Disregarding distortion in the downstream electronics, a capacitor microphone generates principally second-order distortion due to the single-sided construction of its capsule. So if the microphone datasheet specifies, say, 1% distortion at 160dB SPL, we can assume that distortion to be predominantly second harmonic.
This allows us to estimate pretty accurately what the microphone's distortion will be at lower SPL, because for each 20dB reduction in input level we can expect the relative distortion level to reduce by 20dB also. So at 140dB SPL, in this example, the second harmonic distortion will be ~0.1% (−60dB), at 120dB SPL it will be ~0.01% (−80dB) and at 100dB SPL it will be ~0.001% (−100dB).
Unless we somehow compensate for this distortion, this means that at 100dB SPL we can't accurately measure second harmonic distortion (and thus THD) below about 0.01% (−80dB). At 90dB SPL the lower limit is about 0.003% (−90dB).
If we do wish to compensate for the distortion, then we need to know its level more accurately than the above calculation allows. In other words, we have to measure it. With microphones this is not straightforward. The easiest method is to use two loudspeakers, each generating one frequency component of a two-tone test signal. As each loudspeaker will produce harmonic distortion, the microphone distortion is determined from the intermodulation products that it alone generates.