280 



EXPERIMENTAL PROCEDURES 



the point method. But the amplitude measured is 

 not the amphtude at that point, but the greatest 

 amphtude in a band three ping lengths long and 

 centered at the designated point. Corresponding 

 amplitudes are measured for all similar records; al- 

 lowance is made for gain changes and projector-re- 

 ceiver calibration; finally, after converting to deci- 

 bels, the average reverberation levels are plotted as a 

 function of time. As a procedure for plotting rever- 

 beration data, the band method seems definitely 

 superior to the point method. The amplitude cor- 

 responding to a particular point is much easier to 

 obtain with the band method, since it is simpler to 

 pick out the maximum amplitude in an interval than 

 to measure the amplitude at a predetermined point. 

 Also, amplitudes obtained with the band method 

 show much less fluctuation than amplitudes obtained 

 with the point method; an analysis of reverberation 

 records consistently showed a standard deviation of 

 amplitude for the band method of less than 50 per 

 cent of the standard deviation for the point method.^ 

 It is difficult to see the exact significance of the 

 averages obtained with the band method. Certainly 

 the band method does not closely resemble the aver- 

 aging method which was the basis for the theoretical 

 formulas of Chapter 12; the point method, on the 

 other hand, does resemble it. Thus, in order to com- 

 pare the observational results obtained with the 

 band method with theoretical expectations, the sim- 

 plest procedure is to correct the band method results 

 to what would have been obtained had the point 

 method been used. The amount of this correction was 

 deteiTnined experimentally by comparing the results 

 for many records processed by both the point and 

 band methods. Except at very short ranges, it was 

 found that the band method gives results which 

 average quite consistently 7 db greater than results 

 obtained with the point method. Subtraction of 7 db 

 from the band method results thus gives average 

 reverberation levels which are comparable with the 

 theoretical expectations of Chapter 12. At very short 

 ranges, on the other hand, the reverberation is chang- 



ing so rapidly that the band method does not give 

 sufficient detail and does not show any consistent 

 relationship to the point method. 



Some more details of the present UCDWR analyt- 

 ical procedure may be of interest to the reader. The 

 individual records are analyzed by placing the films 

 in a viewer against a graph paper background. Verti- 

 cal and horizontal distances on the film can be meas- 

 ured by counting squares on the graph paper, which 

 is usually ruled in millimeters. In analyzing a record, 

 the analyst first measures the ping length in terms of 

 squares on the graph paper and converts this to 

 milliseconds by comparing millimeters and the dis- 

 tance between points on the timing trace. The num- 

 ber of timing marks in a fixed film length gives the 

 film speed from which a scale of range from midsignal 

 may be constructed. This range scale is set up next to 

 the film in the viewer. At ranges greater than 250 yd, 

 the band method is used to determine the amplitude 

 at the designated range. At ranges of 100 yd and 

 250 yd, however, the point method is used because 

 at these short ranges, as explained previously, the 

 reverberation is changing too rapidly for the band 

 method to give accurate results. 



Despite the simplifications introduced by the band 

 method of averaging, the analysis of a set of UCDWR 

 reverberation records is an arduous and time-con- 

 suming process. In the WHOI system E, the final 

 plot of average reverberation level against range can 

 be obtained immediately from the photographic 

 paper, by drawing a curve through the densest area 

 on the superposed reverberation traces. This system 

 is highly convenient for recording and plotting aver- 

 age reverberation levels; but it does not permit any 

 detailed measurement of reverberation fluctuation. 

 Probably the best system for recording reverberation 

 would combine the advantages of both the UCDWR 

 and the WHOI types. This equipment would make a 

 permanent record of all fluctuation on one recording 

 element, while on the other recording element a 

 smoothed trace would be made from which the final 

 reverberation levels could be readily obtained. 



