WAVEGUIDE AS A COMMIXICATIOX MKDIUM 1237 



piilso. Wc may tluMi set the ratio of signal-wave ami^Iitudc lo tlie recoii- 

 verted-wave amplitude, as given by (5), etjual to unity al n = SO. W'v 

 know the theoretical values of the heat loss* ((u-Hirienls a^ and ai for ihe 

 unused and signal modes I'l'spcclix-ely. Allowing modes! iiicicMsc for 

 surface roughness effects, we will assume the heat loss for the signal 

 (TEoi) wave to be 0.2 dh betwcMMi conversion ])oints (al)oul oOO feet 

 apart) and the loss for the unused mode to be about five times this or 

 1 db between conversion points. Substituting these numbers into (5) we 

 can sohe for the conversion coefficient k which is necessary to obtain the 

 observed eciuality between reconversion and signal wave amplitudes at 

 the 71 = 80 point. Such a calculation yields a value of k of 0.117 and this 

 implies a conversion loss which is 32 per cent of the heat loss for the sig- 

 nal wave. Direct observations of mode conversion (to be described) 

 show^ that the conversion losses in the 5" diameter line must be at least 

 as large as this. Therefore, we have confirmed that the basic mechanism 

 which we are discussing can indeed account for a break-up of the signal 

 uulse of the general type actually observed. 



The calculated ratio of the second reconverted pulse compared to the 

 first reconverted pulse from (0) for the n = 80 condition described above 

 is only —0.5 db, which shows that if we really had a single conversion 

 point which could add exactly in-phase in the manner assumed, Ave 

 would have in our photographs an even worse series of reconverted 

 pulses than actually do appear. 



Again for the ?? = 80 condition in the 5" line the calculated am})litude 

 of the unused mode component relatiA'e to the signal (,'omponent at the 

 same point is —18.5 db for the n"' conversion and, with the S(]uare law 

 response of the display system, it is to be expected that such an ampli- 

 tude would not be observed. This fits satisfactorily with the observation 

 that the pulses in the 5" line, after 40 round trips, are all in the circular 

 electric mode even though the conversion-recon\'ersion process is 

 significant. 



As alreatly implied, the mode conversion situation for the actual 

 500-foot line, which is represented l)y the photographs of Fig. 15, is 

 more complicated than the idealized line just analyzed. First, note that 

 in the case of the nonoverlapping pulses in the idealized line the ])osition 

 of the far end piston (i.e., the spacing between the conversion ])(jints) 

 has no critical effect on the conversion and reconversion amplitudes. In 

 the experiment (Fig. 15), far-end piston movements of a foot or so 

 caused very significant changes in flu; observed conxcrsion and recon- 

 version. This is a consequence of the fact that in the physical 500-foot 

 line, there were a large number of conversion points, and the pulse width 



