I do know that Lewisohn states clearly that this recording is one complete take, and that the following take was useless because John's voice was shredded. Having said that, I'll try and make this next sentence clearer, because it's causing some readers stress ...

>This seems like an edit between two takes, although only one basic take was recorded (the second being abortive due to John's voice being gone!).

I say "seems like" because it is uncharacteristic for John or Paul to sing out of tune like that, and it is odd (fortuitous?) that he corrects it on a beat. This makes it feel like, I say, feel like, the joining of two takes, or of an edit piece. But as this cannot be the case, then John just snapped his pitch down realising his mistake.

Thanks to Stephen Moss for pointing out the error that no-one noticed, all of this was originally attributed to Paul's voice. The order of the "aah's" is clearly John first, Paul and George together next (but one of these is very faint). We discussed this, and came up with the following. Discuss ...

1 bar John (Note A)
1 bar Paul (Note C#) and George faintly, fade in (Note F#) (Making chord A6th)
1 bar George (Note E)
1 bar Paul (goes to Note G) (Making chord A7th)
Then Paul and John jump up notes, I believe
John takes Paul's note (Note C#), Paul goes wild.

Let's see how that grabs people ... either way it's still an anomaly :)

 This has also been suggested as a percussion sound (stick on rim), or hihat.