NON-BINARY ERROR CORRECTION CODES 



1373 



digit will change only one of these two components. Further, an error 

 corresponding to a change in the quinary component can be unic^uely 

 corrected if the error in the decimal digit is assumed to be ±1. For 

 example, if a received 6 is discovered to have an incorrect quinary com- 

 ponent, only a decrease in the quinary component making the decimal 

 digit 5 is a possible correction, since an increase in the c^uinary com- 

 ponent would correspond to the decimal digit 9, a change of more than 

 ±1 from 6. 

 A system is shown in Fig. 2 for taking advantage of these properties. 



BINARY MESSAGE 



INFORMATION 



DIGITS (NOT USED) 



BINARY PARITY 

 CHECK DIGITS 



INFORMATION RECEPTOR 



BINARY 



INFORMATION 



DIGITS 



n-m 



QUINARY 



INFORMATION 



DIGITS 



DECIMAL 

 MESSAGE 



4 



QUINARY 

 DIGITS 



QUINARY 



CORRECTION 



CIRCUIT 



BINARY 

 DIGITS 



20 



ODD OR EVEN 



RECOGNITION 



CIRCUIT 



2n 



BINARY DECODER 

 AND CORRECTOR 



BINARY 

 DIGITS 



^ 



Fig. 2 — Use of binary codes with a decimal channel. 



