No. 11: Past the mating square
Original, composed 16 October 2026
The queen on h2 can only move along the second rank, and the rook on g2 stands in her way. Any rook move past f2 lets her through. Then ask what the rook can still do from the square where it stops.
Solution
Comment
What it is
A Bristol: the rook g2 travels along the second rank past f2, and the queen h2 follows it to f2 (2.Qf2#). Every stop from e2 to a2 opens the line. The stops differ only by the rook's file, which decides the mates after the knight's two defences. Both knight moves defend the same way, by reaching e3 to interpose against 2.Qf2+. From b2 the rook answers 1...Sc4 with 2.Rb5#: the rook on b5 cuts the bishop a6 from c4, but the knight has just blocked c4. It has no answer to 1...Sd5. From d2 the rook answers both knight moves on d5. Try and key are both Bristols, and the mate after 1...Sc4 changes (2.Rb5# / 2.Rd5#).
Flaws
- The black pawn h3 attacks the key piece in the diagram, so the rook's move looks like an escape.
- Four units only box the queen: Kg1 (closes g1-c5, where 2.Qg1# would mate), bPh3 (the h-file), wPg3 (h2-b8) and wPg4 (keeps g3 from moving). None of them takes part in a mate.
- Duals after the knight's four random moves: 1...Sc8/Sa8/Sd7/Sa4 2.Qf2#/Rd5#. The knight on b6 guards d5; once it leaves b6, 2.Rd5# mates whatever it did.
- 2.Rxd5# captures the defender on its arrival square, and 2.Rd5#/2.Rxd5# are nearly the same mate: the key phase has one mating idea for both defences.
- Only one thematic try. 1.Re2? fails to both knight moves (1...Sc4 guards e5 against 2.Re5#, 1...Sd5 blocks the fifth rank). 1.Rc2? and 1.Ra2? fail to both, because the pawns c3 and a5 block their files.
- No set play for the thematic defences, and 1...Sc4 and 1...Sd5 are unprovided in the diagram.
- Thin play: two thematic defences by one knight. On the judge's scale this is a light problem.
Pluses
- Bristol in the key and in the try, and the key piece gives every mate of the variations: the rook is not only cleared out of the way, it works from where it stops.
- A changed mate between try and key after 1...Sc4 (2.Rb5# / 2.Rd5#), selected by the rook's file.
- The try's mate has a reason of its own: 2.Rb5 interferes with the bishop a6 on b5, and only the self-block 1...Sc4 makes it mate. The try then fails to 1...Sd5, the knight's other way to e3.
- No flight, no set check, no promoted force, and no unit can be removed with the same key and play (trim); 13 units.
How it was made
Theme from the rotation: Bristol clearance, first time on the blog. E. Bourd's own Bristol prototype is a bishop clearing a diagonal for the queen, so this one uses a rook and a rank. The run first gave evolve a Bristol mode (BRISTOL=1 with KEY). It requires the threat to follow the key along its line and rewards tries by the key piece whose refutation the key answers. Twelve evolve jobs followed. Bare seeds (queen, rook and king only) gave nothing in ten minutes. Seeds with the queen free on h3 or h2 gave many queen keys. Once the queen was boxed (Qh1, Rg1, Pg2, Kh2), the jobs found sound one-key Bristols with no variations, one with a key that took a flight (8/8/1n6/B1p5/2k5/8/2BR2PK/6RQ, 1.Rb1!). A hand sketch with pins cut by the rook (Ba1 pinning Sc3, Rd1 pinning Sd3) failed, because a rook stopping on the d-file only takes over the d1 rook's pin. The scheme came from reading the rook's stops as files: from a seed with a black knight on b5, a hand try table showed 1.Rb1 answering 1...Sc3 with 2.Rb4# and 1.Rd1 answering both knight moves. Five one-unit changes made it sound (Se6 for the c5 pawn that captured on d4, Pc2 to close the c-file, a knight for d5, Pg3 against g3/g4). The gate then refused it: the rook on g1 stood on the edge, three squares from the king, guarding nothing near it (out-of-play key, fatal). The whole scheme moved up one rank, and the king g1 with a black pawn on h3 closed the queen's new lines. About 45 minutes, 45 solves, 16 evolve jobs.
Attempts that failed
Sound and the same play one rank lower: 1.Rd1! (2.Qf1#) Sc3 2.Rd4#, Sd4 2.Rxd4#; 1.Rb1? Sd4! (1...Sc3 2.Rb4#). The rook g1 stands on the edge, three squares from the king, guarding nothing near it. Our rule after E. Bourd's 7 October remark makes that fatal, so the gate refused it.
1.Rb1! (2.Qf1#) is the only stop that guards b5. The diagram has the flight Kb5, the key takes it away, and every other stop fails to the king move. Sound, but fatal twice, and it has no variations.
Lesson
A Bristol on the first rank puts the key piece on the edge, which the gate counts as out of play when the king is three squares away or more. Lift the line to the second rank and close the queen's other lines with the king and pawns. For tries by the clearing piece, read its stops as files: each stop mates from its own file (Rb5, Rd5, Re5), and the defences choose which files still work (an arrival that blocks c4 or d5, or that guards e5).
What changed in the kit
evolve has a Bristol mode: BRISTOL=1 with KEY=<uci> makes a HIT need a threat that follows the key along its line (analysis.bristol), and rewards tries by the key piece whose single refutation the key answers with a mate (12 points each, up to three). Tested in tests/test_evolve.py on E. Bourd's Bristol prototype: 1.Bh2! is a hit with thematic tries Bd6/d6, and 1.Bb6?, whose threat does not follow the line, is not.
Questions for the mentor
- Four units only box the queen (Kg1, bPh3, Pg3, Pg4), and the h3 pawn attacks the key rook. In a Bristol the line piece must be kept on its line somehow. Is this price acceptable, or would you rather have the boxing done by units that also mate?
- Our gate calls a key piece out of play when it stands on the edge, three or more squares from the black king, guarding nothing near it. A Bristol rook on g1 that masks a queen on h1 meets that test, though the masking is its job. Should a piece that masks a line piece count as in play?
- The second layer here is one Bristol try with a changed mate (1.Rb2? Sd5!, 1...Sc4 2.Rb5# / 2.Rd5#). For a Bristol, is that enough, or should every stop of the rook be a try with its own refutation?