# Get 6 pawns in file A or H

This is somehow inspired by my other chess related question: Possible pawn combinations

How can you get 6 pawns (white or black) on a chessboard on the files A or H using legal moves?
Start with a normal chess board, with the standard configuration of pieces.
If you choose white, provide the list of moves to reach a position where 6 white pawns are on file A or H rows 2 to 7.
Regular chess moves apply.
- if the king is checked it needs to get out of check
- If mate "happens" before reaching the desired position you failed.

Here is a possible layout of the desired position (I chose white):
Since there are 15 captures required, black will end up with just the king. I removed also all the white pieces and 2 extra pawns, but they can still be present on the board.
Also the kings can end up in any position.

As of Sid's recommendation Make it in the lowest number of moves you can.

• Do we have to do it in the shortest no. of moves possible or something like that?
– Sid
Commented Nov 3, 2017 at 10:43
• @Sid. It didn't cross my mind. I was just curious about a solution. But now that you mentioned it.... why not Commented Nov 3, 2017 at 11:48

## 4 Answers

A solution in 31 moves.

https://lichess.org/UTEzGBSi#62
1. Nf3 Nf6 2. Nd4 Nd5 3. Nc6 dxc6 4. g4 Bf5 5. gxf5 h5 6. f4 Nd7 7. e4 Ne5 8. fxe5 Qd6 9. exd6 exd6 10. exd5 g6 11. Na3 Bg7 12. Rb1 Bd4 13. h4 Bc5 14. d4 g5 15. hxg5 Rh6 16. dxc5 dxc5 17. Bd2 Re6+ 18. fxe6 fxe6 19. Ba6 exd5 20. c4 dxc4 21. Ba5 Ke7 22. b4 cxb4 23. g6 bxa3 24. Rb6 bxa6 25. Rxh5 cxb6 26. Rb5 bxa5 27. Qb3 cxb3 28. g7 bxa2 29. g8=Q cxb5 30. Qg4 Ke8 31. Qa4 bxa4
I improved this from 32 moves by having the black player move pawns instead, saving a half move.

A proof that this is optimal.

I'll assume black is the player arranging pawns and white is helping. White needs to get their 15 non-king pieces onto each of the 15 "death squares" for black's pawns to capture, illustrated here with opposite colors. We can count the number of moves for each white piece to get to any such space: 0,1,1,2,2,3,3,4 for pawns, and 1 for each non-pawn piece except the far knight (3) and far rook (2). That forces 26 white moves.

But, some of these are incompatible. The close rook is blocked by the pawn in front that wants to never move, so one of them must take an extra move. The close bishop and knight both share a death square one move away (a3 or h3). Each issue costs a move to resolve. Finally, at least one promotion is required, so some pawn must take at least 5 moves to promote, then one to reach a death square, replacing 4 moves at most with 6, for 2 extra. That's 4 extra white moves, for 30 total.

My solution has the promoting pawn make an extra post-promotion move for 31 total. Can it be cut? The answer is no, and this will take some careful analysis.

The pawn would need to move to its death square right after promoting. Where can it end? The death squares c3, c4, d5, d6, e6 are reserved for the white c through g pawns that cannot reach a different death square without an extra move. So, the promoted pawn must go to a death square on the a-file or b-file or c6. It can't do so by making a diagonal move along the line the black pawn is moving without capturing it. This means it must promote on a8, b8, c8, or d8. The easiest seems to be d8, to promote to knight and jump to c6.

The promoting pawn must start on the h file, the only one that would otherwise take 4 moves to reach a death square. This means it must capture 4+ times to promote to d8 (or more to a8, b8, or c8). White already must capture times with their d,e,f,g pawn in place to get into place. So, white must capture 10+ black pieces. But, black must have their 6 pawns survive, and their king, leaving only 9 black pieces to be captured. So, saving the 31st move is not possible.

• Wow!! This is brilliant!!! Commented Nov 4, 2017 at 16:22
• this is more than I expected Commented Nov 6, 2017 at 8:47
• A tiny detail at the end of the explanation: whatever the number of captures, White's h-pawn can never reach a8 and promote there. Commented Oct 19, 2018 at 10:45

Here's one really ugly solution.

The length is 57 moves. But, I think it can be made tidier.
1. Nc3 a5 2. Nb1 a4 3. Nc3 a3 4. bxa3 b5 5. Nb1 d5 6. Nf3 Bd7 7. Ng1 b4 8. Nc3 Ba4 9. Nb1 b3 10. cxb3 c5 11. bxa4 Qa5 12. Nf3 c4 13. Nc3 e5 14. Nb1 Bb4 15. Ng1 c3 16. dxc3 d4 17. cxb4 d3 18. bxa5 Na6 19. exd3 Rc8 20. Nf3 Ke7 21. Be2 Nf6 22. Ng5 Ne8 23. Nh3 Nec7 24. Ng1 Nb5 25. Nh3 Rc4 26. dxc4 Rc8 27. cxb5 Ra8 28. bxa6 Ra7 29. O-O f5 30. Qd2 e4 31. Bh5 e3 32. Bf3 exd2 33. Be4 fxe4 34. Nf4 d1=Q 35. Nc3 Qd4 36. Bd2 e3 37. Nh3 e2 38. Nf4 e1=R 39. Nh3 Re3 40. fxe3 g5 41. exd4 h5 42. Nf4 gxf4 43. Be3 fxe3 44. Rf4 e2 45. Rg4 hxg4 46. Nd1 e1=R+ 47. Kf2 Re6 48. Ne3 g3+ 49. Kf3 gxh2 50. Kf2 h1=R 51. g3 Rb1 52. g4 Rc6 53. g5 Rbb6 54. g6 Rc5 55. dxc5 Kd7 56. cxb6 Kc8 57. bxa7

