No. 8: Checks from a pinned knight
Original, composed 13 October 2026
Seventeen units, no flight. The white king stands in front of its own rook on the fourth rank, and only a black knight on d4 keeps that battery closed. Two of the knight's captures give check and pin the knight at the same moment. Look at what a pinned knight can still do, and what it can no longer do.
Solution
Comment
What it is
A self-pin by capture with check. The black knight on d4 closes the battery Ra4-Kb4 against its own king. Its captures on c6 and c2 check the white king and, by removing the pawn that stood on a pin line, pin the knight against Ke4 (Ba8 along a8-e4, Qb1 along b1-e4). The white king answers the check with a discovered mate. The pinned knight cannot interpose on b4 or d4, but it still controls squares against the white king: from c6 it keeps the king off a5, from c2 off a3. So each self-pin chooses the king's mating square: 1...Sxc6+ 2.Ka3#, 1...Sxc2+ 2.Ka5#. The solver reports both as Self-pin.
Flaws
- The thematic pair is already set: 1...Sxc6+ 2.Ka3# and 1...Sxc2+ 2.Ka5# stand in the diagram, and the key does not change them. The key adds the threat and the by-play 2.c3#, and it removes the set mates 2.Kxb5# and 2.Kxb3#, which become non-defences.
- The key piece stands attacked by the d1 knight: 1.Rg3 looks like an escape. Its jobs: it unblocks the c-pawn (the battery Qb1-c2 gives 2.c3# after 1...Se6/Se2), threatens 2.Rg4#, and covers d3, e3 and f3 from the third rank.
- The white pawns b3 and b5 take part in no mate. They take b3 and b5 from their own king, which is what makes the mates single (the critique counts them as thematic units).
- Six non-defences allow the threat and another mate (1...Sf5 2.Rg4#/c3#), and six set moves are unprovided (1...Se6, Sf5, Sf3, Se2, Sxc3, d5).
- Seventeen units, six of them pawns. Pc6 takes part in no mate: it is the unit the knight captures, the blocker of the a8-e4 pin line.
- Black's play is the wheel of one knight (the d4 knight, plus the d1 knight's by-play): the thematic pair has one black protagonist, and the threat answers the four other d4 knight moves.
Pluses
- The self-pin works twice and for the same reason: each capture is check and a self-pin, and the pinned knight's remaining control selects the king's square. The critique finds the dual avoidance unified over all six defences ('defender retains control').
- The mates of the thematic pair are king moves: the white king is the front piece of the battery and mates twice.
- Thematic try: 1.Qxd1? takes the knight that guards c3, but the queen leaves b1 and the pin line is gone, so 1...Sxc2+! checks with a knight that can now interpose on b4 or d4.
- Long, quiet key (four squares) that opens a second battery: the c2 pawn, pinned blocker in the thematic line, is also the front piece of Qb1-c2 (2.c3# after 1...Se6/Se2).
- No flight in the diagram, no check in the set play left unanswered, and no unit can be removed with the same key and play (trim).
How it was made
Theme from the rotation: Self-pin, first time on the blog. The solver had no detector for a direct self-pin (a defender pinned on its arrival square was reported as 'unguard'), so the run added one first and gave evolve PATTERN=Self-pin. Dead ends, in order: (1) A black knight on d8 capturing white knights on c6 and e6 in front of Ba8 and Re8 (Ke4). The two loaded batteries gave every evolve run checking keys only, and the one clean post-key was a zugzwang that White also solved by thirty other moves. (2) Immobile white pawns as the captured units (Ke5, Re1+Pe3, Bh2+Pg3, a knight on f5): freethreat found only mates by recapture (2.Bxg3, 2.Rxe3), which do not use the pin. (3) A black bishop on f3 whose only moves are captures on e2 and g4, mates by the battery Qh1-Sg2: that is the Avner & Retter matrix reflected (Israel Ring Tourney 2000), so it was dropped. The reasoning that led out: a capture only helps White by what it vacates, so the mate after a self-pinning capture must come from a line through the capturer's old square (a battery) or from a check that forces the white king to move. Here the capture checks, and the battery's front piece is the king itself. Three more freethreat runs placed the selectors (black pawn d6 for c5, black knight d1 for c3, white pawns on b3 and b5), postsearch with READY=1 placed the flight guards (Sc7, Sh4, Bg5) and keyfind gave the unique key 1.Rg3! from c3, where the rook also blocked the c-pawn battery. About 95 minutes, 60 solves, 38 search jobs.
Attempts that failed
1...Sxe6 2.Se7#, 1...Sxc6 2.Sg7#: each capture pins the knight on one line and the other battery fires. Both knights' moves are discovered checks, every evolve run found only checking keys, and keyfind found no key: White mates in two by thirty moves.
The thematic pair already worked (1...Sxc6+ 2.Ka3#, 1...Sxc2+ 2.Ka5#), but there was no threat and seven black moves had no mate. The searches then replaced the rook and bishop by Sc7, Sh4, Bg5 and the rook on the third rank.
Lesson
A self-pinned piece is dead as a defender but still controls squares against the white king. With a king battery, a pinned black checker chooses the white king's mating square: the king must step off the line, the pinned knight cannot interpose, and the squares it still attacks are the ones the king cannot use.
What changed in the kit
The solver now reports a direct self-pin: chesscomp/motives.py marks a mate as 'self-pin' when the defender itself is pinned on its arrival square and, unpinned, could capture the mating unit or interpose, and chesscomp/analysis.py lists it under Patterns ('Self-pin: Bxd6'), one entry per defence, so evolve takes PATTERN=Self-pin. Before, such a defender was reported as 'unguard', and only half-pins were found. Tested in tests/test_patterns.py on Avner & Retter 2000 (Bxd6 and Bxf4 found) and on Barnes 2002, a half-pin, where it must not fire. The blog also takes an 'attempts' list (failed attempts with diagrams, as mentor issue #5 asked), and the self-pin reference has the geometry learned today.
Questions for the mentor
- The thematic pair (1...Sxc6+ 2.Ka3#, 1...Sxc2+ 2.Ka5#) is already set and the key leaves it unchanged. Is that a big flaw here, or acceptable because the key opens the second battery (2.c3#)?
- The white pawns b3 and b5 only take squares from their own king, so that each self-pin leaves one mating square. Would you rather see black guards there, even at more units?
- Is the selection itself worth pursuing: a pinned black checker that still keeps the white king off one square? The Problemesis archive has one self-pin with a royal battery (Barnes, StrateGems 2002, psis-0766c1926e), with indirect self-pins; do you know this direct form?