Instead, Marshall played 25...Kd6?, which Fritz evaluates at 19 ply as approximately equal, but a deeper search shows White gaining the advantage.

After 25...Ke8! 26.Kg3 Nd7, Fritz gives an evaluation of (-2.64) (19 ply), and gives the following continuation: 27.Nh7 Rf5 28.Qg8+ Nf8 29.h4 Qe7 30.Ng5 e5 31.Kh2 Be6 32.Nxe6 Qxe6 33.Qxe6+ Nxe6 34.g4 Rf7.

An updated Fritz evaluation now shows (-3.54) (19 ply) 35.f5 Nc5 36.Rc2 Ne4, with Black having a large advantage.

Interestingly, none of the annotators for this game made any note of the correct move 25...Ke8!. Instead, they all praised the incorrect move, 25...Kd6?.

Here are the commentators remarks regarding Black's 25th move, 25...Kd6.

<The defense here must always be aggressive! 25...Nf7? is refuted by 26.Nh7 threatening Qf8+ or Qxf6+.> - Frank Marshall

Another commentator gives 25...Kd6! and then states: <Not 25...Nf7 because 26.Nh7! would then win. The simple part here is that 26.Qxf6 allows a knight fork, which Burn bars with his next move.> - Andy Soltis

<Likewise a fine defensive resource. For the moment White cannot capture either rook or knight. If Black had played 25...Nf7 instead, White would have won by 26.Nh7, sufficient material to win the game.> - Yorkshire Post

All of the commentators were in favor of the move Marshall made, 28..Qf8.

<After 28...Rf8 or ...Rf5; 29.Nxe6, Black's position would rapidly disintegrate. The text is therefore the best chance to break the attack.> - Frank Marshall

<Another fine move in what has become a dicey position. On rook retreats, White would have won with 29.Nxe6 and its multiple threats. Now 29.exf6 Qxf6 30.Qxf6! leads to a pawn race that either player might win. But Burn breaks serve first and Marshall finishs with a fine sacrificial attack.> - Andy Soltis

<The only move to save immediate loss. It will be found that the rook cannot be saved.> - Leopold Hoffer

<Commendable steadiness under most exciting circumstances. To retreat the rook to f8 would involve loss of the e-pawn and immediate disintegration of the position.> - American Chess Bulletin

First I will review 28...Rf8. This move seems to be a reasonable choice and does not lead to an immediate disintegration, as the commentators have suggested.

After 28...Rf8 29.Nxe6 a6, Fritz evaluates the position as (1.15) (20 ply), and gives the following continuation; 30.h3 Re8 31.Qg6 Qe7 32.Ng7+ Kc5 33.Nxe8 Nxe5 34.Qg5 Qxg5+ 35.fxg5.

White's advantage has now increased considerably. Fritz gives an evaluation of (1.71) (18 ply), and the following continuation; 35...Bxh3 36.Nc7 Rc8 37.Nxa6+ bxa6 38.gxh3 d4 39.h4 Nd3.

At this point, Fritz gives an evaluation of (2.39) (15 ply), and gives the following continuation, leading to a winning ending for White; 40.Rc2 Rg8 41.cxd4+ Kxd4 42.Ra1. A possible continuation here would be; 42...Nc5 43.Kg4 and a rapid advance of the king side pawns.

The review by Fritz shows the commentators were correct in rejecting 28...Rf8. However, White's win was not as immediate as they believed. In my next post, I will review Black's other choices, 28...Rxf4 and 28...Qf8.

Marshall and the other commentators made no mention of 28...Rxf4. However, Fritz indicated that 28...Rxf4 was Black's best choice.

After 28.e5, Fritz gave the following evaluation and continuation; (.20) (19 ply), 28...Rxf4 29.Kxf4 Nc5 30.Rc2 Nd3+ 31.Kg4 Bd7 32.Rf1 Re8 33.Rf7 Bc8 34.Nf3 Kb6.

An updated Fritz evaluation now shows; (.77) (18 ply), 35.h4 Qb3 36.Nd4 Qb1 37.Rf8 Rxf8 38.Qxf8 Nxe5+ 39.Kg5 Nc6 40.Qf2 Kc7 41.Nxc6 Kxd6 42.h5.

At this point, Fritz shows White to be winning; (2.14) (16 ply), 42...Qd1 43.g4 Qd3 44.Kh6 e5 45.Rd2 Qe4 46.g5. In several lines from this point, the g-pawn will soon cost Black his bishop.

A deeper search may find some defensive improvements for Black, but at this point I believe that after 28...Rxf4, White will win.

Instead, Marshall played 25...Kd6?, which all the commentators agreed was best. However, after 25...Kd6?, Black's position is near lost.

A critical choice awaited Black at move 28. With the aid of Fritz 11, we now can obtain a better review of this position.

Black had three reasonable choices at move 28; 28...Rxf4, 28...Rf8 or 28...Qf8.

