I had not thought much about it, but indeed the actual harmonics are being equalized by the RIAA, so the measurement result is skewed. Using a flat amp would eliminate the equalization and give a true reading of the levels. Good suggestion! Just need to use an amp that is flat up to 48kHz to ensure accurate capture across the Nyquist bandwidth. The flat amp approach would not only give a better readout of harmonics, but also a much higher number that would be easier to analyze.
When harmonics are quoted, they are of course given with the RIAA equalization in place. I assume it would be trivial to add this equalization mathematically to the output to give numbers that would correspond better to what would be quoted by mfrs, reviewers, etc, correct?
What gain would be needed? I suppose correct resistive and capacitive loading is required to get an answer that would be meaningful for comparisons. An RC is needed for the cartridge to be properly loaded, but would introduce some filter effects. Perhaps resistive to match the output, but no capacitor? I suppose the capacitive filter effect could be added mathematically along with the RIAA.