A question of rulebook pedantry:

  1. One can claim a draw when the same position occurs the third time, the same player having the move.
  2. Two positions are not the same if the earlier allows a capture en passant and the later does not, though men of the same kind stand on the same squares in both positions.

So much I understand. However, what if the pawn that could otherwise have captured en passant happens to be pinned against its own king? In this case, one might say that the pawn had the right to capture en passant but could not exercise it. Alternately, one might insist that a right that cannot be exercised is in fact no right.

I suppose that it comes to how one defines the word right.

The reason this is a question of rulebook pedantry is because it might, conceivably, legally affect whether one could claim a draw if the position recurred. For the purpose of the draw rule, are two positions indeed the same if, in the earlier position, a pawn had the right to capture en passant but could not exercise it because of a pin?

To be clear: My question does not regard how fantastically unlikely such a situation would be ever to arise in actual play, nor if it has ever occurred. My question regards only what the rule would be if it did.

Your answer merits extra appreciation if you can support it with any reasonably authoritative reference!


I know no reason that anyone should be following these updates any longer -- and I rather suspect that no one is reading (if you are not reading, you are probably wise) -- but, for the permanent record and in response to a reader's reasonable request, I have deleted parts of the second and first updates. If the deletions still don't suit, further edits by interested others naturally remain welcome.


@CraigYoung characterizes the first update [as originally written] as a "diatribe." I think that he has a point. A Stackexchange question is not the right forum for the asker (me) to extemporize on the philosophy of rulebook writing. As originally written, above, the question was indeed just a question -- and was not originally meant as anything other than a request for information. What the answers seem to me to have revealed however is that no authoritative answer is possible at the present time. The answers occasioned a discussion, and the discussion seemed worth preserving, whence the first update.

I called the original question "pedantic" to poke a bit of deserved fun at myself, of course. My inner David Hilbert, as it were, wants the rules of chess to be logically complete; but the real, original David Hilbert was a wise man, a deep mathematician but also a shrewd judge of his audience. He could approach such questions with a sense of humor and a light touch, understanding that the matter of logical completeness would not fascinate everyone as much as it did him. Dr. Hilbert saved his earnestness for audiences prepared to appreciate it.

If you will let me don the cloak of earnestness for a moment: Each chess player can judge for himself how much or how little chess loses by failing logically to complete its rules, but surely chess gains nothing by this failure. Chess has a body, the FIDE, whose reasonable, permanent decisions in such matters would command broad respect, except that the FIDE seems mired in endless scandal and has never clearly rendered the decisions in question. So, if you're a Hilbert type like me, you're stuck, aren't you?

There is no rule. There could be. There should be. But there isn't. I admit that I don't like that.


@JamesTomasino's fine answer below is recommended to all interested readers. His logic looks pretty good, his angle is attractive, his judgment seems sound, and he may be right. I do not disagree with him, but I am not yet convinced.

The reason is this. There exists a school of rulebook construction with which I admittedly do not hold (and to which I do not suggest that @JamesTomasino belongs). In that school, a rule is to be enforced even if the rule's writer never considered the point to which the rule is applied. More precisely, the rule's language is to be enforced.

Actually, this might be a tolerable way to construe a rulebook, except that we have ample experience to suggest that, in cases such as this, two, different, reasonable people will infer two, different rules from the same, flawed language. Each reasonably believes that he stands upon "the rock of the law." The rules of chess however are to be a logical construct, not a linguistic one. If we are arguing over what the words mean, and if no consensus exists, then it is the words that are at fault. Such words shirk their duty.

Anderssen v. Kieseritzky may be poetic. The rulebook by which they play however is supposed to be precise. If it is not, then this is a real lack in the rulebook.

Here again is the useful reference @JamesTomasino has kindly provided to the FIDE handbook, section 9.2. After reading it thrice, I can only conclude that the rule is [hard for me to understand.] What are "the possible moves of all the pieces of both players"? Taken literally, "all the pieces" (what ever happened to the traditional phrase "all the men"? but let that pass) of one player have no moves at all, since it is his opponent's turn. Now, of course, obviously, that is not what the rule means, but what it does mean is far from clear.

To muddy the waters further, see the interesting discussion between USCF TDs @Andrew has discovered and brought to our attention.

