346 



BELL SYSTEM TECHNICAL JOURNAL 



as the Continental and American Morse codes with an effort to make 

 them as brief as possible without making the reading too difhcuh. It 

 is thought that the two ideal codes chosen are comparable in the 

 matter of ease of reading. In constructing the two-element code, two 

 steps are involved. In the first place it is assumed that the markings 

 and spacings of any integral numlx;r of signal elements' duration can be 

 used so that in addition to the values for markings and spacings as- 

 sumed above, there may be dashes of two, four, etc., units duration. 

 With these assumptions the 26 shortest characters that can be con- 

 structed are next made up. It is found that one character is of 1 unit 

 duration, 1 of 2 units, 2 of 3 units, 3 of 4 units, 5 of 5 units and 9 of ti 

 units duration. The remaining 5 characters arc taken of 7 units 

 duration each. The second step is to ascril^e the 26 letters of the 

 alphabet to these characters in such an order that the most frequent 

 letters correspond to the shortest characters. It is most efficient 

 to use the same spacing as was assumed abo\i' for the Continental 

 two-current-value code, with the addition that spaces of longer 

 duration than three units may be employed within a letter. 



The matter of constructing the ideal three-current-value cotle is 

 similar. First, the 26 shortest characters are constructed. Two 

 characters can be constructed having a duration of 1 unit, four char- 

 acters having a duration of 2 units and eight characters having a 

 duration of 3 units. The remaining twelve characters are taken 4 

 units in duration. Next, the most frequent letters arc assigned to 

 these characters in the order of their duration. It is best in this 

 case to use the same assumptions as to spacings between letters and 

 words as was used above in connection with the threc-current-value 

 Continental code. The use of spaces within letters is not economical 

 in this case. 



A frefjuency table given by Hitt ' was used to determine the relative 

 frequency of the various letters. The average duration per letter 

 was computed from this table and corrected for spaces between words 

 and letters. The resultant average duration is as follows: 



Siunal Ele- 

 ments per 

 Code Letter 



American Morse (two current v.ilucs). .. . 8 26 



Continental Morse (two current values) 8.45 



Ideal (two current values) 6. 14 



Continental Morse (three current values) .V 77 



Ideal (three current values) .? 03 



•Parker Hitt. "Manual for the Solution of Military Ciphers." Army Service 

 Schools Press, Port Leavenworth, Kansas. Second edition, p. 7. 





