172 



INTENSITY FLUCTUATIONS 



be very appreciably reduced below the region of 

 strong thermal gradients. 



Nevertheless, the composite evidence indicates 

 that image interference is probably the most im- 

 portant single factor contributing to the observed 

 fluctuation. There may also be interference between 

 the beamlets into which the irregular surface of the 

 sea breaks up the incident coherent beam. In the 

 deep transducer experiments, the surface-reflected 

 signal showed a very high degree of fluctuation, but 

 at glancing incidence the path differences may not 

 be large enough to bring about random interference. 



If all interference effects could be eliminated, total 

 fluctuation would probably be cut in half.^ 



The remaining fluctuation is essentially frequency- 

 independent. It may be due in part to pitch and roll 

 and in part to the lens effect of microstructure. Elimi- 

 nation of either of these effects is possible in principle, 

 but would require very elaborate additions to present 

 sound gear. 



K Fluctuation by interference can be effectively eliminated 

 by tian.smitting supersonic sound in a broad frequency band. 

 The width of that band should probably exceed 5 kc in order 

 to obtain maximum benefits. 