It seems to me that the answer we are reaching here is that there exists no generally agreed-upon interpretation of the rules in these corner cases. That's too bad. I know what I think the rule should be, but what I think is irrelevant to this discussion, because I am no Staunton or Steinitz and no one cares (or should care) what I think. My chief concern is not whether this rule or that rule might be the better, but only that there in fact be a rule.

Actually, it seems to me that there is a deeper problem. The FIDE Handbook differs from the equivalent publication of the American Contract Bridge League in that the latter draws a clear distinction between the mechanics of the game and guidelines of player conduct. In the FIDE Handbook, the two are sort of all mixed up together.

  • 1
    Great question, I know that a bunch of USCF TD's have discussed some of the intricacies of the rule on the USCF forums. That might be a decent starting point.
    – Andrew
    Commented Jun 9, 2012 at 1:38
  • @Andrew: I have just read through the entire thread to which you have linked. Interesting. I have always cared about game rules, and never have liked the torturing of ugly interpretations from imperfections in the language in which the rules are expressed. The irony is that -- though one or two of the readers in that thread do try torturing rulebook language for a while -- they ultimately, surprisingly give that fruitless discussion up and turn instead to the aesthetical and precedental questions that relate to the rule.
    – thb
    Commented Jun 9, 2012 at 2:00
  • 4
    @thb I was impressed with your question, but after reading your diatribe in the update, I'm left with a bitter taste. If you insist on being pedantic to the point of foolishness, consider this: the player who does not 'have the turn', always has zero possible moves. Ergo that part of the condition always holds true. Commented Jun 23, 2012 at 18:58
  • 1
    @CraigYoung: I think that I agree with you. Moreover, brevity is the soul of wit, saith the bard -- and I've lost it. Somewhere in my two updates is a point I think that I want to preserve, but the updates need radical surgery to be made much shorter, and more importantly to lose their unwelcome character as a diatribe. Help, please.
    – thb
    Commented Jun 24, 2012 at 2:55
  • Regarding the second update, it is good to remember that even Hilbert had to recognize, in the light of Goedel, that completeness is often unattainable.
    – ETD
    Commented Jun 24, 2012 at 3:35

6 Answers 6


The basis of your question comes down to whether a piece is truly in position to execute an en passant if it is otherwised pinned. The answer can be found in the FIDE rules stated just before the explicit mention of the en passant:

"Positions as in (a) and (b) are considered the same, if the same player has the move, pieces of the same kind and colour occupy the same squares, and the possible moves of all the pieces of both players are the same."

The key point is "the possible moves of all the pieces of both players are the same". The en passant passage is mentioned following this to give an explicit example one might overlook.

If in the first occurrence of the position the en passant was not an available move due to another limiting factor, then the list of all available moves would be identical when it appears again.

Thus, draw!

You can find the details in the FIDE handbook, section 9.2.

  • 2
    +1. Your informative answer is well received. My (admittedly tl;dr) comment on it is too long to fit here, so is given in the form of an update to the question above.
    – thb
    Commented Jun 11, 2012 at 15:56
  • Glad you found the answer helpful. If you found it to be the most accurate you expect to find on the question, feel free to accept it. :) Commented Jun 11, 2012 at 21:29
  • 5
    @thb: If your comment is a little too long, you can split it in 2, that happens. If it's way too long, it's probably not well suited to the SE format (if it tends to unrelate to the question, it can go to chat, maybe). Anyway, if the wall of text is not part of the question, then I dont think it belongs to the question, wherever else it may not belong. Commented Dec 11, 2012 at 3:30
  • 1
    Interesting: so two positions P1 and P2 are considered the same even if one player has castling rights in P1 but not in P2, as long as castling immediately is not a legal move (for instance, because one of the intermediate squares in under chess). Very counterintuitive. Commented Jul 26, 2019 at 21:41

The question hinges on whether an "illegal" move is a "possible" move. Strictly speaking, the move is not possible, any more than Bc1-c3 is possible. Both are "illegal" moves.

In the first position, there is no en passant capture that is possible, hence

"... the possible moves of all the pieces of both players are the same."

Until FIDE draws a distinction between these two terms, then, the first position is identical to the later positions, and the draw claim would be valid.

