APPLICATION OF PRINTING TELEGRAPH 609 



ception that automatic transmission from a perforated tape for operat- 

 ing page teletypewriters was substituted for the manual keyboard 

 method for operating tape printers ordinarily used. 



At Rocky Point the power of the radio transmitter was greatly 

 reduced for these tests. The average power radiated in the direction of 

 Rochester was equivalent to 0.7 kilowatt radiated from a nondirec- 

 tional antenna. The average deviation from this value was less than 1 

 db. Under these conditions the average field received in Rochester was 

 42.5 db above one microvolt per meter. The average deviation from 

 this mean value was less than 2 db. A daily half-hour test was made 

 in the afternoon or evening at a time so chosen as to avoid the sunset 

 period of disturbed radio transmission. 



At Rochester, laboratory type receiving equipment was employed 

 for picking up the radio signals and demodulating them to voice fre- 

 quencies. The voice-frequency signals were then used to operate 

 standard voice-frequency carrier terminal equipment at Rochester. 

 This was modified for two-tone operation in a manner similar to that 

 shown for the transatlantic receiving terminal in Fig. 4. 



The teletypewriter signals were sent out from New York at 60 

 words per minute from an automatic tape transmitter. The copy re- 

 ceived over the radio circuit was subsequently compared with simul- 

 taneously recorded copy which was not sent over the radio circuit. 

 Keyboard errors which occur occasionally in perforating the tape for 

 automatic transmission appeared on both copies. Disregarding these 

 errors and counting all which did not appear on both copies, it was 

 possible to obtain the per cent of errors caused by radio transmission. 

 During the half-hour daily test period, about 10,000 characters were 

 sent. It is apparent that rates of error which were less than about 0.1 

 per cent could not be determined accurately. 



Before making the half-hour test each day to determine the per cent 

 of errors received at Rochester, measurements were made of the amount 

 of signal and of noise in the output of each voice-frequency filter. The 

 signal-to-noise ratio thus measured was assumed to be the value ob- 

 taining over the succeeding half hour of test. The nature of these 

 measurements was such that the data were somewhat scattered. How- 

 ever, by suitable smoothing procedures the approximate curve shown 

 in Fig. 5 was plotted. 



At Houlton, Maine, routine radio noise ^ observations are made four 

 times each day on a loop antenna and hourly on the wave-antenna 

 system, as a part of the operating procedure in maintaining transatlan- 

 tic telephone service.'^ It seemed desirable to find out whether these 

 data which extend over several years could be utilized to extrapolate 



