This relation gives the following: 



Aa o = ± 0. 2 db/kyd for a^ = 12 db/kyd 

 Aa Q = ± 1 db/kyd for a k = 40 db/kyd 

 Aa o = ± 20 db/kyd for a h = 200 db/kyd 



The above computation considers only the accuracy 

 with which the reverberation curves can be read. Spurious 

 resonances caused by inhomogeneities in the cavity equip- 

 ment or in the liquid sample will give rise to additional 

 inaccuracies in the measurements. The encircled points 

 in figures 9 and 10, marked 18. 5° C, were recorded on 

 two different days, and show that the measurements can 

 be repeated within a few db, at least in the part of the 

 curve of interest for the data reduction. Inspection of 

 figure 10 shows that the measured points are essentially 

 contained in the interval Aa Q = ± 5 db/kyd for the part of 

 the calibration curve above 5. 1 kc/s, which is the part of 

 greatest interest for the data. 



The excess attenuation, a ex is computed as the differ- 

 ence between a, , and a , and Act must therefore be 



tot o ex 



larger than Aa o . It has been assumed that excess attenua- 

 tion is present in a sample, when the measured value for 



l ex 



Remarks Concerning the Bubble Measurements 



The original measurements on a captured bubble are 

 shown in figures 5 to 7. The transducers were adjusted 

 with no glass rod in the cavity and not changed during the 

 measurements. The value for a obtained by this setting 

 was used as the value of a , and the excess attenuation, 



74 



