Chess Archaeology HomeChess is a scientific game and its literature ought to be placed on the basis of the strictest truthfulness, which is the foundation of all scientific research.W. Steinitz

5-4 Score — Sixth Match, 1834 La Bourdonnais-Macdonnell Matches
Researched by Nick Pope

(Return to Main Index)

The earliest published account of the totals for these matches appears in Bell's Life in London:

On the occasion of La Bourdonnais' recent visit to London, he played 88 games, in the Westminster Chess Club, with Mr. M—, the first English player. Of these, 14 were drawn; and of the remaining number, La Bourdonnais won 44, and lost 30. A selection of 50 of these games was made, and printed by Mr. Lewis; but it is matter of great regret to the Chess World that he did not publish the whole of the 88.

The earliest known statement about the match is dated November, 1834, and comes from the Report of the Westminster Chess Club which concurs with Bell's Life in London, as they should, as George Walker supposedly authored both accounts.

To The Editor Of "The Chess Player's Chronicle."

Hereford, May 21, 1844.
Sir,—[...]

I take the same opportunity of informing you that I have in my possession a document which may be of much interest to the Chess world. It is a printed Report of the Westminster Chess Club, of the date November, 1834, the very year in which the great matches between those august champions, De la Bourdonnais and M'Donnell, were played. This gives a statement of the final result different from any that I have observed in the "Chronicle." I transcribe it to the letter:—

"Eighty-eight games were played by these gentlemen; of which number fourteen were drawn. Of the remaining seventy-four, M. de la Bourdonnais won forty-four and lost thirty."

This statement corresponds with my own recollection exactly. I was a member of the Club at the time, and for a considerable period afterwards; and I never heard its accuracy impugned. I beg to remind you, that the death of the lamented Mr. M'Donnell did not take place till the September of the following year, 1835; and if the report had been incorrect, is it not highly probably that he would himself have set it right? As I think it very likely that you may be able to obtain a copy of the same report from one of your metropolitan friends, I am unwilling to commit so precious a record to the hazard of the post. I will merely add, that I perfectly recollect that Mr. M'Donnell was much ahead in the last games played, but something was urged in excuse of his great opponent's defeat on the ground of domestic anxieties.

I trust that on such a subject you will not deem these few remarks troublesome or intrusive, and I remain,
Mr. Editor, your obedient Servant,    
AN AMATEUR.

This total of 88 games is maintained by George Walker until at least July, 1838:

The total number of games played by La Bourdonnais and M'Donnell was 88. Of these Mr. Lewis printed fifty, and we believe we gave the whole of the remainder before they were re-edited in an improved form, and printed with the numerous other games of Mr. M'Donnell, by Mr. William Walker.

It is also worth pointing out that George Walker published repeatedly that McDonnell had won eight of the games played in the last match. Presumably eight of the last eleven decisive games or eight of the last twelve games (including a draw).

In the beautiful series of games played last year in the Westminster Chess Club, between the great La Bourdonnais and the first player in England (need I name Mr. M'D******?), this opening appeared to be a decided favorite with the former; who, as I observed, at this point, uniformly played K.B.P. one sq. As I have alluded to this interesting trial of skill, I think it right that a correct statement of the result should be recorded. The contending champions played, in all, 88 games; of which number 14 were drawn. Of the remaining 74, M. De La Bourdonnais won 44 and lost 30. I cheerfully admit La Bourdonnais to be the stronger player, but cannot believe the above proportion forms a correct inference as to their relative degrees of skill. Mr. M., in the beginning, was naturally diffident of his powers, on finding himself opposed to him whose reputation has justly spread throughout Europe as "the greatest living player," and his own fine play consequently appeared to much disadvantage. Of the first 21 won games, Mr. M. lost 16; but gaining confidence as he proceeded, his play improved, and of the last 12 games of the 84 [sic], he won, I believe 8. On the next occasion of La B.'s visiting England, the contest will be certainly renewed. Mr. Lewis has published a selection of 50 of the above games, which form a volume no Chess-player should be without; but, in common with most other members of the Westminster Chess Club, I feel both regret and disappointment that the remaining games were omitted. Let us hope that the first impression of Mr. Lewis's book may be very shortly exhausted; and that, on his then going to press for a second edition, he will favor us by printing the whole of the 88, in the exact order they were played.

