12 
PACIFIC SCIENCE, Vol. XXII, January 1968 
recordings was 18.75 cm/s ec (7.5 in/sec). The 
tanks were 1.2 m X 0-6 m X 0.6 m deep, 
with a capacity of 758 liters. Water tempera- 
tures ranged from 21° to 23°C. Individual 
fish of the same species usually differed in size 
by no more than 7.5 cm. 
All Bimini observations and experiments 
were carried out in November 1965 and Jan- 
uary 1966. Specimens of B. vetula and M. 
piceus were captured at depths of 2-7 m by 
hook and line and recorded hand-held under 
water in concrete tanks, 1.8 X 0*9 X 0.6 m 
deep, using an Ampex (601) or Uher tape 
recorder. Specimens ranged from 18 to 26 cm 
in total length. The record level was adjusted 
so that sounds of normal fish peaked at no 
more than -1 on the VU meter of the Ampex, 
and at about the same level on the Uher (VU 
meter not calibrated). 
A series of operations were carried out on 
R. rectangulus, B. vetula, and M. piceus. Sounds 
from fish were recorded before and after re- 
moval of part or all of the pectoral fin on one 
or both sides of the body. The role of the air 
bladder was determined by piercing the drum- 
ming membrane with a syringe and replacing 
most of the gas with water. The record level 
remained constant during all recordings (after 
an initial adjustment for each normal fish), so 
that relative changes in intensity of sounds pro- 
duced by normal and operated fish could be 
measured. Only one recording system was used 
in any set of experiments. All fish were given 
at least 10 minutes to recover from handling 
and operations. No anaesthetic was used. The 
first five pectoral fin sounds produced by normal 
and operated fish were usually chosen for anal- 
yses. However, a few sounds of the fish were 
always masked by louder noises caused by tooth 
stridulation or violent contractions of the body. 
These could be recognized easily and were not 
included. Therefore, only the first five pectoral 
fin sounds free of these disturbances were ana- 
lyzed. 
Sound durations were measured from oscil- 
lographs made with a Fairchild oscilloscope 
(701) and Grass Kymograph camera (C-4) at 
film speeds of 250 mm/sec. The effect of op- 
erations on the overall intensity of fish sounds 
recorded on tape was determined. A General 
Radio Co. Impact-Noise Analyzer (Type 1556- 
B) was connected to the output of the tape 
recorder and a peak sound pressure value was 
determined for the loudest of the first five pec- 
toral fin sounds of a normal fish. The peak 
sound pressure of the same fish after the opera- 
tion was also obtained. The peak value for the 
normal fish was considered as 0 decibel, while 
the value for the operated fish was considered 
as positive db (if the value exceeded that of 
the normal fish) or negative db (if the value 
was less than the normal fish). These changes 
in relative sound pressures (and all others men- 
tioned below) were measured in db relative to 
0.0002 microbar, but are not related to the 
original underwater sound pressures. Compar- 
isons between pressure levels of normal and 
operated fish could be made because recordings 
were carried out at the same record level and 
with equal distances between the sound source 
and the hydrophone. 
Relative sound pressures at various octave 
band frequencies were also measured from tape 
recordings. The output of a General Radio Co. 
Octave Band Noise Analyzer (Type 1558-A) 
was connected to the input of the Impact Ana- 
lyzer. Sine wave signals of either 400 or 1000 
Hz were applied to the input of the Octave 
Band Analyzer when in the "all pass" filter 
position, and with the preamplifier set to the 
20 kHz weighting (essentially flat response 
from 20 Hz to 20 kHz). The Impact Analyzer 
was then calibrated to give a peak sound pres- 
sure value 3 db higher than the root mean 
square value shown by the Octave Band Ana- 
lyzer for the sine wave. After calibration, the 
tape recorder output was connected to the input 
of the Octave Band Analyzer and readings for 
the fish sounds determined from the Impact 
Analyzer at various filter positions on the Octave 
Band Analyzer. The loudest of the first five 
sounds produced by each of ten normal fish in 
each species was measured, and considered as 
0 db. Then, the sound pressures of the same 
sound in each octave band were compared to 
the level obtained for the unfiltered signal. The 
sound pressures of the filtered signal were al- 
ways less than the total sound pressure, and this 
decrease was measured and expressed as nega- 
tive db relative to the total sound energy. 
The behavior of fish in the field was studied 
using scuba gear or, in clear waters, from 
