This is a chess puzzle someone showed me a long time ago. If you already know this puzzle please give the others a chance to solve it first. This puzzle was originally created by Otto Gallischek and it was published in Weser-Kurier on 1960-02-25. The problem can be seen here on the website Yet Another Chess Problem Database. It is also featured in this Chess Base article.

There is one small change though-this puzzle has a White queen instead of a rook on h3, as the original problem has.

Above you see a position where white is obviously on the winning side. The queen is one step from giving a checkmate. And the heroic h pawn made his way from h2 to c7 and is waiting for a promotion. However, black saw a last chance and moved Rf3+ in the last turn. The rook can't be taken, because black couldn't move anymore which is a stalemate (= draw). If the king moves away the rook will continue to give checks.

Can you stop the rambling rook and find a way for white to win the game anyway?

PS: Computers are not allowed, and no cheating.

  • 11
    $\begingroup$ Oh my gosh black is playing amazingly well lol! $\endgroup$
    – warspyking
    Aug 31, 2015 at 17:35
  • $\begingroup$ Is this like a 1 or 2 move thing? Or can I list a bunch of moves? Also do we assume black plays perfectly? $\endgroup$
    – warspyking
    Aug 31, 2015 at 17:36
  • 1
    $\begingroup$ @warspyking This is obviously not from a real game ;). And there are "a little" more than 2 moves involved. Don't want to give too many hints just yet. And yes, we assume that black tries to survive as long as possible. $\endgroup$
    – Sleafar
    Aug 31, 2015 at 17:40
  • 3
    $\begingroup$ @mikeTheLiar, I suspect that it indicates whose turn it is. $\endgroup$
    – Kevin
    Aug 31, 2015 at 19:12
  • 2
    $\begingroup$ @Evargalo I've managed to track down the original author. I've added his name in alongside the original source with a couple of links. It turns out that it is a rather famous rambling rook problem $\endgroup$ Aug 17, 2019 at 13:14

2 Answers 2


If white can ever stay out of check for one turn, then it can promote its pawn and put black in checkmate. So in order to play perfectly, black must attempt to put white in check with every move. In turn, white should make sure that black has only one option for a check in the next move, or else the rook will "break free" and have much more influence over white's moves.

White's overall game plan is to sacrifice pawns and maneuver the king until black falls into a simultaneous capture+checkmate.

enter image description here

 1. e3          Pawn moves forward to block rook.
 1. ...   Rxe3+ Rook captures pawn and places king in check.
 2. c3          Pawn moves forward to block rook.
 2. ...   Rxc3+ Rook captures pawn and places king in check.
 3. Ka2         King begins a "serpentine" maneuver, which will continue until move 9.
 3. ...   Ra3+  Rook takes only available move to put king in check, continuing until move 9.
 4. Kb1   Ra1+ 
 5. Kc2   Rc1+ 
 6. Kd3   Rc3+ 
 7. Ke2   Re3+ 
 8. Kf1   Re1+ 
 9. Kg2   Rg1+ 
10. Kf3   Rxg3+ Rook captures pawn in row 3 and places king in check.
11. Ke2         King begins serpentining back to column A.
11. ...   Re3+  Rook continues to pursue.
12. Kd1   Re1+ 
13. Kc2   Rc1+ 
14. Kb3   Rc3+ 
15. Ka2   Ra3+ 
16. Qxa3#       Queen takes rook and places black king in checkmate.
  • $\begingroup$ Your last king move is a2, not a1. Also, it might be nice if you labeled which piece is moving, especially for the chains. $\endgroup$
    – JonTheMon
    Aug 31, 2015 at 18:26
  • $\begingroup$ Oops, fixed. The outcome should be still be the same though. $\endgroup$
    – Kevin
    Aug 31, 2015 at 18:27
  • 1
    $\begingroup$ @LuxxMiner, it's just how I imagined it. Thanks a lot :-) $\endgroup$
    – Kevin
    Aug 31, 2015 at 18:57
  • 6
    $\begingroup$ Here is another animated variant I prepared. $\endgroup$
    – Sleafar
    Aug 31, 2015 at 19:17
  • 1
    $\begingroup$ @Sleafar Thanks, this is way healthier for the eyes! :P $\endgroup$
    – user14478
    Aug 31, 2015 at 19:19

Unfortunately, and I'm just going to plain say it outloud, @Kevin's answer is actually wrong, but only by a tiny bit.

As @Kevin originally, and incorrectly, assumed was that the rook had to be captured with checkmate. While that is generally true for most rambling rook puzzles (In fact, Tim Krabbe himself invented the term "rambling rook!), it turns that that is not the case for this particular rambling rook.

The above statement also applies the the assumption that every Black move must be with check.

However, I do congratulate @Kevin im very clearing explaining the concept of the rambilng rook and how to solve the puzzle.

The only part of his answer that is wrong are the last moves of White and Black. Instead of playing 16.. Ra3+, Black has one more trick up his sleeve:

16... Rxc7!, extendeding their life by one more move, and thus invalidating @Kevin's answer.

Here is the corrected solution, in full.

enter image description here

  1. e3 Rxe3+
  2. c3 Rxc3+
  3. Ka2 Ra3+
  4. Kb1 Ra1+
  5. Kc2 Rc1
  6. Kd3 Rc3+
  7. Ke2 Re3+
  8. Kf1 Re1+
  9. Kg2 Rg1+
  10. Kf3 Rxg3+
  11. Ke2 Re3+
  12. Kd1 Re1+
  13. Kc2 Rc1+
  14. Kb3 Rc3+
  15. Ka2 Rxc7
  16. Qh8+ Rc8
  17. Qxc8#
  • 4
    $\begingroup$ I don't think this invalidates the existing answer. The deviation you propose is trivial, and though the mate is indeed 1 move longer in your variation, I would (subjectively) judge this variation to be subobtimal for black as Ra3 might see Kb1, where Rxc7 practically forces white to play a winning move. $\endgroup$ Aug 6, 2019 at 14:39
  • 2
    $\begingroup$ The question is about winning the game. As I don't see any part asking for the fastest mate, nor the existing answer claiming to deliver it, the existing answer is truely a valid answer to the question. In fact I am now starting to think this whole answer could have been a comment on it ;-) -- To avoid confusion, I did not downvote as it is a nice touch. $\endgroup$ Aug 6, 2019 at 15:29
  • 2
    $\begingroup$ I'll have to disagree. The crux of the answer is what we really should care about here, so on PSE this would be trivial nitpicking. This, for sure, should have been a comment and not a competing answer. As an analogy, suppose PSE had this great mathematical puzzle, in which I was the first to find the brilliant trick to it and thus solve it, but forgot to carry a one right at the end of my answer. Is my answer invalidated by a competing answer that fixes the trivial error? $\endgroup$ Aug 17, 2019 at 14:21

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.