Naked Pair
A Naked Pair is two cells in the same unit — row, column or box — that each contain exactly the same two candidates and nothing else. Between them those two cells must hold both digits, one each, even though you don't yet know which way round. That means the two digits are spoken for: they can be removed from every other cell of the unit.
The reasoning is a short counting argument. Say r4c2 and r4c7 both show only 3 and 8. If any other cell in row 4 were 3, then both pair cells would have to be 8 — impossible, since they share the row. The same goes for 8. So 3 and 8 come off every other cell in row 4, whatever else those cells contain.
Spotting Naked Pairs requires pencil marks: look for two cells in a unit whose marks are identical two-digit sets. Be precise about the definition — both cells must contain only those two candidates. A cell showing 3, 8 and 5 does not form a naked pair with a 3/8 cell (though it might be part of a triple). If the pair sits in a box-line intersection, it can eliminate in both the box and the line at once.
Use Naked Pairs as your first stop when singles and locked candidates stall. They are common, cheap to verify, and the eliminations they produce — especially stripping a candidate from a crowded cell — often reopen the path to singles.
- Bivalue cells (exactly two candidates) are the raw material — note them as you mark up the grid.
- Two identical bivalue cells in one unit is the whole pattern; check row, column and box membership for extra eliminations.
- The pair cells themselves keep both candidates — you still don't know which cell gets which digit, and you don't need to.