COUPLED WAVE THEORY AND WAVEGUIDE APPLICATIONS G97 



increasing wavelength values, a general result for small holes in the side 

 wall. As the number of wires in the gi\'en aperture was reduced, markedly 

 increased coupling resulted. This was due to the fact that the coupling 

 loss per hole varied approximately as the fouilh power of the hole dimen- 

 sion perpendicular to the electric vector, whereas the overall power loss 

 varied only as the scpare of the luimber of holes in the loose coupling 

 region. (Equations (39) and (40) describe the effects of inimbcn- of cou- 

 pling points more precisely.) At 50 holes, the transfer loss was about 5 db 

 and the wave in the driven line was reduced by about 2 db; the slope of 

 the I El I \'ersus Xo plot was the same as for 102 holes. At 34 holes, the 

 transfer loss was about 2 db and the wa\'e in the driven line was reduced 

 b}^ about 5 db; in this case, however, the undri\'en line wave loss in- 

 creased with increasing Xo . Since coupling increases with increasing Xo 

 we deduced that the total coupHng was greater than required for com- 

 plete power transfer and the bracketed expression of (39) and (40) was 

 greater than 7r/2. On the diagram of Fig. 16, the presumed operating 

 point was near ex = 2.2 radians. At 25 holes, Fig. 37(d), the transfer 

 loss was about 5 db and the wave in the driven line was reduced l)y about 

 2 db; as in the 34 hole case, the undriven line wave amplitude decreased 

 with increasing Xq and hence with increasing couphng. Again the in- 

 tegrated coupling appeared to be in the region between x/2 and tt. The 

 driven line wave loss was headed for a low value at the long-wave end 

 of Fig. 37(d), and it seems clear that periodic energy exchange is realized 

 in practice. 



The losses associated with this energy exchange may be inferred by 

 comparing the total power output of the undriven and driven lines to the 

 input power. Assuming that the forward waves in the driven and un- 

 driven lines contain all the output power, (i.e. neglecting reflection, back 

 wave in the undriven line and waveguide losses) the following table gives 

 the losses observed in the above described experiments: 



These losses may be due to circulating currents in the wires, in which 

 case the loss would be expected to increase with increasing coupling. 



Good agreement between the observed and theoretical directivities has 

 been found in the loose coupling case, but when appreciable power is 



