530 



HERSEY AND BACKUS 



[chap. 13 



bubble hypothesis, this is conflicting evidence. Either the depth behavior or 

 the frequency characteristics of this layer must be, in fact, otherwise than our 

 analysis methods indicate ; perhaps both are. 



It should be noted that the frequency scale of our analysis was insensitive to 

 small changes in the region between 3 and 6 kc/s. This is because the frequency 

 scale was linear and designed to cover a band from to 30 kc/s. 



In addition to the problem of recognizing frequency migration, the correct 

 interpretation of depth migration is difficult at these lower frequencies. We 



100 

 200 

 300 

 400 



-40 -20 *20 +40 



SUNSET 



TIME IN MINUTES RELATIVE TO SUNSET 



AREA A- SOUTH OF NOVA SCOTIA 

 AUGUST 1954 



Fig. 18. Depth and peak frequency versus time of observation relative to sunset of the 

 principal scattering layers in area A of Fig. 14 (cf. Figs. 16 and 17). For the low- 

 frequency layer (B), the figure shows the depth of the greatest intensity of scattering 

 ("peak") and the apparent depth where scattering begins ("top"). 



have always used a weakly directional receiver (see Fig. 15a). If the layer is 

 patchy, which we believe it is, such a transducer will not clearly define the 

 limits of the layer with each observation, since it will be sensitive to side 

 echoes from nearby patches. The layer can be better defined if the sampling 

 rate is high enough to identify side echoes, as in echo-sounding. Unfortunately 

 the explosive technique has limited the sampling rate to about one observation 

 every two minutes. 



Quite apart from problems of discriminating variations of frequency in the 

 layer and discriminating against side echoes, there is yet another cause for 



