I don't know what a "bent" transfer function is, but I'd need to know the specifics of the test. What frequencies, what levels, what amounts of each distortion type, and so forth. I know you can do the opposite: create more IMD than THD using a "full wave" multiplier. But anything that changes the waveform enough to add THD will add similar amounts of IMD. Not that John Atkinson is what I'd call a reliable audio reporter anyway.Again your statement is false. You can have rather low IMD while THD might be ten times higher or more. Digital audio is a good example, as aliasing is a form of IMD, while THD is almost non-existent. You can't have it both ways!!
Instead of trying to make the world wrong why don't you read the article at the link instead of the innuendo? (BTW a bent transfer function is one that has a bend in the linearity curve) Its all there. I'm not a fan of John either (having nothing to do with his testing); if he makes a report on a test, then the test is pretty reliable (I caught him once on a bug in one of his tests, but the bug related to a design flaw in his test gear at the time, so that was understandable).
Again: Read Norman Crowhurst. This will give you a good grounding.
If you want to know how to build an amplifier (not that you do, this is academic) that can have low IMD while THD might be much higher, you have to start with zero feedback, and then sort out where the distortion sources are and reduce or eliminate them. Victor Lamm knows his stuff- very impressive specs for an SET.
So I just provided evidence, but you aren't interested. That is the definition of a skoftic which is very different from a skeptic.
JA and John Curl **are** evidence. Apparently you don't know the difference between evidence and proof. I think you are asking for proof but at the same time your website isn't that either. Any of your sound files can be doctored and there is no way a person going to your site has any way of knowing if they are the real thing or not. IOW, your site is 'evidence' as well, not proof. I also detect a bit of hubris- that somehow you place yourself above accepted masters in the field.
John Curl is not just evidence but proof in that his designs are well-recognized decades on. Clearly he knows his stuff and he's been at if for decades. You're really going out on a limb trying to make me wrong by attempting to impugn John Curl!!
I recommend you get an oscilloscope and look at what waveforms look like. A 'scope allows you to do that- to view the same signals that might be driving an amplifier or loudspeaker. You can also view distorted waveforms on a scope and compare them to undistorted waveforms. I think you might already know this although your posts seem to belie it; so if you are looking at a distorted waveform, how can you tell? If you are able to answer that it might give you a clue as to why distortion is much more audible than you think.
One other thing: you are ignoring is how the ear works. It is far more sensitive than test equipment to higher ordered harmonics and there is a reason for that. Again, an example of not knowing what you don't know. Without that understanding, I can see it would make it hard to see how the ear could detect distortion of only 0.005% or the like. I suggest you read up (perhaps set up that demo I posted earlier) and see for yourself.