Chess.—We cannot tell our Correspondent where to get the French poem he alludes to, but believe it may be procured through any foreign bookseller. The author is the celebrated Mery, the title "Le Revanche de Waterloo," and the subject, we are told, for we have not yet received it, is the victory of La Bourdonnais over M'Donnell. We hope that the author is honest enough to sing how COMPLACENTLY the French hero went through the first twenty-one games, of which he won no less than sixteen, through Mr. M.'s nervousness: how SMILINGLY he pursued the match through the next stage in which, though M.'s play was fast improving, the Frenchman kept his advantage; lastly, how FROWNINGLY Monsieur performed the third act of the drama (the said third act being equal in length to the other two, which comprised forty-two games, of which our lamented countryman won twenty-two; and last of the lastly, by way of epilogue, we hope M. Mery laments the fact, in suitable rhymes, that M. La Bourdonnais, having lost eight games out of the last eleven they played, was unfortunately compelled to business to depart for Paris at an hour's notice, leaving the match unfinished. We think it probable that, unless M. Mery has done justice to both sides, a friend of our's will translate his poem, cum notis variorum, which the truth will be told in round terms.

The fact was, the parties played always on terms of strict equality (and that, although of the 88 games thus played in all, M. La B. won at first in a large proportion, yet of the last 42 our lamented countryman won 22, and of the last eleven won eight.

Using Chess Studies, which appears to give a correct accounting of the first five matches, we arrive at the following subtotal of 41 wins for Bourdonnais, 22 for McDonnell and 13 draws. This leaves us twelve games that should exist for the sixth match, i.e. three wins for Bourdonnais, eight wins for McDonnell and one draw.

 Match  LCM wins  AM wins  Draws 
11654
2450
3651
4837
5741
Subtotal412213
"6"381
 Total 443014

This set of totals lines up perfectly with the prior statements of McDonnell winning eight of the last eleven (not counting the draw), and eight of the last twelve (counting the draw).

However, when one factors in the games from the sixth match as given in Chess Studies to the above subtotal total we end up with:

 Match  LCM wins  AM wins  Draws 
11654
2450
3651
4837
5741
6450
 Total 452713

Chess Studies gives 45 wins to Bourdonnais instead of 44. As matches one through five appear to be correctly given in Chess Studies let us look to the sixth match, which as published by Chess Studies gives four wins and not three to Bourdonnais.

If there is a mistake it would appear to be that one of these four wins has been erroneously assigned to Bourdonnais. The four wins in question are games 78, 79, 80, and 82.

First off we will rule out the Queen's Gambit (game 78) as being incorrectly assigned for a couple of reasons. The first being that McDonnell always accepted the gambit pawn when offered and the one time McDonnell played the attack Bourdonnais declined the pawn (game 38). The second reason being that this is one of two games from the sixth match which were also published by Lewis in his book. Lewis also gives it as Bourdonnais-McDonnell making it less likely to be a game with the players being incorrectly assigned.

This leaves us with three Evans' Gambits, games 79, 80 and 82, as candidates for a color reversal. In checking prior research into this topic, Murray noticed that game 80 has "internal evidence" of being wrongly assigned.

The exact details of the matches are not known,[13] but De la Bourdonnais won a considerable majority of the games.
[...]

[13] The authorities are: (a) Greenwood Walker's edition; (b) Report of Westminster Club, 1834 (by Geo. Walker); (c) Geo. Walker's Chess Studies; (d) Geo. Walker in CPC., iv, 369; (e) Reprint of games in CPC., ii and iii (? supplied by Lewis); (f) Palamede, 1836, 26 (De la Bourdonnais); (g) Palamede, 1844, 266 (Saint-Amant says De la Bourdonnais told him that he allowed MacDonnell some games in the last match). Cf. CPM., 1864, 72, 115, 161 (Geo Walker), 161, 203, 232.

Five matches of 21, 9, 11, 11, 11 games (excluding draws) were played out, and part of a sixth. All agree as to the score of the first four matches (I, B. 16, M. 5, Drawn 4; II, 4, 5, 0; III, 6, 5, 1; IV, 8, 3, 7. According to (d) the other matches resulted: V, 7, 4, 1; VI, 5, 4, 0. The score of the existing games of VI is, however, 4, 5, 0.

