Up to the present, the acoustics of auditoriums has rested on a purely empirical basis. A more satisfactory situation would be reached if we knew a physical quantity which had some definite value for all rooms, regardless of size, which are considered acoustically good. Such a condition is that ∫ t0t1 Lt dt = −32.6 where t0 is the time a sustained source of sound, Ē, is cut off; t, the time the sound becomes inaudible: and Lt, the loudness of the sound at any instant t. For a test tone of 1000 cycles we take Ē to be 35.3×1010 cubic feet of sound of threshold density per second, which is a fair average of sources met with in practice. All rooms which have been considered acoustically good satisfy this condition. This implies that the ear is a ballistic instrument during the period of hang over of a sound, and suggests that rooms should be adjusted so that the loudness of all pure tones shall decay at the same rate, independent of frequency. The absorption material used must have a definite frequency characteristic to obtain this result. This characteristic turns out to be very near to that which an audience of people exhibits. An audience is recognized as being nearly ideal absorption material.