values of these quantities may not be significant (particularly in the theoretical 

 case) their magnitudes are useful for purposes of comparison. In all cases the 

 experimental a's are greater than the theoretical values, a result which is to be 

 expected since the theoretical a's represent only the deviation of the theoretical 

 curve from a straight line and do not take fluctuations into account . The trends 

 established by the theoretical a's are followed by the experimental a's, e.g., 

 a (A3) exceeds a (A 2) which exceeds a(Aj) both theoretically and experimentally. 

 (The sole exception is the theoretical a (A3) for Curve 60.) Also a (3000 cps) 

 is greater than a (1300 cps) which exceeds a (700 cps) theoretically and experi- 

 mentally. The two most interesting observations to be made are (1) the experi- 

 mental a's increase with frequency being greatest for 3000 cps, and (2) the ratio 

 of the experimental to the theoretical a's increases markedly with decreasing 

 frequency in all cases, viz: the ratio is between 1 .5 and 2 for 3000 cps but in- 

 creases to 5 at 700 cps . These observations presumably arise from two causes: 

 the greater simplicity of the 700 cps theoretical curves and the greater sensi- 

 tivity of the higher frequency signals to perturbations such as those caused by 

 waves and surface reflections . 



A comparison between Area I and Area II can be obtained from Curves 58 and 

 62, both at 3000 cps . The theoretical a's are greater for the Area II curve 

 than for the Area I curve, reflecting the change in shape of the theoretical 

 curves, a consequence of the different acoustic velocity profiles and duct 

 depths used in the computations . However, as discussed in Appendix B, the 

 difference between the velocity gradients was close to the limit of experimental 

 resolution of the bathythermograph . The experimental a's, in agreement with 

 the discussion in the last paragraph, show the same relative behavior as the 

 theoretical a's (increasing from Aj_ to A3) but the differences between the two 

 areas as shown by the following table do not appear to be significant. 



Experimental a 





Area 



1 



Area 2 



V 



4.3 





3.8 



V 



4.3 





4.1 



A 3 



4.6 





4.6 



In fact there is closer agreement between the experimental a's in the two areas 

 than between the corresponding theoretical values. 



12 



Arthur ZD.Hittle.Ilnr. 



