Breaker and O'Neil 



so that the Nyquist frequency of fj^ = 1/(2*2 sec) = .25 HZ corresponds 

 to a u of 9.6. Figure 1 shows this to be adequate to prevent serious 

 problems with aliasing (P(f) is monotone decreasing for u >1.38 ). 



In no case was it possible to tow the projector at depths 

 greater than 50 feet. This of course raised the problem of the 

 surface reflection. A pulse length of 10 milliseconds was felt to 

 be the minimum necessary to permit reliable pulse detection. Thus, 

 considerable direct -transmission/ surf ace -reflected- transmission 

 overlay was inevitable. 



Two unanticipated noise problems were encountered about 

 which little could be done, given the time constraints of the 

 experiment. The first was simply that the weather was worse than is 

 usual in the Tongue of the Ocean. Sea states rose as high as three 

 late in the experiment, and rain squalls occasionally blotted out 

 transmission altogether. The other problem was noise leakage to the 

 receiving amplifiers and recorder from the ship's ground circuit. 



On the whole, however, the experiment went well. It 

 produced a total of 124 reels of magnetic tape containing 64 hours 

 (real-time) of analogue data. During this time approximately 

 115,000 pulses had been transmitted. 



REDUCTION AND ANALYSIS OF DATA 



At this time, the reduction and analysis has not been 

 completed. The work which has been completed, however, will be 

 described, while a sketch will be given of the work planned. 



Because pulsed signals were used the data were inherently 

 suited better to digital than to analogue processing, but the 

 quantity of analogue data tapes, their noisiness, and the relatively 

 high acoustic frequencies all presented serious problems in con- 

 verting the data to digital form. 



Direct digitization of the analogue tapes was felt to be 

 infeasible due to the volume of data. Moreover, only about 0.5% 

 of the total tape time was devoted to data of interest (i.e., 

 10 ms/2 sec) so that many hours of computer time would have been 

 required to sift the output tapes for pulses. 



Figure 3 is a block diagram of the system (designed by 

 Dr. M. F. Gordon of Bis sett -Herman) developed to deal with this 

 problem. It detects pulses and digitizes the pulse envelopes. 

 These are recorded in computer-compatible format on tape, one pulse 



376 