Neither Marshall or any of the other commentators made any mention of 28...Rxf4. However, Fritz indicated that Black has good drawing chances after 28...Rxf4.

Here is one possible continuation: (.49) (23 ply) 28...Rxf4 29.Kxf4 Nc5 30.Rc2 Bd7 31.Rb1 b5 32.Kg4 Nd3 33.Rxb5 Qf8 34.Qxf8 Nxe5+ 35.Kg3 Rxf8.

At this point, Fritz indicated: (.55) (24 ply) 36.Ra5 Kb6 37.Ra1; (.59) (26 ply) 37...Nd3 38.Rca2 Nc5 39.h4 e5 40.Rb1+ Kc6 41.Rxa7; (.59) (23 ply) 41...d4! 42.Nf3 Rg8+ 43.Kh2 Kd5 44.Ra5 Rc8 45.Re1 dxc3 46.Rxe5+ Kd6. Fritz now indicates Black has good drawing chances after either 47.Ra2 Rh8, or 47.Re2 Ba4.

However, Fritz indicated 28...Rf8 would give Black good chances for a draw. My initial review of 28...Rf8 had indicated it would lead to a loss for Black.

Fritz indicated the following: (.95) (23 ply) 28...Rf8 29.Nxe6 b6 30.Rhd1; (.81) (21 ply) 30...Kb7 31.Nxf8 Qxf8 32.Qxf8 Nxf8 33.Rxd5 Ne6 34.h4 Kc7.

At this point, a possible continuation is: (.74) (24 ply) 35.f5 Bb7 36.Rd2 Nc5 37.Rd4 Be4 38.f6 Rg8+ 39.Kh3 Bxg2+ 40.Kh2 Bf3 41.Rg1 Rxg1 42.Kxg1 Bh5.

Fritz indicates White's advantage has now increased, but Black still has drawing chances after: 43.Kf2 a4 44.Ke3 a5 45.Rd2 Ne6 46.Ke4 Bg6+ 47.Kd5 Bf7.

Instead of this line, Fritz indicated Black should play: (.82) (21 ply) 30...Ba6 31.Qg6 Rg8 32.Nd8+ Kc7 33.Qxg8 Rxd8 34.Qxd5 Bb7 35.Qxc4+ Kb8.

Fritz evaluated this position as equal: (.00) (23 ply) 36.Kf2 Rc8 37.Qe2 Nc5 38.Kg1 Rg8 39.g3 Ne6 40.Qe3 Qa2.

However, 28...Qf8 could have lead to such complications that no one could calculated the final result.

After 28...Qf8 29.exf6 Qxf6 30.Qg8?, the commentators noted:

<Here White misses his best chance. He should have exchanged queens, remaining with the exchange to the good, and then advance his h-pawn.> - Yorkshire Post

<Obviously he should have exchanged queens and advanced the passed pawns. The chances are in favour of a successful issue.> - Leopold Hoffer

<He would probably have fared better by exchanging queens and trusting in the race of pawns to save the day.> - American Chess Bulletin

<A let-down after his superb attempt to turn the tables. 30.Qxf6 was indicated, followed by a nerve-wracking race to queen, with a result that would be very difficult to forecast.> Frank Marshall

<Now 29.exf6 Qxf6 30.Qxf6! leads to a pawn race that either player might win.> - Andy Soltis

After 28...Qf8 29.exf6 Qxf6 30.Qxf6 Nxf6, the players would have a Pawn race position where no one could be certain of the outcome:

click for larger view

After 28...Qf8 29.exf6 Qxf6 30.Qxf6 Nxf6 31.h4, Fritz finds the position to be in favor of White, but the complications are enormous.

Fritz indicates the following continuation: (1.66) (26 ply) 31...a5 32. h5 b5 33.h6 b4 34.h7 Nxh7 35.Rxh7 b3.

It looks like we are getting somewhere, but the complications continue.

At move 36, White has several choices. Fritz prefers 36.Re1 or 36.Rh8.

We will look at 36.Re1 first. Fritz indicates (.80) (22 ply) 36...a4! 37.Nxe6 a3 38.Nd4+ Kb6 39.Rh6+ Ka5 40.Rd6 a2 41.Rxd5+ Ka4 42.Rc5 Ba6 43.Nc6 Bb7 44.Nb4.

Black now has 3 reasonable choices; 44...Rg8+, 44...Rc8 and 44...Ka3. Fritz indicates a drawn position is reached after 44...Rg8+ 45.Kf2 Rxg2+ 46.Ke3 b2 47.Nxa2 Rg3+ 48.Kf2 Rg2+. I did not have Fritz do a deep review of the other two moves.

After 36.Rh8, Fritz indicates a drawn position is reached after 36...a4 37.Nxe6 a3 38.Nd4+ Kb6 39.Rh6+ Ka5 40.Ra1 a2; (.31) (26 ply) 41.Rd6 Ka4 42.Rxd5 Bb7 43.Rb5 Be4 44.Rb4+ Ka3 45.Rxc4 Bd5 46.Rb4 Kb2 47.Nxb3 Bxb3 48.Rxa2+ Rxa2; (.00) (30 ply) 49.Re4 Kxc3 50.Re3+ Kc4 51.f5 Kc4 52.Kh3 Kd4.

