CARRIER TELEGRAPH METHODS 525 



per second and words per minute respectively at which these arrangements 

 operated with 10 per cent distortion. In taking these data all other vari- 

 ables such as noise and level or frequency changes were absent. (Similarly 

 in the other columns only the variable mentioned was allowed to change.) 

 Column D gives the order of preference for arrangements having the same 

 channel filter loss characteristic (but not necessarily the same total frequency 

 band) on the basis that the highest speed is the most desirable. These 

 data apply when no flanking channels were present. The one-source 

 two-band arrangement and the single-sideband arrangement with —6 db 

 spacing carrier were the fastest of those having the same channel filter loss 

 characteristic, as might be expected from the widths of the transmitted 

 sidebands. However, for a given band width, the single-sideband arrange- 

 ment with — 6 db spacing carrier was the fastest. If frequency band width 

 is of paramount importance, consideration should be limited to arrangements 

 with the same band width, which would change the order of preference. 



Columns E, F, and G apply, respectively, in place of columns B, C, and 

 D, when flanking channels were in operation. According to column G, the 

 single-sideband arrangement lost its place in preferential rating due to 

 interference from the adjacent channels; and the on-off and two-source 

 arrangements were the best, as might be expected, because their carriers 

 were located at midband. 



In column H are listed the ranges of levels over which the received signal 

 power could vary without causing more than 10 per cent distortion. The 

 order of preference listed in column I indicates that arrangements with 

 limiters were the most stable. 



In column L are listed the ranges of carrier frequency variations which 

 could be tolerated without causing the distortion to exceed 10 per cent. 

 In column M it is seen that the two-source arrangements performed better 

 than the others when the mean frequency changed. 



In column P the arrangements are rated on the basis of the resistance 

 noise which could be tolerated on the line compared to the linear on-ofif 

 arrangement. For example, the frequency-shift arrangement in item 1.11 

 could tolerate 2.7 db more noise than the on-off arrangement for 10 percent 

 distortion at 60 w.p.m. According to column Q the one-source two-band 

 arrangement performed the best in this respect. 



Table VIII is arranged similarly to Table VII except that the maximum 

 total distortion is 20 per cent throughout, and additional data are given in 

 columns J, K, N, O, R, and S to cover speed at 120 w.p.m. 



Discussion 



It is believed that the circuit arrangements tested were reasonably repre- 

 sentative of those commonly used in carrier telegraph practice, so that 



