I think my basic problem was in writing about the winter following the second summer as 'the second winter' -- because, as eventually established above, the 'years' in this calculation have winter first and summer second, so the third year *starts* with a third winter.