I played it over and over and kept finding a win for Black ... so I put it on my Chess Master which won against itself twice. But I am not a master or anything.

In any case Capablanca was in an inferior position and it was worth a try by Thomas - Capablanca played the whole game as if he was half asleep, or tired, ill, or uninterested - very dull chess by him - and I mean we all have such days - and Thomas was I believe pretty good player in those days.

But these endings as are extraordinarily difficult to evaluate OTB - in way that is how we should view it. I mean as if one was playing OTB - not according to long analysis - of course the analysis is interesting.

Please provide your analysis showing a win for Black after 47...Rb6. Note Alfred Brinckmann's tournament book analysis (my post of 11/26/06), where he reviewed the move 51...Rb6, which may transpose with your suggestion of 47...Rb6. If you have an improvement in this line, I would be very interested in seeing it.>

I'll study this later - I didn't see the idea of bringing the Black King down maybe it is a draw. These endings are are very difficult to judge -ultimately it boils down to judgement.

But Thomas was in no danger of losing so should have given it a go with 47 Rb6

I seem to have something promising for Black following Seirawan's analysis 47...b2 48.Rd1 Kb4 49.Rb1 Kb3 50.f4. It is:

50. f4 Ka2 51.Rf1 Rc1 52.Rf2 Rg1

Note: 52.Rf2 Ka1 (or a3) 53.Rxb2+ Kxb2 only leads back to the line by Seirawan (and your improved Kg4, <Pawn and Two>). With Ka2/Rc1 the Ka1/Rxb2/Kxb2 maneuvre can be delayed and the rook can be repositioned on the first row.

First, I tried Re1, but this seems to lead after f5 to the Fritz-main line by <Pawn and Two>, same positions like after <55.Kxh6 Re1 56.g4 Rxe3 57.h4 (.00) (22 ply)> or to 52.Rf2 Re1 53. f5 Rxe3+ Kf4 with the problem for Black how to place the rook. It seems to be even "advantage White".

With 52.Rf2 Rg1 White seems to be in danger, however I played it only against my old Fritz 7 (got it for one Euro) and analyzed loosely with him. Fritz tried 53. e4 and I won it against him in Blitz game settings, so far from being well analyzed, maybe it is just a variant or even a bad idea. I looked at other options and 53. f5 Ka1 54.Rxb2+ Kxb2 does not yet allow f6. 53. forward King-Moves h4/g4 have disadvantages, too, because the g-pawn has to be moved immediately, after ...Kxb2 or it is lost. Comments?

It is interesting to read Thomas's comments about this game. In the book, "Delights of Chess", by Assiac (Heinich Fraenkel), he noted that Thomas had added to the end of the score the following comment: <Draw agreed. A case of funk. Black should win the ending.>

Thomas certainly could produce some fine games, including wins over Botvinnik, Capablanca, Tartakower and Flohr. However, he had some draws against top players where he had a winning position, but could not score the full point.

Consider the games, Alekhine vs G A Thomas, 1926 and Flohr vs G A Thomas, 1933. In both of these games, Thomas had a winning position. Regarding the Alekhine game, Thomas commented; <Interesting game. Black missed a forced win in the endgame, but White should have won earlier.> On his missed win with Flohr, Thomas commented; <Draw agreed, but Black can win by Rg8+.>

If you provide additional analysis with the move 47...Rb6, please indicate the differences in your suggested line and Brinckmann's 2nd variation.

In reviewing the position after 47...Rb6, Fritz found another drawing line. After the moves 47...Rb6 48.Rd1 b2 49.Rb1 Kc4 50.Kh4 Kd3 51.g4 Kc2 52.Rf1 b1(Q) 53.Rxb1 Rxb1 (if 53...Kxb1 .00 17 ply) 54.Kh5 (.00) (18 ply).

Your suggested line is: 47...b2 48.Rd1 Kb4 49.Rb1 Kb3 50.f4 Ka2 51.Rf1 Rc1 52.Rf2 Rg1.

Your latest suggested line is: 47...b2 48.Rd1 Kb4 49.Rb1 Kb3 50.f4 Ka2 51.Rf1 Rc1 52.Rf2 Rh1.

A review by Fritz shows the following: (.00) (20 ply) 52..Rh1 53.f5 Kb3 54.Rxb2+ Kxb2 55.Kh4 Rh2 56.Kh5 Rxg2 57.Kxh6 Rh2 58.f6 Rxh3+ 59.Kg7 and the position is a draw.

Fritz also shows that 52...Rh1 53.Kh4 Ka3 54.Rxb2 Kxb2 55.Kh5 is equal.

Thomas certainly could produce some fine games, including wins over Botvinnik, Capablanca, Tartakower and Flohr. However, he had some draws against top players where he had a winning position, but could not score the full point.>

Thanks for this (and the links to the games) - I analysed it some more and decided Black could win but it wasn't that easy.

I only really guessed (o.k. I did some analysis) it might be a win for Black ...this kind of ending - many in fact require judgement and sometimes it is best to take a draw. The Bard would have said Thomas had decided that "Discretion was the better part of valour"! Funk - no - I have taken draws when I was fairly sure I might be winning - from fatigue or caution - in a tournament it is often a good policy.

Based on Thomas's comment on the score sheet, it appears he later thought that Black should have been able to win. Unfortunately, he did not provide any analysis.

A. Brinckmann and Y. Seirawan have provided analysis of this ending (see my post of 11/26/06). My review of their analysis indicates their suggestions all end in drawn positions.

My other analysis, some of which is posted here, have all started with advantage for Black, with many variations winning for Black, but when the search went deeper, down a very narrow path, all eventually ended up in clearly drawn positions.

If you have any new analysis or suggested lines that could indicate a win, I would be interested in reviewing them.

I did review your suggestion of 47...Rb6, and that also ended in a drawn position (see my post of 10/01/07).

It would seem that white needs a pair of advanced connected pawns to ensure at least a draw in the above. The e and f pawns are the best candidates. Could 53. e4 followed soon by e5 be an improvement over going after the black pawn on h6? Is Fritz overrating the importance of material in the line above?