In this their second game played a year later, Marshall returned the favor with interest, blowing out von Scheve in just 19 moves.

This win, at least temporarily, put Marshall in third place just ahead of Janowski and only a quarter point behind second-place Pillsbury (who was a over a point behind first-place Maroczy). Marshall's high standing didn't last long, and he was brought back to earth in the next round when he was beaten by Mieses.

This game featured what was best and worst in Marshall's game at this early stage of his career. His early play was hyper-aggressive but unsound, but he was able to turn the game around when von Scheve faltered badly on move 15 and then blundered away the game on his next move.

1. d4 d5
2. c4 e6
3. Nc3 Nf6
4. Nf3 Be7
5. Bf4

click for larger view

5... c6

This move came in for considerable abuse from Gunsberg and Soltis. Soltis claims that the text ignores the "rule of thumb" in the QGD that Bf4 should be met by c5 and Bg5 by c6. But this is all quite unfair both because von Scheve back in 1902 could hardly be expected to know modern chess theory and, more importantly, because 5...c6 is not at all objectively bad. It left the position as:

click for larger view

While White still retains a small opening edge, von Scheve's position was not at all bad.

6. e3 Nbd7
7. Rc1 0-0
8. Bd3 dxc4
9. Bxc4 Nd5
10. Bg3 N7f6

click for larger view

White still retained a small opening edge, and nothing extraordinary had happened yet. But now, things got crazy:

11. e4?!

click for larger view

"?"--(Soltis)

"White deliberately gives up pawn for the sake of a rapid development." (Gunsberg)

Soltis was almost livid in his outrage at this move in his wonderful book on Marshall. But, again, Soltis was here being unfair. The move is almost certainly not entirely sound. But Marshall surely knew that. He gets some compensation and attacking chances for the pawn. I would bet that even if Marshall had gotten into a time machine and read Soltis' book, he would still have played 11. e4. Even later in Marshall's career, he might well have done this. Attacking wizards get away with stuff like this.

As a quick sidebar, I would not that most of Soltis' analysis in this and his other books on great players is wonderful. But this game is an exception. He, like Gunsberg, missed--among other things--the actual losing move.

11... NxN

Accepting the pawn, reasonably enough.

12. bxN Nxe4
13. Bd3 Nf6

click for larger view

Marshall's pawn sacrifice on move 11 was perhaps speculative, now Marshall seemingly off the rails:

14. h4?!

click for larger view

"?!"--(Soltis)

"White threatens a possible sacrifice of his Bishop on h7..." (Gunsberg)

A true coffee-house move. Marshall like King-side attacks, but 14. Ne5 or 14. 0-0 looked more sensible. But one thing can be said for the text; it obviously unnerved von Scheve, who now fell to pieces.

With 14...c5 or maybe 14...b6, von Scheve would be nicely placed and Marshall would have been hard-pressed to justify his super-aggressive play. Gunsberg claimed here that 14...h6 was "absolutely necessary," but this is nonsense. Black is probably still for choice after 14...h6 15. Ne5, but Black had better options. But instead of those better options, von Scheve played:

14... Qa5

click for larger view

Gunsberg notwithstanding, Black is not only not lost but was pretty clearly still better (though not as well placed as he would have been after 14...c5 or 14...b6).

15. Be5

click for larger view

Gunsberg, who was still laboring under the delusion that White was near-winning, said that 15...h6 was the only way to "prolong the game."

More nonsense. While Black would probably still have been better after 15...h6, he would do better than 15...c5 or 15...b6 or maybe 15...Bd7. The big question here is whether von Scheve could just have played 15...Qxa2 (16. Ng5 h6 17. Bb1 Qd5 18. 0-0! My best guess is that Black would still emerge ahead, but I'm not sure I would want to venture this against an attacking genius such as Marshall.

But all this became moot when von Scheve played the much weaker:

15... Ng4?

click for larger view

Gunsberg called this the "losing move," though claiming [mistakenly] that Black was already in trouble.

Soltis also seems to believe the game was now lost for Black since Marshall now had:

16. Bxh7+!

"!"--(Soltis)

As <fredthebear> so nicely put it, Bxh7+ was "begging to be played."

click for larger view

16... KxB??

After this, Black is dead--especially with Marshall at the helm as White:

click for larger view

None of the commentators raised a peep about 16...KxB. But what about 16...Kh8. Suddenly, there is no clear win for White. If 17. Bb1 Black seems far from dead. For what it's worth, Stockfish rates the game as 1.50 and Fritz as 0.59. Black can play 17...f5 and if now 18. 0-0 (probably best since the h-file now doesn't look like a way to win) 18...Qd8 19. g3 b6 with perhaps c5 to follow. Is White then better, for sure. Does he have a win? Here I think Fritz and Stockfish have it about right.

Whatever the conclusion about 16...Kh8, it was clearly the only way for von Scheve to offer any resistance.

After 16...KxB, von Scheve got slaughtered:

17. Ng5+

click for larger view

Black is almost certainly dead here, but von Scheve decided to lose instantly.

17... BxN?

Bringing Marshall's Rook into the attack. 17...Kg8 was--though unlikely to save the game--the only way to prolong the game even a little but.

18. hxN+

click for larger view

Now the only question was how many moves would Marshall require to checkmate the Black King.

18... Nh6

As good or bad as anything here.

This left:

click for larger view

19. Qd3+

And here von Scheve resigned.

But I am shocked that Marshall--of all people--did not find the faster mate beginning with 19. RxN+ (19...gxR 20. Qh5 etc.).