I should add, however, that I am not a FIDE-certified Arbiter. An Arbiter would be the definitive source for an answer to your question.


Following the thread you linked and doing some digging from there, I found a PDF of a 2011 Arbiter's Notebook column which addresses this issue.

Although the player of the white pieces will lose the right to castle in both diagrams on the next move, the rule states that in the actual position the player has not yet lost the right to castle. This means that if same positions appear again on the board, these positions will be considered as different.

His first example position appears to have been butchered at some point in its hosting history, but the second one seems to be intact and makes his point clear; the "right" to castle contributes to the position's identity even though castling is an illegal move. I see no reason to suppose this does not extend to en passant, contrary to the other answers here.

Whether this "illegal right to castle" also covers positions where there is a piece between the rook and the King is unclear to me. I suspect Gijissen would say it does not, but I find it hard to accept that illegal movement through check is fundamentally different from illegal movement through pieces.


You never know when such rules will become relevant. In the past you could seal an illegal move (back when adjournments occurred), but that gave the player a chance to seal such an illegal move and then spend his adjournment time thinking of options. There was a puzzle I saw once which emphasized castling is definitely on a rank by showing how to castle up the board Ke1-e3 and Re8-e1 to give checkmate (notation: O-O-O-O-O-O-O-O).

More real was the problem Viktor Korchnoi had in one of his world championship matches with Anatoly Karpov. He wanted to castle queen-side as Black, but Karpov had a bishop on the h2-b8 diagonal, controlling b8. Could he castle? He couldn't recall the rule and had to ask the arbiter.

  • 2
    This story about Korchnoi strikes me as almost certainly apocryphal; can you cite a source? It seems highly unlikely that any IM, let alone a World Championship Challenger with the experience of Korchnoi, should be ignorant of the castling restrictions. I won't bother enumerating them here, but the move is perfectly legal.
    – jaxter
    Commented Oct 2, 2016 at 0:41
  • @jaxter this is such a well-known story, I'd be surprised if it were a hoax. Just Google for korchnoi karpov castling
    – Glorfindel
    Commented Oct 2, 2016 at 9:20
  • 3
    @Glorfindel Well, I had hoped the responder would follow the answer guidelines and provide a source, but be that as it may, my research reveals that Korchnoi himself states in his autobiography, "Chess is My Life", that he'd never come across the situation in nearly 3,000 games, and wanted to be sure of the rule before making the move. I'll be more careful about asking for sources in the future...
    – jaxter
    Commented Oct 2, 2016 at 17:14
  • @MarkH: I really would like to see that puzzle you mentioned with castling up the board with Ke1-e3 and Re8-e1. Could you show it to me? I remember that once I saw such a puzzle, and it was very nice, and it came with a nice story. But I cannot find it any more.
    – LeibnizGW
    Commented Feb 18, 2018 at 17:21
  • @KnightoftheSquareTable There's one such puzzle on Wikipedia.
    – bcsb1001
    Commented Feb 18, 2018 at 17:35

Thanks for the question, and for the very competent answers. It's not pedantry: it's a genuine corner-case for chess-engine programmers and retrograde analysis enthusiasts. A couple of years ago I asked international arbiter Stewart Reuben (FIDE rulebook committee chairman) who confirmed that the rule works as James Tomasino surmised: i.e. if, because of some pin or check, an en passant capture was never possible, then that position is considered to be the same as later copies.


A pawn that is captured "en passant," starts on the second rank and ends on the fourth rank, having gotten there via the third rank.

The "en passant" rule allows the opponent to stop the pawn on the third rank, where he can be captured. (In war, if one army starts to march, it might provoke a "fire" from the opposing army.)

In a three move repetition rule, the pawn on the fourth rank started there, because it had not been captured en passant. If you have three repetitions that lead to a draw, it's because all pieces started on the same square.

The three move rule specifically applies when no pawn had been moved and no pawn or piece captured. An "en passant" situation is a violation of both these conditions.

  • 2
    (A little late to the party) I think the OP is suggesting that because the pawn is pinned, the en passant right could not be exercised. This means it must still be on the 2nd rank, not the 4th.
    – jaxter
    Commented Oct 2, 2016 at 0:43

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.