According to the current FIDE Laws of Chess:

3.9.1 The king is said to be 'in check' if it is attacked by one or more of the opponent's pieces, even if such pieces are constrained from moving to the square occupied by the king because they would then leave or place their own king in check.

3.9.2 No piece can be moved that will either expose the king of the same colour to check or leave that king in check.

It wasn't always so.

When was it legally possible in a FIDE rated game of chess to have your king attacked by 3 of the opponent's pieces? Please give an example illustrating this.


1 Answer 1


In my thread on the English Chess Forum, which seemed to make the world go crazy on the subject, I gave all the major and minor events in the history of the “legal” triple check that my extensive research has uncovered. This loophole was even talked about recently in Episode 20 of "The Chess Pit" at the 12:13 mark. Here is all the information in chronological order.

The dubious phrasing in FIDE's laws of chess goes back to the 1984 FIDE Chess Congress when it was implemented. This is indicated by its first appearance in the 1985 laws-many thanks goes to @Stephen for pointing this out in a comment. Here is the 1985 law on checks, in whole. (All linked laws come from the Chess Arbiter's Association website.)


9.1 The king is in check when the square it occupies is attacked by one or two of the opponent's pieces; in this case the latter is or are said to be "checking the king."

9.2 Check must be parried by the move immediately following.

9.3 If the check cannot be parried, the king is said to be "checkmated" ("mated"), (See Article 10.1.)

9.4 Declaring a check is not obligatory."

Beforehand, the 1980 laws instead stated "The king is in check when the square it occupies is attacked by an enemy piece." This itself is as vague since it can be interpreted to allow a king to be left in double check. However, since double check is widely known to be possible, such an argument obviously wouldn’t stand. Even so, it is no wonder then why they later rephrased it to "one or two” for clarity. But that still wasn't enough, as the saga of the “legal” triple check shows.

Within a few years, at least some people had realized the implications of the law’s wording. This is evidenced by a letter sent by Robert Norman, who is from Dar Es Salaam Tanzania, to "CHESS Magazine", then called "Pergamon Chess,” that was published in the June 1988 issue. This is the oldest known chess problem that utilizes the "loophole."

[Title "Robert Norman, CHESS Magazine June 1988, White To Move And Win"]
[FEN "1n5k/1r6/5K2/4Q1Pb/8/8/8/8 w - - 0 1"]

1. g6 Nd7+ 2. Kf7+ Nxe5++ 3. g7+! Kh7 4. g8=Q+ Kh6 5. Qg7#

Thanks to Jonathan Bryant on Twitter, we have a visual of Norman's letter.

enter image description here

The chess magazine EG reprinted it in April 1989, saying this about it: "Definitely an oddity. Ridiculous, irrelevant, amusing or disturbing, as the reader chooses."

Of course, it is arguable as to whether or not it can be considered legal. It can be countered that triple check is indeed accounted for in the letter of the law. The aforementioned EG magazine summarized it neatly: "The argument is that check by three men is not in check. Well, one can easily counter that check by three men includes check by two and is therefore covered (not 'specifically excluded') under the Laws, which do not need to be changed." Even under the counterargument, the idea of the triple check alone was enough of an ambiguity to change the law's wording, which definitely says something about it.

Norman’s problem is an online rarity online, so here are a few links to it.

1- A Dutch blog for bizarre chess problems. #2-An obituary for Robert Norman. #3-Page 29 in the 12/2001 issue of Internet træk.

A few years later in 1992, the idea again appeared in CHESS Magazine, then called Maxwell Macmillan Chess, in the column "Addicts Corner" by Richard James and the late Michael Fox. I found the relevant part of the February issue in an old Google forum.


Right, N. Short, J. Speelman, J. Nunn, and the rest of you clever clogs. Break your heart on this: White: Ke3, Qb2, Ne5 Black: Kh8, Re8, Rg8, Ba7, Nb6, Pg7, h7 Black incautiously played a double check 1...Nc4+?? How did White crash through to victory? Admit it; you're baffled. Deduct one hundred Elo points [and read on]:

Dead easy. White gets out of check by 2. Nf7 mate. Yes, it's perfectly legit. Consult Article 9.1 of the laws: "The King is in check when the square it occupies is attacked by one or two of the opponent's pieces..." It says nothing about a threefold attack, so White's move takes him out of check..."

[Title "Kevin Thurlow, CHESS Magazine February 1992"]
[FEN "4r1rk/b5pp/1n6/4N3/8/4K3/1Q6/8 b - - 0 1"]
[startflipped ""]

1... Nc4+ 2. Nf7#!

It seems that Thurlow's problem spread like wildfire, and FIDE finally got wind of it. In a short time, the wording was soon changed to the modern wording at the 1992 Fide Chess Congress for the 1993 laws-"The king is in "check" when the square it occupies is attacked by one or more of the opponent's pieces"

You can find various references to the subject on the English Chess Forum. Most are by Kevin Thurlow in reference to his problem; others are by Stewart Reuben, an English arbiter who was on the rules commission, and helped with the fix after its 1992 rediscovery.

In conclusion, from 1985, when the effect of the 1984 FIDE Chess Congress change took place a king could hypothetically be left in a “legal” triple or higher check in a FIDE rated game until 1993 when the law was again rephrased at the 1992 FIDE Chess Congress.

