ii8 SCIENCE PROGRESS 



and subsidiary foci, the main foci being the points at which 

 light would be concentrated by reflectors of the same shape. 

 Watson's experience in the Illinois auditorium would seem to 

 show that if sufficiently drastic measures are taken to destroy 

 the main concentrations the subsidiary ones are eliminated also. 



The effects of diffraction have been further considered by 

 Tallant (14), who shows the importance of these in estimating 

 the acoustics at points not far in front but at a considerable 

 distance laterally from a speaker. His analysis confirms the 

 view of Lord Rayleigh that a whisper is essentially a directed 

 sound and shows that this applies, though in a less degree, to 

 sounds of lower pitches. He concludes that the floor of an 

 auditorium should be fan-shaped. 



The question of acoustic isolation is a subject by itself, and 

 it has received attention from W. C. Sabine (15), P. E. Sabine 

 (16), F. R. Watson (17), and others. Lack of space precludes 

 any account here of the experimental work done, but the con- 

 clusions reached are broadly these : 



Sound-proof partitions should be as rigid and free from air- 

 passages as possible. Any necessary openings for pipes, venti- 

 lators, doors and windows should be placed in outside or corridor 

 walls where a leakage of sound will be less objectionable. 



In the case of a sound generated in the building structure, 

 e.g. by the vibration of machinery fastened to the floor, a break 

 in the structure must be arranged so as to interpose a medium 

 differing in elasticity and density. A layer of hair-felt or similar 

 air-filled material might be placed between the base of the 

 machine and the floor, but bolting through this material would 

 almost destroy the insulation. 



Special attention must be paid to the ventilation system ; 

 without such attention it is a waste of effort to construct sound- 

 proof walls, doors, etc. The system must either be omitted 

 altogether or constructed in a special manner. Pipes small in 

 diameter and extending without break from the air-supply 

 chamber to the room have been successfully used. 



The conclusions of this review of existing data may be 

 summarised thus : 



Sufficient information is available to enable satisfactory 

 acoustics to be ensured in advance of construction, the general 

 principles to be followed being : 



(i) The size of the room must not exceed certain limits. 



(2) Curved walls are almost always bad, a partial exception 

 being the case of a paraboloidal wall behind the speaker. If 

 there is doubt in any particular case the matter can be settled 

 by constructing a small model of the room and photographing 

 sound waves in it. 



(3) When the size and shape^have been fixed the materials 



