I realized I had assumed that the change from e to i occurred before
the i-mutation, but it could be the opposite. The change would then be
something like: *e > *ea > *ee > *e > i, which makes more sense to me.
Yet again it might be nonsense !