Ok, but so does a floor or wall, I would have thought?
I feel that I must missing something extremely obvious here...
EDIT: having re-read some articles on the topic, it seems that the discrepancy simply comes down to an assumption in the classical theory that the reflection is incoherent with the direct sound. See for example the
article that
@March Audio borrowed graphics from earlier, which states, "the effective gain realised from
incoherent summing is +3dB for each additional boundary..."
This would seem to be a wrong assumption in the case of subwoofers placed against boundaries IMHO, since for frequencies with wavelengths that are short (<1/4 wavelength or so) relative to the distance between the sub and the boundary (i.e. all typical frequencies reproduced by a subwoofer), summing will be coherent, giving a +6dB SPL increase.
This would also seem to me to be consistent with the fact that a halving of the radiation angle should result in a doubling of the % efficiency.