Both 40...Ne7 and 40...Qxf4 are good enough to win.

Mason had destroyed his own position with 37.cxb4 and 38.Nc2. His position was good enough for a draw if he did not make these moves.

At move 42, Winawer could finish off the game with 42...d4 43.Qxc4 Qg4+ 44.Kh2 Qxf4+ 45.Kg2 Qxe5.

Instead, he gave Mason a short reprieve with 42...Qh1+ 43.Kg3. Had the game been resumed with the correct position, Winawer could have continued with 43...Qc1!, and then if 44.Qf2 c3 or 44.Nc2 Qxb2, Black is winning.

<No doubt the reader will see what neither of the players saw at the time--to wit, that this is what is usually known as a "false" or "impossible" move. It was not discovered to be such until some days after the game had been played and when the official copy of the score came to be examined. It was a powerful stroke, and in every sense of the word a successful one, as it left White absolutely without resource.>

Based on his above remark, it seems that Mr. Mason had quite a sense of humor.

The editor of the tournament book, J.I. Minchin, then added the following note:

<The only explanation that can be afforded regarding this singular impossible move having been made and not detected is that the adjournment had taken place just previously, and the position of the pieces was probably taken down incorrectly on the diagram handed to the member of the Playing Committee, and Black's Knight was placed on d7 instead of e7, thus affording Mr. Winawer the opportunity for his brilliant coup, but for which the game would have probably been drawn.-Editor.>

Mr. Minchin is incorrect in stating that a draw would probably be the result, if the correct adjourned position had been played. As indicated in my previous post, Winawer still had a winning position after 43.Kg3.

Also, as noted by capanegra, the above game score is incomplete. However, it is not just the move 43...Nc5! that is missing, but also the moves 44 to 49.

Mason did not abandon the game after 43...Nc5; instead he played on until he was checkmated.

The final moves of the game were: 43...Nc5! 44.Qg2 Ne4+ 45.Kf3 Qe1 46.Qh2 Nd2 47.Kg2 Qf1+ 48.Kg3 Qf3+ 49.Kh4 Qg4x.

Read Sneaky Pete's post in

E von Feyerfeil vs Lasker, 1889

There is no specific mention of this situation following adjournment. But if the Rapidplay rules for illegal move applied to games played at classical time controls, if the position was incorrectly set up after adjournment and neither player nor the arbiter notices, then presumably the illegal move stands after both players make their next move.

I cannot see there being much of a change in the rule if an abnormality is noticed in a game then the game reverts back to where the abnormality occurred.

* Naturally, computer assistance was not a thing at this point, so diagrams were used, and an error (which is believed to be what happened here) in the diagram would easily go undetected at resumption. (Nowadays, I would imagine adjourned games - if they even occur these days with FIDE's silly new time limits - would easily be double-checked at resume time by feeding the moves themselves back through a computer to reach the correct position ...)

* Adjournments could be long, sometimes days. It would not be unreasonable for such an error to be forgot about by resumption in this case (although it seems likely that the adjournment here was but two hours long, as the schedule for London 1883 was noon to 5 PM, with a two hour dinner break followed by continued play till 11 PM, only after that would the game have gone to a later date).