redundancy per sample. More complicated schemes 

 would pi'esumably be more successful in taking advantage of the large 

 frame-to-frame redundancy which undoubtedly exists. 



ACKNOWLEDGMENT 



Many of the ideas expressed in this paper are due to B. M. Oliver, 

 whose resourcefulness is hereby gratefully acknowledged. 



* The expression used in evaluating tlie correlation between frame 1 and frame 

 2 (an}^ two frames) is 



rp rp2 



_ -^ 12 ~ -^ 1 (-7) 



where T12 is the optical transmission of frames 1 and 2 in cascade, Ti is the average 

 of the individual transmission of frames 1 and 2, and Tn is the average of the 

 transmissions of two cascaded slides of frame 1 and two cascaded slides of frame 

 2, respectively. In all cascade transmission measurements, the two frames must 

 be in precise register. 