Additionally, this caused some effects on chess composition other than Norman's and Thurlow's joke. There exists a chess variant called "Bosma Chess" that was created in 1992 by Dutch problemist R. Bosma that is based on the "loophole." Bosma also noticed the odd wording of FIDE laws, after it was changed in 1992, so he incorporated it into fairy chess.

In Bosma Chess, a king is not in check if it is attacked three or more times. If a king is triple checked, it can't be captured. Here is an example problem that utilizes underpromotion to get the point across.

[Title "Henk Le Grand, 1st Price 151th Thematoernooi Probleemblad 1993, Mate In 2, Bosma Chess"]
[FEN "7K/1NRNkP1R/5p2/b1P2B2/2n5/8/8/3q4 w - - 0 1"]

1. Rc8! null (1... Bd8 2. f8=R#) (1... Qxd7 2. f8=N#) (1... Nd6 2. f8=B#) 2. Re8#

The announcement and award results for the Bosma tourney can be viewed on the Internet Archive: https://archive.org/details/bosma-chess/

They can also be viewed in the Die Schwalbe Chess Problem Database..

Finally, fast forward to 2008, 30 years after Norman's problem was published, a variation of it appeared in an obscure and now dead blog called "Far Off Chess." that was created by IM Jens Kristiansen. In a post entitled "In the twilight zones of chess rules", which is about old and new ambiguities in the rules, Kristiansen created a version of Norman's study from his hazy memory. It is as beautiful as the original and worth showing.

[Title " Jens Kristiansen, Far Off Chess 12/13/2008, White To Move And Win"]
[FEN "7k/rn6/5KP1/p1p4b/8/8/8/7R w - - 0 1"]

1. Kf7 Nd6+ 2. g7+! Kh7 g8=Q+ Kh6 4. Qg7#

This is the known history of the "legal" triple check. Whether or not you would consider it legal, the idea was and still is an amusing nitpick at the exact wording of the FIDE laws of chess.

The problems seen in the answer, and some more, can all be viewed under the keyword "Bosma" on the Die Schwalbe Chess Problem Database. It might also be worth reading this MatPlus forum thread for further discussion on the topic.

In the comments, @bof has discussed the effects of the triple check, which was only lightly discussed in the English Chess Forum. This section will supplement that discussion.

When triple checked, kings become normal pieces that can castle. Under the 1984-1992 laws, castling into, out of, and through check along with having adjacent kings is legal. However, the main interest is about king captures, which are not forbidden either. The answer is simple--if a king is captured, that side cannot be checkmated. This is the only possible conclusion going strictly by the law's letter. This is different from Bosman Chess's immune kings.

Here are problems of mine that show these bizarre conclusions.

Castling Into Check

[Title "Me, chess.stackexcange.com 6/15/2020, Mate In 1"]
[FEN "8/8/8/8/8/7n/5Rrp/2k1K2R w K - 0 1"]

1. O-O#

Castling Out Of Check

[Title "Me, chess.stackexcange.com 6/30/2020, Mate In 2"]
[FEN "4r3/4N3/8/b7/5pp1/3P1Bkp/2n5/2N1K2R w - - 0 1"]

1. Nf5+ Kxf3 2. O-O#

Castling Through Check

[Title "Me, chess.stackexcange.com 8/12/2020, Mate In 2"]
[FEN "8/8/8/5B2/5p1Q/5kn1/2NRb2n/31K2R w K - 0 1"]

1. O-O+ Nhxf1 (1... Bxf1 2. Rf2#) (1... Ngxf1 2. Qf2#) 2. Qg4#

Adjacent Kings

[Title "Me, English Chess Forum 6/13/2020, Mate In 1"]
[FEN "8/8/8/8/8/3p4/3rp3/RK1kN3 w - - 0 1"]

1. Kc2#

Lastly, here are some bonus problems. On page 3 of my English Chess Forum thread, it was asked if a quadruple check could be done, which I did, but I later managed a quintuple check!

[Title "Me, English Chess Forum 4/30/2020, Mate In 4, Dedicated To Marken Foo"]
[FEN "1RnkBB2/2p5/8/4K1Nr/4N3/6R1/1n5q/bQ2r3 w - - 0 1"]

1. Qd3+! Nxd3+ 2. Ne6+ Kxe8 3. Nf6+ Kf7 4. Rg7#

Per another request, I also showed a triple check via en passant.

[Title "Me, English Chess Forum 6/13, Mate In 2"]
[FEN "3r4/3B3N/4n3/3K1k1P/2pp2p1/3r1bP1/2Q1PR2/8 w - - 0 1"]

1. e4+! dxe3+ 2. Bxe6#

Lastly, as suggested by @supercat in a comment, here is a mate in 1 that removes two checks for a mate.

[Title "Me, chess.stackexcange.com 6/13/2020, Mate In 1"]
[FEN "8/7B/3B2NK/8/5k2/6P1/5Q2/8 w - - 0 1"]

1. Ne5#
  • Quadruple or quintuple check is far from the maximum. The only limit to the number of checking pieces in Bosma Chess is the number of White units (8 without promotions, 14 with promotions) e.g. from 8/4R3/6B1/3Q4/1R2N3/3PkPN1/8/4K3 1.Nd6 forces 1...Ke4 (into an octople check, hence not in check) when 2.Re8 is stalemate but 2.d4 Kf4 or 2.Rb3 Kxd3 will mate in a few more moves. Note that 1.Nc3 allows 1...Ke2, and 1.Nc5 allows 1...Kxd3.
    – Evargalo
    Jan 29 at 12:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.