Which wing stalls

[24Carrot]

It helps to think of the level turn first. In a level turn, once the bank is established the ailerons are returned to nearly neutral. To maintain the bank angle the AoA of the faster outside wing must be slightly lower than the slower inside wing so there is a very little up aileron.

Now imagine the aircraft rising. The airflow is more nearly parallel to the pitch of the wings, so the AoA is reduced on each wing. The reduction in AoA is the climb speed divided by the wing speed. The outside wing is faster and sees a smaller reduction in AoA and lift. This explains overbanking.

Some up aileron on the outside wing corrects the overbanking, but leaves the inner part of the outside wing at a high AoA, which could then stall when the pilot pitches up to regain lift.

Now imagine the aircraft descending. The airflow is now less parallel to the pitch of the wings, and so the AoA increases. Once again the faster outside wing sees the smaller effect, and there is underbanking.

Some up aileron on the inside wing corrects this, leaving the inner part of the inside wing at a high AoA, which could then stall.

My explanation doesn't exclude the down aileron wingtips stalling. For that we have to look at the size of the forces. Taking any of these turns as an example, and comparing the lift on each wing section to a non-turning manoeuvre at the same g-factor, the inner sections have some asymmetric lift. The outer sections have ailerons applied to cancel out their own asymmetric lift, and then a bit more to counter the rolling moment from the inner sections. But because they are further out, the outer sections have greater leverage, and a smaller asymmetric lift will do the job. So the inner sections have the highest asymmetric lift, and one of them will have the highest Lift Coefficient of any wing section, and (theoretically) it is the only stall candidate.

Personally I never liked the spiral staircase, though I agree with Unhinged that it does explain the change in relative airflow. I think it is more natural to think of the angles made by the ascent/descent speed and the different wing speeds, even if it is the same thing in the end.