What an ending! After all the excitement and complications, it seems to me a draw would have been a fair result. Burn missed his opportunity at move 30, but it may have only been good enough for a draw. Fritz has given us a look at what might have happened had Burn played 30.Qxf6!!

However, Fritz indicated White still had the advantage, and should have played: (.51) (22 ply) 31.h4! Nd3 32.Rcf1 Qxc3 33.Rf3; and now: (.96) (20 ply) 33...e5 34.Qe8+ Kc7 35.fxe5 Qxe5+ 36.Qxe5+ Nxe5 37.Rf8; or (1.01) (20 ply) 33...Qf6 34.Rxd3 cxd3 35.Rc1+ Kb5 36.Qh7 Qd4 37.Nf3, with considerable advantage for White in both lines.

Additional analysis is needed, but it may prove that Burn's 30.Qg8 was actually a stronger move than 30.Qxf6.

Instead of 31.h4!, Burn played 31.Qe8+?, after which Fritz indicated the game is equal: (.00) (22 ply) 31.Qe8+ Bd7 32.Qxa8 e5 33.Rhf1 Nd3 34.h3 Nxc1 35.Kh2 Ne2 36.Qxa7 Nxc3 37.Nf3.

After 30.Qg8 Nc5 31.Qe8+ Bd7 32.Qxa8 e5, Marshall stated: <Now I am in my element again. This time it is White's Queen that is out of play, and Black's attack must win out, although he is a Rook and the exchange down!>.

Marshall's evaluation of this position is incorrect. White should play 33.Rhf1!, with an equal position: (.00) (24 ply) 33...exf4+ 34.Kf2 Qxg5 35.Rb1 f3 36.gxf3 Bf5 37.Qe8+, or (.12) (24 ply) 33...Nd3 34.h4 Nxc1 35.Qb8 Ne2+ 36.Kf2 Qxf4+ 37.Kxe2 Qg4+ 38.Nf3.

Marshall then stated: <If 33.Rhf1 exf4+ 34.Kf2 Nd3+ with a winning game.>

Marshall's suggestion is incorrect. If 33.Rhf1! exf4+ 34.Kf2 Nd3+??, the position is winning, but not for Black!

In this line Black must play 34...Qxg5, and the position is equal.

After 33.Rhf1! exf4+ 34.Kf2 Nd3??, White can play 35.Kg1 Nxc1 36.Rxc1 Qxg5 37.Rb1 with a winning position.

Fritz gives this continuation: (1.80) (24 ply) 37...f3 (not 37...Qe5 as the Queens will be exchanged after 38.Qxb7+) 38.Qxb7+ Kd6 39.Qb8+ Ke6 40.Re1+ Kf7 41.g3 f2+ 42.Kxf2 Qd2+ 43.Re2 Qxc3 44.Qxa7 Qf6+ 45.Kg1 Qd6 46.Qd4. Fritz shows this position to be winning for White after: (4.30) (22 ply) 46...Bf5 47.g4.

This was a wonderful fighting game with attack and counter-attack. Marshall noted that it was one of his most exciting games. He said Burn's 20th move 20.Bxh7+, <made it the kind of position where anything can happen.>

This table would have two sides. One side would be rough and was intended for everyday use. This was the side that they would use for preparing food, eating most meals and other domestic chores.

But the other side was much smoother and highly polished. This was the side used when important guests arrived and the hall was used for a banquet.

From this we get the expression - "to turn the tables" on someone, meaning a reversal of fortune. If you were important, the table would be turned polished side up as a mark of respect. But if the Lord considered you to be unimportant, the table would be turned the other way - ie rough side up.

<[pawn and two's]Posts are from lots of different days as well.>

Great Analysis! Not only computer analysis but also historical comments for the game. Please, keep up the good work!

Sounds contrived. I imagined it to be referring to turning a game table around, switching sides with one's opponent.

So, following your post, I did my own internet and book research and found three explanations. Most scholarly sources are similar to yours, but there are also references to double-sided tables and one suggestion (possibly fanciful) that it relates to Roman wives trying to explain to their husbands why they have bought expensive furniture. Not sure I buy that one.

But one intriguing suggestion I found was this. In the Penguin dictionary of English idioms it says "The idiom originates in the one time custome of turning the tables round in the middle of a game of chess or draughts so that the player who had the inferior position could then take advantage of the stronger position of his opponent."

Could this be true? If so, some of our database games might be incorrectly labelled. Instead of "expert vs NN", we might have "expert vs NN/ NN vs expert" in the same game.

But then again I was surprised to learn that white didn't always move first, which has also come out of this forum. So I suppose that nothing much would surprise me about the history of this game of ours.

Great game by both sides.