Thanks to <jnpope> and the O'Keefe project I've found another 1862 reference to this game (perhaps the earliest know thus far?), in G.B. Fraser's 2nd article in the <Dundee Courier and Argus>:

http://www.chessarch.com/excavation...

Fraser claims the game was play in St. Helena, about 1817.

Winter quoted from BCM 1900, pages 53-56 and corrected a date for a source as follows: "This will be found reprinted in Waifs and Strays, the second edition of which does not give the date and place of first appearance of the sketch, but a footnote indicates that this must have been the American Chess Monthly, in which case the date cannot have been earlier than 1857 <[July, 1860, in fact]>."

BCM Social Chess (1900) unnamed reviews refers:

"Waifs and Strays" (2e unknown date) which contains a footnote referring to

ACM 1857

But no explicit citation, just a statement about the earliest it could <possibly> have been. I assume the 1857 used is the start of publication? And the parenthetical, [July, 1860, in fact] was introduced by Winter?

Why didn't Winter spell it out, instead of burying the primary source in an obscure parenthetical to an obscure footnote.

BTW - the BCM review contains this:

It's nice to spell it out explicitly.

Thanks for the ACM ref.

https://books.google.de/books?id=eC...

Please see this comment, and the posts it references, for the possible actual source of the game itself, with proper attributions:

Napoleon Bonaparte vs General Bertrand, 1820 (kibitz #108)

Well sure, assuming he actually played this game, he was able to play chess. He knew the rules (though some say he bent them to his will, knowing nobody would dare to protest). But he wasn't great or even good. He was just an average patzer like you and me.

The first moves are okay (which is already quite surprising). <5...Bc5?> is a mistake, but Napoleon plays <6.c3?>, missing the pawn-winning <6.Bxf7+!> (Black can't take <6...Kxf7> because of <7.Qh5+>, winning the bishop and the central pawn).

<9...cxb2??> looks awesome, but it turns out to be a blunder, because after <10.fxe5! bxa1=Q 11.Qd5> (D) Black will have severe problems not to get mated or have his kingside eaten. However, instead, "Napoleon" plays the painfully obvious trap <10.Bxf7+?>

click for larger view

And then we've got <13...a5??>, one of the absolutely worst moves in all times. It is so bad that it makes me doubt the authenticity of the whole game, it makes it look much more like someone found that mate in that position and just needed Black to do *some* move -- and if patzers don't know what to move, they push an edge pawn. But <13...Qxe5> is just painfully obvious and would have been played even by beginners, making the whole mate impossible. Or -- assuming this is authentic -- did Napoleon subtly indicate that he wanted to win now and Black did him that little favour?

The five-move mate itself is really nice, beginning with a rook sacrifice. But again, I have strong doubts that Napoleon actually found and played it. On the whole, this is another classic early game that turns out to be not only probably unauthentic, but also just isn't very good, even if it's me saying so.