1238 THE BELL SYSTEM TECHNICAL JOURNAL, NOVEMBER 1954 



was not small enough to produce non-overlapping effects. If all of the 

 conversions in the 500-foot length were between TE„i and a single addi- 

 tional mode, then it may be shown that the piston position which would 

 cause in-phase addition of components from one conversion point on 

 successive traversal s of the 500-foot line would also cause in-phase 

 addition for the components from all of the other conversion points. 

 However, in the physical line significant conversion takes place between 

 TEoi and three or more other modes, each of which has a different phase 

 constant and each of which requires a different piston setting for in-phase 

 (or out-of-phase) addition of components generated on successive trav- 

 ersals. Whereas it is not possible to relate the shuttle-pulse observations 

 of Fig. 15 to a simple quantitative theory, the general behavior of the 

 experimental line is in qualitative agreement with the conversion- 

 reconversion explanation. 



The favorable piston setting in Fig. 15 represents an infinite hne in 

 which approximate amplitude cancellation of mode conversion effects is 

 achieved, whereas the unfavorable piston setting corresponds to an in- 

 finite line in which approximate amplitude addition of mode conversion 

 effects is experienced. Since the former is optimistic and the latter is 

 pessimistic compared to what might be expected in a physical long line, 

 it has been our practice to average the loss data obtained at the fa\-orable 

 and unfavorable piston positions. 



The expression (5) for the ratio of the signal pulse to the reconversion 

 pulse shows that appreciable loss between the conversion points, repre- 

 sented by the factor ^~'-"^~"i^^^ ^g an effective way of reducing the recon- 

 verted wave effects. During the early experiments it was found that 

 mode filters did reduce the influence of the far end piston. In Fig. 15, 

 rows 3 and 4, show records of the 5" line pulse transmission with a single 

 mode filter at the sending end. The mode filter did not completely 

 eliminate the phasing effects of the piston, and this may be expected for 

 at least two reasons: (1) the mode filter itself may cause some mode 

 conversion, and this mode conversion component will be influenced by 

 the piston setting, and (2) the attenuation introduced by the mode filter 

 is not sufficient to completely eliminate the wave components in the 

 unused mode. 



Even in the presence of the best mode filter now available, row 4 of 

 Fig. 15 shows that the conversion-reconversion phenomenon does take 

 place. The reconverted pulse amplitudes show up at 7,000 yards in row 

 4 as small distortions on the right hand side of the signal pulse, and these 

 distortions grow in magnitude as we proceed to the right in the series of 

 pictures representing more trips past the conversion points. 



