Seeking a particular problem

Question

I recall seeing a problem in Games magazine, probably in the early 1990s, that required the solver to deduce that a capture en passant was possible (not just because that's the only way to get a solution, but using retrograde analysis). I recall that all the pieces (including pawns) on the board were congregated in one corner — if I recall correctly they were all in the first three files and the last four ranks. The solver was required to mate in a small number of moves, possibly two. Does anyone know what problem it is?

Answer 1

I found a problem that perfectly fits your description! I’ve played the position back a move for the double step.

[Title "Kurt Keller, Stuttgarter Nachrichten & Südwest Presse (Ulm) 1993, Mate In Three"]
[FEN "5RbN/5ppk/5P1P/5P1K/8/8/8/8 b - - 0 1"]
[startply "1"]
[startflipped ""]

1... g5 {Now White mates in three.} 2. fxg6+! fxg6+ (2... Kxh8 3. gxf7 Kh7 4. fxg8=Q#) 3. Nxg6 Bf7 4. Rh8# 

If the Games magazine flipped the position, then this is a problem with all pieces in one corner, on the first three files and last four ranks, a short number of moves, and an en passant key! Even if it wasn’t flipped, your mind might have altered your memory. Since the problem itself is from the early 1990s, it’s safe to assume that the Games magazine reprinted it for show.

Addendum 7/2/2022: I came across a mate in two by Pal Benko that also fits the fill perfectly. (Once again, I've played it a move back so the replayer works properly.)

[Title "Pal Benko, 1996?, Mate In Two"]
[FEN "kbN5/npp5/B1N5/P1K5/8/8/8/8 b - - 0 1"]
[startply "1"]
[startflipped ""]

1... b5 {Now White mates in two.} 2. axb6! cxb6+ 3. Nxb6# (2... Nxc6 3. b7#)

While the YACPDB database actually lists the source as "Games", the Die Schwalbe Chess Problem Datase sheds series doubt that on that publication date/source.

To quote it, translated from German:" In 'My Life' [,by Paul Benko,] the original source is "Games 1996" - it is unclear which publication is meant. The name Benko does not appear in the index files of the 'Games and Puzzle Journal' [Jelliss], according to 'Chess Periodicals' [Felice] there were some magazines that could fit the name (see p.90ff, entries 1016ff), but their publication period ends well before 1990."

The Games and Puzzles Journal archive is here: http://www.mayhematics.com/p/p.htm I found it not in any 1996 issue. Either it is in another issue, with a wrong date given, Benko forget the source to his own problem, another unknown source named "Games" is meant, or one of many strange possibilities is the answer to this problem's origins.

Answer 2

If it's from a source like Games Magazine, which does not have a regular chess column, then I guess the problem was not a first publication but a reprint of this Amelung classic, which fits the description almost perfectly up to reflection (and perhaps the reprint was reflected):

[Title "Friedrich Amelung, Düna-Zeitung 1897, Mate In Two (after ...g5)"]
[FEN "7R/6pp/5P1k/4PK1P/8/8/8/8 b - - 0 1"]
[startply "1"]
[startflipped ""]

1... g5 2. hxg6! Kh5 3. Rxh7#

There is no other possible Black move to reach the position after g5: squares f6 and h6 are blocked, so the pawn didn't reach g5 by capturing; bPh7 is on its starting square; Kg6-h6 would have WK and BK attacking each other; and Kg7-h6 would require White's previous move to be with the checking f6-pawn, which is impossible because e5,f5,g5 are all blocked (the first of these is the only reason the e5-pawn is needed in the diagram).

As it happens, in this Amelung problem (unlike the more recent Keller and Benkő problems suggested by Rewan Demontay) Black's g5 is not a blunder: any other move would also allow the same Rxh7 mate in the next move or two.