At a recent tournament the topic of: “is castling legal if more than 10 moves have been played and only then is it noticed that the K&Q were reversed?” was discussed. A few NTDs weighed in on the subject and the opinions were split. Some pointed out that the rules require that the K be in its original position and others argued that for that game d1 or d8 was the original square. None of us could find what we thought was clear unambiguous wording in the rules. It was also pointed out that if a player notices that his opponent’s pieces are incorrectly set, that it would likely be to his advantage to say nothing if castling would be illegal. Comments?
After 10 moves (by both players), you play them as they lie. And the castling rule says nothing about the King starting on the e file. So if the King starts on d1, you castle “long” to the King side by King to f1 and Rook to e1 and “short” to the Queenside with K to b1 and R to c1.
On the one hand, the rules don’t offer any specific restriction on where the pieces have to be for castling. So, I can see the argument for simply allowing castling as defined in Rule 8A2.
On the other hand, if this is taken to a logical extreme, then castling would be legal from virtually any starting position, no matter how far afield it strayed from Rule 3C, so long as the king and rook haven’t moved. I have a hard time believing that this was the intent of the rulebook authors, and I harbor grave reservations as to whether the Rules Committee would see it that way.
If a player started the game with, say, his QN and QR inverted and all other pieces according to Rule 3C, and the irregularity was not discovered until after move 10 was completed, would he be able to then castle queenside? And, if not, why does that not apply to the king as well? What about starting positions where the back rank has more complicated irregularities than what is under discussion here? (Solely based on some of the incorrect piece setups I’ve seen at national scholastic events, I suspect there have been games conducted with a back rank that more resembled a FRC position generated from a DGT 960 clock than the prescription in Rule 3C. I’ve even seen the occasional first-rank pawn.)
In summary, I’d probably disallow castling. If your starting position does not match Rule 3C, and you don’t notice it within the first ten full moves, you have to live with the consequences - which, IMO, include forfeiting the right to castle in the situation Mr. Stenzel describes. However, I think either ruling is certainly defensible…which means the situation could stand clarification, either way.
While true, the illegal position quoted above is also much more likely be discovered before 10 moves is reached. As you can probably guess, the situation was at a scholastic event and rendered moot by subsequent information.
I’m not too concerned about this but I and a few other TDs wouldn’t mind having clearer wording in the rule book.
I’d like to see this ambiguity addressed as well. Ideally, I think it should be addressed in the FIDE Laws of Chess, and the US Chess rules would follow the FIDE rules in this case.
There is nothing in rule 8A2 that requires the king to be on the e-file and the rook to be on either the a- or h-file. Additionally, neither rule 8A3 nor 8A4 state that castling is illegal if the initial position has been set up incorrectly.
I note that blitz rule 3a explicitly states that if the king and queen are swapped, then it is legal to castle short on the queenside and long on the kingside. I also note that this is directly opposite the FIDE rapidplay rule A.4.a.2 that states that “in case of incorrect king placement, castling is not allowed.” To me, this seems to be an unnecessary difference in rules.
If I had to offer an interpretation (and I’m glad I don’t), I’d argue based on blitz rule 3a and the fact that 8A2, 8A3, and 8A4 don’t disallow castling if the initial position is correct, that the player is allowed to castle as long as it complies with 8A2. Do I think this is the way things should be? No.
(I believe there is a proposal to change the FIDE Laws of Chess to require illegal moves to be corrected within ten moves, as the US Chess rules require. I hope that if the laws are so changed, the question of castling with incorrect initial piece placement after ten moves will also be addressed definitively.)
I’d go with 8A and state that a swapped K&Q discovered after 10 moves results in allowing castling short with the a-rook and castling long with the h-rook. For that matter, a swap on the e and f files (K and KB) would result in the f-file king castling long to the d-file and castling short to the h-file. A king on the a, b, g or h files would only be able to castle in one direction (unable to move two squares in the other direction). Before doing such castling there would need to be notation that substantiates that the king really did start somewhere other than the e-file (to prove that the king had not simply been moved there). Players able to keep such notation are usually observant enough to find the discrepancy before ten moves have passed.
FIDE’s rules about going back to the last legal position (regardless of number of moves played since the illegality) renders this moot for them (the game would end up being restarted).
This entire discussion is rendered more complicated by the possibility that FIDE and USCF may differ, and/or by the possibility that FIDE and/or USCF may have changed in the last two years or so, and/or by the ridiculosity of the whole situation to begin with.
Bill Smythe
Under FIDE’s Laws of Chess (7.2) “If during a game it is found that the initial position of the pieces was incorrect, the game shall be cancelled and a new game shall be played.” As previously noted, under US Chess rules (11F), “If the error is discovered after the completion of black’s 10th move, the game shall continue.”
Yes, this is true of the current Laws of Chess, but it may change in the next revision cycle.
