# [Retros] [tros] chess puzzle

Noam Elkies elkies at math.harvard.edu
Sat Mar 15 00:21:57 EDT 2008

"Frank Tuyl" <tuyl at tpg.com.au> writes:

> I happened to come across a chess puzzle, published in a Dutch newspaper

> a couple of years ago, and after many hours I have not been able to crack it.

> After some googling I found your site and was just wondering whether this

> would be a way to get the solution from someone. (white to play and win)

> White: Ka6, Qh2, Nb4, a7

> Black: Ka8, Qg6, Nf8, d6

Doesn't seem to be a retroanalytic puzzle, which is what this mailing list
is for. Anyway, material is even, Black's King has no moves, and the board
is wide open but for now both Queens are out of play, with neither side
having a safe check except for Black's threat to free his King with ...d5+.
This will also answer the otherwise desirable check 1 Qh1+. So White
would like to play Nd5, blocking that check and threatening mate in one from
either b6 or c7 -- but 1 Nd5 lets Black play another check, 1...Qd3+, and then
after 2 Kb6 Qxd5 only Black can win. But if Black didn't have this check
then Nd5 would win outright, even without using White's Queen. So it's
worth the Queen to make it possible...

1 Qh6!! Qxh6 2 Nd5 and 3 Nb6# or 3 Nc7#

Black can defend either b6 or c7 with either Queen or Knight (Qe3 or g7/h7,
Nd7 or e6) but can't stop both mates. Black doesn't have to play Qxh6, but
White is also threatening 2 Qxf8+ and mate, and Black can't hold everything:
1...N-any 2 Qxg6; 1...d5+ 2 Qxg6 Nxg6 3 Nd5 and mate; 1...Qf5(f7,g8/e8?)
2 Qh1+ (this is good now that the forced reply ...d5 is no longer
a cross-check) d5 3 Qxd5+ Qxd5 4 Nxd5 and again mate. If 1...Qe8? then
2 Qh1+ mates on the diagonal, and if 1...qg8 then 2 Nd5 is a faster win.

Any idea who is the composer and when/where this study was first published?

--Noam D. Elkies