[greybeard]

Hi Bill, and thanks.
My idea was that if the axis of the capstan was rotated horizontally, away from the feed direction of the cable, with the angle chosen correctly, the offset of one turn would exactly equal the thickness of the cable, so the outgoing line would always be in the same position.
I did think this morning that the reason my idea was wrong was that on meeting the angled capstan, the cable encounters a slight turning force which, due to the stiffness of the cable, will result in a slight curvature being transmitted back to a point before the capstan is touched. So the cable is given a small lateral displacement before it becomes "locked" to the surface of the capstan.
Hope that makes sense.
I'm going to try angling the capstan downwards instead of horizontally(or as well as), so the effect is in the same plane as the curvature of the surface, and see if that makes a difference.
John