As JR is saying, harmonic distortion can be more audible at lower frequencies than the same percentage distortion at higher frequencies. To be perceived as the same loudness as a midbass note, a very low bass note has to be much louder in terms of dBSPL. For example, to be perceived as 80 phons (phon is a measure of perceived loudness), a sine wave at 90 Hz has to be about 93 dBSPL. A sine wave at 30 Hz has to be about 110 dBSPL to be perceived as 80 phon. In contrast, at 1000 Hz, an 80 dBSPL sine wave is perceived as 80 phon. This 17 dB increase in perceived loudness going from the fundamental to the third harmonic makes the distortion much more audible. See this equal volume graph:
http://en.wikipedia.org/wiki/File:Lindos4.svg1% third harmonic distortion of a 31.5 Hz fundamental means the 94.5 Hz third harmonic is 40 dB lower than the fundamental. This would only be perceived as about 23 dB lower due to the ear's lower sensitivity at low frequencies, which would correspond approximately to 7% or 8% harmonic distortion.
http://www.sengpielaudio.com/calculator-thd.htmWhat this means for you is that a 1% 95 Hz third harmonic of 31 Hz sounds much louder relative to the 31 Hz fundamental than, say, a 1% third harmonic of 500 Hz would sound relative to its fundamental.
I can't tell you at what percentage level the harmonic distortion would have to be for you to perceive a note as distorted. There are other factors, however, and probably the physical distortion due to changes in voice coil / magnet overlap at extreme cone displacement, and possibly cone flex at loud volumes could be as or more audible than the 1% electronic distortion in your signal. In real life, some people like the sound of distorted bass notes, probably because it occurs so commonly that we have all gotten used to it.