Greenwood Walker says that he took down all the games as played, and gives the score of 83, and mentions one (No. 14) as omitted because it was badly played. His total score is 41, 29, 13. The other editions of the games add No. 85 (won by M.), which was first published in CPC., ii, 232 (where it is not described as a game of the match). The total score is given in (b) as 44,30,14 = 88 games; in (c) as 46, 26, 13 (but the games themselves give 45, 27, 13) = 85 games; (d) 44, 28, 13 = 85 games; (e) 44, 28, 13 = 85 games. In (a) M. plays first in games 70-74; (c) reverses the players in 71 and 73, (e) in 73 only. From internal evidence (c) seems to be right, and if we correct the totals in (a) accordingly, and an obvious misprint in the result of game 82, and then add the results of games 14 and 85, the revised totals of (a) are 45, 27, 13, and agree with the corrected figures for (c). All the editions make B. play first in games 77-80, and internal evidence supports them, though it was clearly impossible that B. could have played first in four consecutive games in the ordinary course of events. Centurini (CPM., 1864, 232) suggested that B. gave M. the odds of three games in the last match (to be of 15 games), and that these were assumed to be the games that should have come between 77 and 78, 78 and 79, and 79 and 80. This would make the total score 45, 30, 13 (cf. (b)], and agree with M.'s statement that he won eight of the last twelve games [cf. also (g)] This may be the explanation, but it is also possible, since Geo. Walker only obtained the games of the last match en bloc, that they have been disarranged; some may be wrongly ascribed, thus, internal evidence suggest that M., not L. opened and won game 80. According to (f), B. played other than match games with M., and even attempted to give him odds. B. speaks here of a total of 100 games.

In looking over these three candidate Evans' Gambits two of them reach this position: 1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.b4 Bxb4 5.c3 Ba5 6.0-0 d6 7.d4 exd4 8.cxd4 Bb6 9.Bb2.
The move 9.Bb2 seems to be Bourdonnais' second favorite method of attack after 9.d5. At no time did McDonnell show any inclination in playing 9.Bb2 (or 9.d5 for that matter). This alone would make it unlikely that either game 79 or 82 have been mistakenly assigned.

The third Evan's Gambit, game 80, however, arrives at the following position after 1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.b4 Bxb4 5.c3 Ba5 6.0-0 d6 7.d4 exd4 8.cxd4 Bb6 9.h3.
The move 9.h3 also occurred four times in games 65, 66, 72 and 83. In every case this move, 9.h3, was played by McDonnell as White and was never attempted by Bourdonnais. It would seem unlikely that he would have played it now.

If game 80 was correctly assigned making either 79 or 82 incorrectly assigned, it would also then mean that Bourdonnais had adopted McDonnell's move, 9.h3, in game 80 and that McDonnell must have adopted Bourdonnais' move preference, 9.Bb2, in either game 79 or 82. It is far more reasonable to conclude that game 80 was wrongly assigned.

Perhaps this was the "internal evidence" suggested by Murray?

Based upon the opening repertoire evidence alone, the conclusion is that game 80 has been wrongly attributed as Bourdonnais-McDonnell, and should be corrected as being McDonnell-Bourdonnais.

This reassignment of game 80 then "corrects" Chess Studies and gives the following totals of published games from each match:

 Match  LCM wins  AM wins  Draws 
11654
2450
3651
4837
5741
6360
 Total 442813

This maintains the possibility that the earlier statements of McDonnell winning 8 of the last 11 (not counting the draw), and 8 of the last 12 (counting the draw) and the earliest totals for these matches, i.e. 44 wins for Bourdonnais, 30 for McDonnell and 14 draws, as being true.

This leaves us with the mystery of the three "missing" games (two additional wins for McDonnell and a draw).

As a side-note, I had developed custom software which allows me to compare the relative playing strength of players from different eras. The data was generated using Deep Rybka 4.1x64 at a fixed depth of 14-ply for each move in a game. Then I created a strength-of-response matrix which allowed me to construct Markov chains to represent each player as he had performed in a specific event. These player-event derived Markov chains were then used to populate the custom software to pit these historic players against each other in a mock tournament (thousands of times) to create comparison rankings.

On a lark I used my technique to create a baseline for comparison of the two players in the sixth match, using games 77, 78, 79, 81, 82, 83, 84 and 85 (as "pairing known"). I then compared each of these games against the baseline and, not surprisingly, in every instance the greater probability estimate for the pairing matched the games as published in Chess Studies.

The following series of graphs shows comparisons of each game against the baseline for each player. The first half of the graph shows a McDonnell-Bourdonnais pairing, the second half shows a Bourdonnais-McDonnell pairing. Below each graph is the Chess Studies published pairing, followed by the absolute combined changes, or delta, between the individual game (for both pairings), followed lastly by a pairing estimate based upon those absolute delta values.









I then compared game 80 (which up to this point has not been factored into any computations) against the baselines. The results support the theory that the color allocation given in Chess Studies was incorrectly given for game 80. And while not a mathematically conclusive test, it was interesting to note that the software suggests the pairing to be McDonnell-Bourdonnais based solely upon the strength of the players as determined by data derived from the other published games of the sixth match.


[Archive] [Excavations] [Gallery] [Journal] [Library] [Links] [Legend] [Market]
© 1999-2021 Jacques N. Pope. All Rights Reserved.