112 



SCIENCE PROGRESS 



Extracting from this table the times of reverberation when 

 the acoustics were declared satisfactory, we have : 



This result is of considerable practical value, especially 

 when one notes that the maximum departure from the mean 

 is -13 sees., and the average departure only -05 sees. 



Closer inspection of the large table reveals some interest- 

 ing facts. The greatest deviation from the mean was in 

 the case of room i. In this room only it was suggested 

 carrying the experiment further, viz. by removing two 

 more cushions. The reverberation period was then 1-22 

 sees., and it was pronounced unsatisfactory; 1-22 is farther 

 above the mean than -95 is below, and if one more cushion 

 had been removed instead of two the result might have been 

 closer to the mean value. In every room the chosen 

 condition came nearer the mean than any other condition 

 tried. 



As the rooms were of different sizes and the final numbers 

 of cushions were 6, 5, 15, 10, 5, there could no question of 

 prejudice. In three cases the final condition was reached from 

 overloading, in two from underloading. Also before the experi- 

 ments no explanations were given as to the method to be 

 pursued. 



This surprising accuracy of musical taste is probably the 

 explanation of the rarity with which it is entirely satisfied, 

 particularly when architectural designs are left to chance in 

 this respect. 



To explain the necessity of extending the reverberation in- 

 vestigation to different pitches Sabine cites the following 

 example : 



In an empty room with hard walls such as a church the re- 

 verberation for a violin (middle register) and a double bass-viol 

 are about the same. The introduction of cushions diminishes 

 the reverberation for both, but in different proportions, so 

 that the reverberation for the double bass is now twice that 

 for the violin. The presence of an audience increases the 

 disproportion still further. Since a difference of 5 per cent, in 

 reverberation is a matter for the approval or disapproval of 



