The relevance of performing reverberation time measurements at very low frequencies became an issue in Sweden when the national standard recommended that impact sound insulation should be evaluated from 20 Hz for sound classes above the minimum requirement. Even though the standard states that L'_n,T is not to be normalized with respect of reverberation time for frequencies below 50 Hz, it could be argued to include such a correction term to handle any possible variation in the absorption properties of the room. But this can be done only if the reverberation time can be accomplished with reasonable accuracy. The present paper presents an empirical study where reverberation time has been measured from 20 Hz in two different bedrooms with more than 100 microphone positions in each in order to determine the spatial variation. A comparison is made between the uncertainty as a function of frequency and it is indicated that the standard deviation is larger for the lowest frequencies, below 50 Hz, compared to higher. From an engineering point of view, this can be compensated by adding additional positions to the already existing ISO measurement procedure