That's correct. Sudden and narrow dips in on-axis response cause THD+N as a ratio to shoot up even though its amount can be the same. It looks alarming there because it is a percentage. If you look at the 96 dBSPL graph where actual distortion level is shown, the spike is hardly there:
The Harmonic Distortion (HD) measurement is intended to show where there might be potential problems caused by HD.
Problems occur when the HD is no longer masked by the fundamental (here around 1kHz).
Therefore the percentage display is the more important display, because here the HD are always displayed in relation to the fundamental.
Since we are looking for potential weak points of a loudspeaker, we consider the worst possible case. This means that we assume a single tone around 1kHz as a masker tone - for example a synthesizer C6 tone.
A single tone is a bad masker, so in this case HD is only badly masked.
When the masker tone generates 60dB sound pressure (see green no. 1 in the diagram), then HD2 is masked at 2kHz up to a sound level of 8dB (test tone level). Which corresponds to a perceptibility threshold of -52dB/0.25% HD2 at 60dB.
At a sound pressure of 70dB@1kHz (see green no. 2 in the diagram), HD2 is masked up to -42dB/0.8%.
At a sound pressure of 90dB@1kHz (see green no. 3 in the diagram), HD2 will be masked up to -35dB/1.8%.
Please do not interpret the results from the diagram as absolutely valid values, but much more as a threshold with tolerances (depending on the individual listener).
Source: Zwicker, Fastl - Psychoacoustics
Related to the 805S the 1% HD2@86dB at 1kHz should not be a big problem yet.
Just one narrow band masker is sufficient to raise the threshold of harmonic distortion significantly.
I have described
this in more detail in this post
or
here in relation to a piano stroke.
In short, the dip around 1kHz (and the percentage increasing HD2) should not be ignored, because that's where a woofer surround resonance probably causes increasing HD2 which could possibly be audible.
