500 



PROCEEDINGS OF THE AMERICAN ACADEMY 



taken to avoid such variation as far as possible. It was found possible 

 to reduce this to a very small amount, by always keeping the bellows 

 full by gently and continually moving the feet. The same result 

 could also be secured by filling the bellows full of air, and then allow- 

 ing the pressure to diminish gradually, not forcing in air while the 

 beats were being counted. This sufficed, however, only when the 

 duration of the sound was brief. 



In countiug the beats we were usually able to count either ten or 

 twenty in a series, and six or eight series were taken in succession 

 before passing to the following note. The extreme difference in the 

 duration of a series of ten or twenty beats was very rarely over five 

 tenths of a second, and generally was less than this. The times were 

 estimated by means of a stop-watch reading to eighths of a second. 



Three complete and separate sets of comparisons were made at 

 different dates between all the forks and the corresponding reeds of 

 the cabinet organ. The rate of the highest note measured — that is, 

 C^, an octave above middle C — was in each case taken as that on 

 which to construct the calculated scale, as experiment showed that the 

 beats were better defined in the upper portions of the scale. The 

 calculated vibration periods of the isotonic scale thus determined were 

 compared with the periods of each one of the notes comprised in that 

 scale as given by the organ, and the differences noted. The organ 

 had, of course, been tuned from a C fork. 



The following table gives the difference between the observed and 

 calculated values of each note, assuming in each case, as just stated, 

 that the scale was based on the high C. The differences are called 

 positive when the observed value is the greater. 



