How do n- and p-channel fets cancel distortion?

[jwb]

I've seen some tangential claims that an n-channel amplifier stage can cancel the distortion of a p-channel stage, or vice-versa. I don't know if these claims are mentioned briefly because it's obvious and everybody understands, or if it's because nobody can explain it. But I would like to know, so I'm asking ...

Why do the Aleph amps have p-channel inputs and n-channel outputs?

Why do Krell amps have two diff amps in the voltage gain stage, one all-npn and one all-pnp?

If I built a diff line amp from all n-channel jfets, will there be distortion products that could have been cancelled out if I had mixed in some p-channel devices?

PS I'm not talking about push-pull, or complementary, circuits. I understand how those can cancel even-order distortion. I'm talking about distortion cancellation between seperate stages of an amplifier.

 

[Nelson Pass]

Why do the Aleph amps have p-channel inputs and n-channel outputs?

Because that's the most convenient way to get full swing (rail
to rail) gain while maintaining high power supply rejection.

The Alephs do not rely on P and N channel cancellation for
lower distortion, as the gain curve of both varies in the
same direction for a given signal.

 

[jwb]

In the interest of being specific, from your Zen V4 article:

Third, the P channel non-linearity will tend to operate in opposition to the N channel distortion of the gain Mosfet, giving some distortion reduction due to cancellation

I don't understand how or why that would work.