80 



EXPERIMENTAL PROCEDURES 



range of approximately 100 yd. During a run, the 

 sending vessel keeps its projector trained at all times 

 on the receiving vessel. This aiming is done by means 

 of a pelorus, mounted either on the flying bridge or 

 vertically above the sound projector to eliminate 

 parallax. In a recently developed installation, selsjm 

 repeaters cause the sound projector to follow auto- 

 matically the changes in bearing of the pelorus. In 

 former installations, an operator in the ward room of 

 the sending ship had to match the two bearings by 

 hand. At WHOI, projector training is now completely 

 automatic, with the help of a radio compass. 



In transmission work at supersonic frequencies, 

 ping lengths are usually either about 50 msec or 

 about 10 msec. It is believed that as the signal length 

 decreases below 50 msec aural perception of the re- 

 sulting echoes becomes more and more unsatisfactory 

 (see Volume 9 of Division 6). Very short pings have 

 an important use, however, in transmission studies. 

 If the water is fairly shallow and the ping length is 

 sufficiently short, the directly transmitted and the 

 bottom-reflected signals can be examined separately. 

 This is possible when the time resolution of the fol- 

 low-up circuit is sufficient to resolve the time differ- 

 ence of arrival between direct signal and bottom-re- 

 flected signal. The minimum requirements of resolu- 

 tion in a specific case will depend on the geometry of 

 the paths, depth, range, and refraction pattern of the 

 ocean. 



For very long ranges, the signals often arrive with 

 badly distorted envelopes and with tails known as 

 forward reverberation. When such tails are present, no 

 resolution of the electrical circuit will result in satis- 

 factory separation of the two sound paths. In the 

 absence of such tails, resolution is frequently possible 

 even with fairly long, square-topped signals ; in other 

 words, it is possible to distinguish three portions of 

 the received signal trace, the direct signal alone, the 

 composite signal, and the bottom-reflected signal 

 alone. 



Signals are emitted at a rate of about one signal per 

 second. Once a minute pinging is interrupted, and a 

 long signal of 10 seconds' duration is sent out. These 

 long signals serve two purposes. First, a received long 

 signal often provides a very instructive, graphic il- 

 lustration of the degree of coherence of the transmis- 

 sion. Furthermore, this long signal makes it possible 

 to correlate the received short sound signals with the 

 radio signal, and thus to determine the range at which 

 the signal was received. In a transmission run carried 

 out at 5,000 yd, by the time a sound signal arrives at 



the receiving ship, three additional signals have al- 

 ready been put into the water. The once-a-minute 

 breaks facilitate the identification of particular sig- 

 nals. 



The overall accuracy of the determination of the 

 transmission loss of an individual signal has been im- 

 proved steadily in the course of the transmission pro- 

 gram. A distinction must be made between the deter- 

 mination of the absolute transmission loss, and the 

 determination of the difference in transmission loss 

 between two signals received at the same range, or 

 one signal received at different ranges. The deter- 

 mination of relative loss is not affected by errors of 

 calibration, while the determination of the absolute 

 loss is affected by calibration errors. The uncertainty 

 of calibration in the earlier data taken both by 

 UCDWR and WHOI is very large and probably ex- 

 ceeds 10 db in many instances. Improvement in pro- 

 cedure has cut this uncertainty down to approxi- 

 mately 1 db. Both absolute and relative determina- 

 tions are affected by training errors of the projector 

 and by the horizontal directivity of the hydrophone. 

 Training errors are small at long range where the 

 bearing is changing slowly. At ranges of the order of 

 100 yd, where the bearing changes rapidly, training 

 errors can be significant even when great care is used 

 in following the target. The uncertainty of training 

 has been almost eliminated by improved instrumenta- 

 tion and is probably well within 1 db at the present 

 time. Even in earUer work, training errors probably 

 never caused an error in received sound level much in 

 excess of 1 db. The most recent hydrophone models 

 in use are practically nondirectional at 24 kc, but the 

 directivity of the CN-8 model used in earUer studies 

 introduced errors of about 2 db. Thus, the experi- 

 mental error of most of the transmission loss deter- 

 minations at UCDWR is probably about 2 db, while 

 for the most recent data the experimental error is 

 probably considerably smaller. 



4.3.3 Analysis of Data 



This section will be concerned with the analysis of 

 data in which single-frequency supersonic sound is 

 received by one of the recording systems with a linear 

 response. The procedure used in the analysis of runs 

 carried out with sonic sound will also be sketched. 



Figure 11 shows that the received soimd intensity 

 is subject to rapid changes in intensity, which obvi- 

 ously cannot be related to observed changes in range 

 or temperature distribution. Figure 12 shows the re- 

 ceived amplitude of a continuous, 10-sec, 24-kc signal. 



