Abstract: On the basis of several years of experiences in the measurement of R.T. in auditoria (with empty and full audience), and of the accumulated data experimenting with different sound sources, in this article the deviations of measurement results are examined and analysed. The coefficients of variation of measured R.T. due to recorded reading's deviation are within about 5% under carefully controlled conditions; while the coefficients of variation due to non-uniformity of sound field in normal halls come within 10% approximately. With regard to the deviations among measured results from different sound sources, they are always below the deviation scope mentioned above. Therefore no significant difference can be observed. The sources of sound, both in a steady state and in an impulse state, used for comparison, include an orchestra playing staccato chords, a white noise and a warble tone from a loudspeaker, a blast from a sport pistol and a burst of a children's toy balloon.
Occasionally large deviations are observed in the R.T. measurements in the same hall by different authors, there fore a normalized method for R.T. measurement should be considered.
In order to find the reverberation time in the full occupied condition, cotton padded garments are used to simulate the acoustic absorption of real audience and the results are quite satisfactory.