I was browsing through a couple of old "acoustics" text-books over the weekend, and came upon a description of some experimental work which attempted to quantify the degree to which the ear generates harmonics of pure tones. It struck me as rather elegant, and the method used was as follows:
1. The wanted tone is applied to the human subject at a known sound-power level (raised in stages to eventually generate an audio level v distortion curve).
2. A second pure tone, set to a frequency
slightly different to the harmonic to be investigated, is brought up in level until the subject reports a
low-frequency beat of maximum amplitude: at this point the second tone is at the same level as the harmonic generated by the ear.
3. The experiment is repeated at several wanted frequencies and for as many harmonics as desired.
The results suggest quite high levels of "distortion", even at modest sound-levels (c. 50dB spl).
A related topic, did you know that hearing loss with age is (on average) quite dramatic: around 30 to 40dB at 4 to 6kHz for 60-year-old men!
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2802451/
John