146 DBS. GUY BARLOW AND II. B. KEENE ON THE ANALYSIS OF SOUND. 
to examine the separate responses. The results of this sweep are given in fig. 7, 
where it will be seen that no important vibrations were missed, and that their 
frequencies were obtained with almost the same degree of accuracy as in the more 
careful analysis. As would be expected, the amplitudes as determined in this rapid 
sweep were generally less than before, and could only be determined roughly. 
Other experiments which cannot be described here have shown that all responses, 
however small or irregular, which occur in an analysis can always be accounted for. 
In fact, analysis has sometimes indicated the presence in the source of quite unsus¬ 
pected vibrations which have been afterwards shown to exist by other means. 
Replacing the galvanometer by a telephone, a weak sound of machine-like quality 
was heard, the frequency I ) = 261 being most pronounced, and also a marked throbbing 
which was probably associated with the heating of B and C (89 and 100). 
(4) Analysis as affected by a General Background of Sound. —The object of these 
experiments was to determine to what extent a definite vibration could be masked by 
the presence of a general “ background ’’ of sound. A suitable background was found 
to be produced by placing a roaring bunsen near the suspended microphone. As heard 
in the telephone this background completely masked the note from a 256 maintained 
fork, so that one could not detect when the fork started or stopped. On analysis it 
was found that the background alone gave a disturbance on the galvanometer at all 
frequencies of interruption, while the fork response was 15 times that due to the back¬ 
ground, showing that in this case the analysis is vastly superior to the ear using the 
telephone. Similar experiments were made with a 512 fork, but it was found much 
more difficult to mask this note, two roaring bunsens close to the microphone being 
required. The analysis appeared less efficient in distinguishing the fork note from the 
background, the ratio of fork response to background disturbance being of the order 
8:1. The galvanometer disturbance due to the background showed a maximum when 
the speed of interruption was of the order of 1000/sec. Previous experiments have 
shown that this is the resonance region of the microphone. 
The masking effects of backgrounds appear to be of considerable interest and im¬ 
portance, and require fuller investigation. 
Analysis of Sound in Water. 
It was decided to give special attention to low frequencies ranging from about 
5-150/sec. A number of low-frequency sounders and receivers of different types were 
used, and will be described before the experiments. 
(a) Sources of Sound. 
(1) Cylindrical Sounder. (A type used by Lord Rayleigh for experiments in air, 
‘ Phil Mag.,’ 1907.)—This was a metal can maintained in bell-like vibration by means 
