I think this assumption is the only way you can rationally explain most letter overmarks. It doesn't make sense for a die to be sent from say Exeter to York (in the case of the y over E 1696 2/6d). A far more plausible explanation is that the dies were ready for despatch from the Tower mint when an urgent request came through from a provincial mint for more dies and they recut one sitting on the shelf to fill the order. Transfer of dies between provincial mints only makes sense once operations were being wound down at the end of the recoinage given the scale of the operation and hence demand for dies. The provincial mints closed starting with Norwich and York in April 1698, Chester in June, Exeter in July and finally Bristol in September 1698. The only likely overcut mint letters are therefore like to be B over E for coins dated 1698 being geographically close and which would be conveniently dropped off when the dies were returned to the Tower. The others are quite remote from each other, so as there are no 1698 coins with the overmark, it is quite possible that transfers between provincial mints never took place at all. Even if you assumed that provincial 1698 coins were struck using 1697 dies, the rarity of 1697 overmarks suggests that most were made in 1696 at the height of the recoinage, thus reinforcing the theory that the recutting took place at the Tower.