At a recent scholastic tournament, we had a young man castle on the 6th rank. The king and rook were on the correct files so why not try it? The best part was he didn’t believe the original tournament director when she told him that wasn’t legal and wanted to talk to a different TD.
How often does this happen at adult tournaments? Changing the adult rules is one thing; however, the tail that wags the dog has their own scholastic rules variations (as does CCA). So, how would changing the rules to conform to the “cookbook”–the rules have to address every situation–address the adult/scholastic rules split?
BTW, I really don’t have an opinion either way on what the rule should be. I just wonder if changing the adult rule is a practical method of addressing what may be a scholastic concern.
While I am sympathetic to the observation that the rule book sometimes tries much too hard to cover every possible situation, I think it is quite reasonable to expect the rules to spell out whether a move is legal or not. I don’t think it will take a ridiculous number of words to correct the situation. There is a difference between “unambiguous” and “comprehensive.” The US Chess Official Rules of Chess should not strive to be comprehensive, but they absolutely should be unambiguous.
Harold beat me to the punch by posting this. I was the floor chief and was called over to make a ruling. It is interesting that the NTDs on this forum are also somewhat divided.
As pointed out here (and as was discussed at the tournament), each proposed solution has interesting consequencs.
A ruling that one cannot castle banishes the king to the center forever. Such a ruling could also encourage a somewhat unscrupulous player to deliberately not correct the incorrect starting position knowing that after move 10, they are going to get one heck of an attack!(and all he has to do is say I didn’t notice!)
On the other hand, allowing castling, could lead to some strange happenings. A K at b1 castles by going to d1 (and the rook goes all the way from h1 to c1!) and cannot castle on the other side.
Note that 8a says “original square.” Does that mean e1 and e8 or does that mean wherever the player put it.
A few other comments:
When I was called to the board, both players agreed that the W king and Q were reversed, but the B K + Q were not. Visual inspection seemed to confirm this. Black was already up a Q.
As a last ditch effort, I decided to look at the ply count and scoresheet. Initially I did not. Normally scholastic players who set up K + Q wrong do not have good scoresheets! 
When I looked at the scoresheet and plycount, I determined the K + Q WERE set up correctly. At some point, W played the Q to e3. Black played Nxc2+ Kd1 and Nxe3. I pointed this out to the players and both said “Oh yeah, you’e right.”
At this point I ruled no castle, K must move, and 2 minutes. I also ruled that I wanted to go to the bar, but as the round was not yet over, I could not enforce that ruling!
Dave Hater
I have no idea why 11F isn’t sufficient to rule on this.
The absence of detailed, written, instructions (in 3C) of how the initial position is set shouldn’t require anyone to become confused; there is a diagram showing the correct initial position, and any deviation from it is covered in 11F.
I think you’re all overthinking it. After 10 moves, castling would be allowed, following rule 8A2; previous to 10-moves, the game should be annulled and a new game played.
Am I over-simplifying it?
As I said earlier, “Before doing such castling there would need to be notation that substantiates that the king really did start somewhere other than the e-file (to prove that the king had not simply been moved there). Players able to keep such notation are usually observant enough to find the discrepancy before ten moves have passed.”
Without a scoresheet there would have been uncertainty about where the king was set up, and such uncertainty would mean that castling short to the queenside would not have been valid. As David’s example shows, players able to keep a scoresheet well enough to show where the king really started are generally also able to not start a game with the king and queen swapped.
I suspected this had arisen from a scholastic event. It’s a corner case, to be sure. I, too, am not terribly concerned about this.
However, I do remember seeing one game at a National Elementary where there was a rook on a2 and a pawn on a1. As it happened, just as I was approaching the board, Black fell into an extended Scholar’s Mate (1 e4 e5 2 Bc4 d6 3 Qh5 g6 4 Qf3 Nc6 5 Qf7#), which relieved me of any duties beyond collecting the result slip. 
The game ends with a checkmate coming from a legal move. Fortunately, nobody asked whether or not the Qxf7 was illegal because the position was illegal. If you verified the result (normal for a national) by asking Black who won, with the response that White did, then it is also a resignation in the event the Qxf7 was deemed illegal.
I don’t know why that would have been unfortunate. Had one of the players asked this question prior to verifying the result, I would have had them play a new game, since it would have been discovered during the first ten moves.
My verification involves confirming that the slip is properly completed, including having both players agree the correct result is circled. Generally, I don’t even look at the board. The odd position of the a-pawn drew my eye, though.
When at a scholastic tournament I verify the result by having both players confirm it I always look at the board to see if the position is actually as claimed by both players. At our recently completed MLK Holiday Scholastic there were three instances of which I am aware in which both players agreed that the position was a checkmate when in fact it was not. In those cases I make a note of who the losing player was and speak to his coach about it later.
I will just say there are very good practical reasons for not looking at the board when verifying a result at a national scholastic.