> Quote - I don't think I've seen this problem with propagatingScale, but I wouldn't swear by it.

Yes, propagatingScale seems to avoid the problem (I was mistaken about that before).

Quote - It may be possible to compute a combination of x-, y- and z- scale values for the thumb that would make it scale consistently with the hand.

That possibility also crossed my mind, though I haven't plucked up the courage to dive in and try it as yet.

Attached above is ERC injection I have so far, in case it is any use to you as a basis for modification. Of course it is meant  to be used with the smoothScale channels removed from the hands and fingers.