I know this has already been addressed but, I can't stop myself from responding anyway.
Two perfectly co-located subwoofers in an anechoic room increase peak SPL at any given frequency by about 6dB's because you gain 3dB's from the doubling of cone area (it's like adding 3dB's of efficiency) and because you added a second amp, 3dB's for that.
Go back to a real room and the matter becomes far more complex. If you co-locate the subs in a room, you do typically see an increase, but because the room is also dominating the response and sound, because dips are often caused by cancelation, you may not see any increase at certain frequencies. Just like EQ can't fix a cancelation dip, neither can more sub output.
However, once you start spreading those subs out you can improve smoothness and mitigate those cancelation dips from modes. So you see an increase from that just because the response becomes flatter, and you see an increase because it is more sub in the room. In a room where the subs can be roughly time and level aligned, you do tend to see a sizable increase potential, but again, in practice, that isn't how it works and what happens is maybe 1-2dB's (that can be associated with the increased drivers and amps rather than smoothness).
Now, because smoothness in a room is a function of the number of low-frequency sources defined as 1/N, the more the better. The sweet spot according to Geddes is 3 if you can use the mains as LF sources, and 4 if you are Harman and rely only on the subs.
In other words, buy the highest output subwoofer you can afford, then start saving for two more, not one more, if you want the best response and maximum output. Or, don't try to achieve a flat/smooth response over a wide range of seating positions, stick the subs right next to your chair, and call it a day. Everyone else be ....well you know.