Interactive on Lichess

Here's a picture of the final position:

• Nice. I added a link to Lichess with your sequence to make it easier to view and go through it. Commented Nov 3, 2017 at 12:08
• really nice work. I will try to find a better solution. Commented Nov 3, 2017 at 15:25

Here is a solution in 51 moves

1. Nf3 h5
2. Ne5 h4
3. Ng4 h3
4. gxh3 Rh4
5. Nf6+ Nxf6
6. a4 g5
7. Nc3 Ne4
8. Nd5 Ng3
9. fxg3 Bh6
10. gxh4 e5
11. Ne7 f5
12. Bg2 f4
13. b4 e4
14. Bb2 e3
15. dxe3 Qxe7
16. exf4 d5
17. fxg5 d4
18. gxh6 d3
19. cxd3 Bg4
20. a5 Qf7
21. a6 Qh5
22. Rf1 Nxa6
23. Bd5 c5
24. Rf3 Nxb4
25. Qc2 Nxc2+
26. Kd2 Nd4
27. Ra4 Nxf3+
28. exf3 b5
29. fxg4 bxa4
30. gxh5 Rd8
31. Bb3 axb3
32. Bd4 a5
33. Bc3 a4
34. Bd4 Rxd4
35. Ke3 Re4+
36. dxe4 a3
37. Kf2 a2
38. Ke2 a1=Q
39. Ke3 Qe5
40. Kd3 Qf5
41. exf5 b2
42. Ke3 b1=Q
43. Kf3 Qg1
44. Kf4 Qg6
45. fxg6 c4
46. Kf5 c3
47. Ke6 c2
48. Kd6 c1=Q
49. Kd5 Qc7
50. Kd4 Qh7
51. gxh7

https://lichess.org/TOve2wLp

Position at the endgame!

[]

I tried optimizing it down under 50, excruciatingly hard!!!

The reason I went for

h-file instead of a-file is simply because I couldn't find a pattern under 54 moves.

• @Marius that's great. Also, no hints, please :) Commented Nov 3, 2017 at 16:24
• Ok. Ok. I will delete the comment :) Pretend you didn't see it. Commented Nov 3, 2017 at 16:25
• Did you try XHAVINGXTHEXBLACKXCXPAWNXCAPTUREXWHITEXPIECESXONXDXANDXEX, XSOXASXTOXAVOIDXHAVINGXTOXGETXITXALLXTHEXWAYXTOXEIGTHXRANKX? Commented Nov 3, 2017 at 23:21

Here, "kiloNewton" shows how White can achieve this at move 52.

The first 51 moves are:

1. Nf3 a5 2. Ng1 a4 3. Nf3 a3 4. bxa3 b5 5. Ng1 b4 6. Nf3 b3 7. cxb3 Ra4 8. bxa4 c5 9. Ng1 c4 10. Nf3 c3 11. dxc3 e5 12. Ng1 Bb4 13. cxb4 Qa5 14. bxa5 e4 15. Nf3 e3 16. fxe3 d5 17. Ng1 d4 18. exd4 Nd7 19. Nf3 Nc5 20. dxc5 Ba6 21. Ng1 Bd3 22. exd3 f5 23. Nf3 g5 24. Ng1 h5 25. Nf3 Rh6 26. Ng1 Rb6 27. cxb6 Ne7 28. Nf3 Nc6 29. Ng1 Na7 30. bxa7 f4 31. Nc3 f3 32. Be2 fxe2 33. Nf3 exd1=N 34. Ne5 Nb2 35. Nf3 Nc4 36. dxc4 g4 37. Ne2 gxf3 38. Be3 fxe2 39. Rd1 exd1=N 40. Bf2 Nc3 41. Be3 Nb5 42. cxb5 h4 43. Bd4 h3 44. Be5 hxg2 45. Bf4 g1=N 46. Be3 Ne2 47. Bf4 Nd4 48. Bd2 Nc6 49. Bf4 Nb8 50. Be5 Na6 51. bxa6 Kd7

(As Marius pointed out, White's 52nd in the linked web page is not needed for this puzzle.)

• It's actually 51 because the last move is just white castling. But maybe it can be even lower than that because in my question there is no need for white to keep the king side rook for castling. Commented Nov 3, 2017 at 12:26
• Hey, that link is not opening for me. Commented Nov 3, 2017 at 15:27
• It works for me, but @ABcDexter makes a good point -- it is best to summarize the content here in your own words in addition to providing a link. Commented Nov 3, 2017 at 16:09
• Even if this is in fewer moves as Sid's answer, I will not consider it because it feels like cheating. +1 for the good find, but that's about it. Commented Nov 3, 2017 at 16:23
• That method could also be done a few moves quicker. If the last promotion is a rook instead of a knight it could get into the position to be taken in two moves instead of five so the pawn would take (and get into position) at move 48. The reason it isn't in the linked game is again the castling requirement. A rook would have prevented the castle. Commented Nov 3, 2017 at 16:34