168 Lecture 10 
coastal areas is subject not only to considerable variation in the size of its par- 
ticles (from mud to pebbles) but also toirregularities such as sand waves due to 
undercurrents, it would seem somewhat pedantic to attempt to scale down sand 
and mud particles. It was considered reasonable to use fine sand, particle size 0.1 
to 0.3 mm, linear, which is small compared with the shortest wavelength of 
sound (about 1.5 mm) employed in the experiments. In the early stages of the 
investigation it was found that the bottom-to-surface oscillograph records were 
very similar whether the steel bottom of the tank was covered with fine well- 
wetted sand (free from entrapped air) or with a 1, -in. -thick layer of rubber 
sheet. This was very fortunate, for the removal or insertion of the rubber sheet 
was a very simple matter compared with the corresponding operation with a layer 
of sand, and incidentally was much more reproducible. In what follows, I shall 
therefore refer to the bottom covered with the rubber sheet as equivalent to that 
covered with well-wetted, air-free sand or mud. The steel bottom has been re- 
garded as acoustically hard, equivalent to rock, but this of course was not ob- 
vious and required some experimental proof. In the first instance it was nec- 
essary to find out how much high-frequency sound (0.5 to 1.0 Mcps) was actually 
transmitted along the bottom of the steel tank containing water. A small (1 cm in 
diameter) quartz receiver was attached in good acoustical contact to the under- 
side of the steel bottom, while a similar quartz transmitter was arranged face 
down in midwater on the movable trolley. As the transmitter was moved away 
from a point directly above the receiver, the received signal strength was meas - 
ured by the CRO and an attenuator. At a frequency of 1 Mcps, the attenuation 
was found to be 40 db/m. With the steel bottom covered by a sheet of '4-in.- 
thick rubber, the attenuation increased to 300 db/m. These attenuation values 
relate to sound incident normally on the steel or rubber surface, and it is rea- 
sonable to assume that they may be even greater at small grazing angles. As a 
further check on the advisability of using '4-in. steel as equivalent to hard rock 
bottom, comparative observations were made when the bottom was sheet steel 
uf in. thick or concrete 24 in. thick (equivalent to 2000 ft, full scale). Bottom-to- 
surface records in bothcases, the formerinthe large 20 ft steel tank and the lat- 
ter in the small concrete tank, showed similar characteristics as regards the 
number of maxima in the same depth of water at the same frequency. Similar 
records were also obtained when the bottom of the steel tank was covered with 
‘/,-in,-thick plate glass. In all the above tests point (omnidirectional) transducers 
were used, and both cathode-ray-oscillograph photographic recordings and 
logarithmic (db) pen recordings were made. In Fig. 10.8 is shown a series of com- 
parative bottom-to-surface records, using a Bruel and Kjaer logarithmic re- 
corder, in which the nature of the bottom and the frequency are varied. In all four 
series of records, a point source was used; the bottom was successively aia 
steel, 0.1-in. rubber covering 1, in, steel, 0.1-in. rubber covering concrete, and 
bare concrete. The frequencies used are, left to right, 430, 189, 160 (approxi- 
mate), 92, 52.5, and 25.6 kcps. These records were all made in the small con- 
crete tank so that all coverings mentioned, rubber and/or steel, were overlying 
the thick concrete bottom. Important features to be noted are (1) the general re- 
duction in the number of maxima or minima as the frequency decreases (or wave- 
length increases) and (2) the reductionin the number of maxima and minima when 
