844 BELL SYSTEM TECHNICAL JOURNAL 



about 1.4. Phase velocity in these rods was not measured and is not 

 available from theory. Referring to Fig. 4, however, we must assume at 

 least 0.47r radians of phase retardation to explain the increased gain. When 

 the pattern for the 6X rod is compared with the sharpest pattern (j8 = 6.5) 

 in Fig, 3, the observ^ed characteristic is sharper than expected even with a 

 phase retardation of tt. The amplitudes of minor lobes are in good agree- 

 ment. Attenuation, as revealed by the amplitudes at minima in the 

 patterns, is apparently appreciable but not serious. 



The principal defect of the uniform polyrod is the strong minor lobes. 

 This is remedied by tapering the amplitude of radiation symmetrically 

 about the midpoint, as suggested in Fig. 5. To obtain such tapering let 

 us start at the waveguide end with a relatively thick rod. From Fig. 1, 

 this tends to retain a larger fraction of the power and should therefore not 

 radiate so strongly. Let us decrease the cross-section gradually in pro- 

 gressing along the rod, thus increasing the power radiated. Upon reaching 

 a point near the center, we find the power in the rod already considerably 

 diminished by the radiation which has already taken place. Beyond this 

 point, gradually decreasing radiation is automatically secured with a uni- 

 form cross-section as a result of previous radiation. 



This line of reasoning, calling for a polyrod tapered down in cross-section 

 only in the first half of its length, is verified experimicntally. Since detailed 

 field analysis is not available for the polyrod, the most favorable proportions 

 have been found empirically. Three examples will be described. 



Figure 7 shows a 6X rectangular polyrod linearly tapered for a little more 

 than half its length from a base ^X square to a rectangular section |X by ^X. 

 the remainder being uniform. The tapering is confined to the magnetic 

 plane. Measured phase velocity and directional pattern are included in 

 Fig. 7. By reference to Fig. 5, the observed minor lobe amplitudes cor- 

 respond to a value of a somewhat less than 0.5. The gain, considerably 

 improved over the uniform rod, nnplies from (2) a value of 1.86 for A in 

 remarkable agreement with Fig. 4. 



Figure 8 shows data on a 6X cylmdrical polyrod linearly tapered for about 

 half its length from a diameter of 0.5X to 0.3X with the remainder uniform. 

 The pattern is very similar to that of the preceding example; the gain is 

 slightly reduced, and A = 1.66. From Fig. 1, e = 2.5, about half the 



power is internal for — = 0.5, while less than one-tenth is internal for 0.3. 

 X 



Agreement between Figs. 2 and 8 for phase velocity is fairly good. 



Figure 9 gives information'" about an 8.65X radiator which resembles the 



" Supplied by Mr. C. B. H. Feldman. 



