Fastidious or morbid travelers have been known to shine a black light in their hotel rooms, revealing unspeakable stains and smears everywhere. Turning an engine on this game provides a similar experience. Notes by Tarrasch and Georg Marco are as much of a mess as the game itself.

Tarrasch and Janowski commit several blunders and a battalion of smaller errors, and are also guilty of some major strategic misjudgments. Things begin going wrong early, and rarely go right thereafter. It appears that both players drastically overestimate Black’s risky early bid for counterplay, causing them to miss multiple opportunities for White to obtain clear advantages or even near-wins (12.a5). Then both sides neglect the center, then Janowski allows his bishop to be driven into a cul-de-sac, then Tarrasch fails to take full advantage (24.h4). After three consecutive White mistakes (27.Bxg5, 28.Nf3, 29.b3) the initiative passes more or less permanently to Black. (Marco thinks 25.a5 is White’s critical mistake. It’s not a good move, but in this game, it hardly stands out.)

In the slow maneuvering that follows, Janowski allows, and Tarrasch overlooks, a chance for White to break open the position. Tarrasch nevertheless probably ought to be able to survive, but he makes makes several more unforced defensive errors. After they both miss the immediately decisive 60....Be1+, Janowski is finally able to infiltrate his queen and bishop and force resignation.

The final tableau is striking, and the game looks grand from afar. But if you want to remain lost in admiration, keep your distance, and for God’s sake leave your flashlight at home.

The annotators, Tarrasch and Marco, repeat most of the players’ errors and add a few of their own, repeatedly drawing attention to imaginary turning points and missing real ones. Though neither does well, Tarrasch comes off somewhat better than Marco, probably because, as the game’s loser, he’s less prone to overestimate it.

Tarrasch and Janowski were great masters, and Tarrasch and Marco were terrific annotators. Everyone is familiar with Tarrasch’s writing. Marco is less well known today, but he was very well-regarded in his own time, and his wit and humor shine through in the notes here. Marco’s annotations in the Carlsbad 1907 tournament book, available in Dale Brandreth’s translation, are a treasure. So, what went wrong? As for the game, (i) everyone has bad days, and (ii) chess really has come a long way in the last hundred years. As for the annotations, I think Marco was badly handicapped by his belief that he had a masterpiece on his hands, while Tarrasch was led astray by a flaw that afflicted him as a player and an writer: he thought deviations from his narrow conception of correct play would, or should, always be punished. Since he made several “incorrect” moves by his own criteria in this game, he was strongly inclined to underestimate his own position, and to believe that he deserved defeat. Finally, of course, an engine can make almost any human look bad, over the board or on the printed page.

Perhaps not relishing the thought of facing his own Open Defense, Tarrasch played the unusual 5.Nc3, and, I suspect, almost immediately decided he shouldn’t have.

After 8.a4.

ST: One of the positions in which the strength of the defense d6 comes so fittingly to account. The white king’s bishop is endangered. Those most natural and powerful developing and attacking move, 8.d4, cannot be played, for after 8….Nxd4 9.Nxd4 exd4 10.Qxd4 c5 and …c4 the bishop is lost. On the other side, Black threatens to exchange the bishop advantageously with …Na5. The rook’s pawn’s move should remove both dangers, but even this does not appear to be good; the sacrifice of the King’s pawn, with which it is bound up, is not entirely correct. White has no prospect of an attack, however, after a tame continuation like d3.

After 8….b4

GM: If 8….Rb8, Black would have relinquished the a-file to Black; besides, after 8.axb5 axb5, the pawn on b5 is weak and must advance sooner or later to b4. Why, then, not play …b4 at once, says Janowski, rightly, to himself. How well-founded his view is, can be seen after the 10th move of Black.

<8....b4 also has its drawbacks; Marco’s overestimation of the move has a lot to do with his subsequent mistakes in analysis.>

9.Nd5 Bg4

ST: A strong move, developing, attacking, and calling into question the whole way of playing 8.a4. It would not have been so good to capture the pawn on e4, for White would have won back the pawn with an attacking advantage by d4.

<This move turns out to be a poor fit with 8....b4, leaving Black with serious weaknesses on the queenside. SF10 thinks the game is even after 9….Na5 10.Ba2 Nxd5 11.Bxd5 c6 12.Ba2 c5.>

10.Nxf6+?!

ST: Otherwise c3 comes into consideration, a move employed in the later game Tarrasch-Capablanca, that is still worse, just because the King’s pawn can then be captured with good play for White.

GM: In my opinion, Black is considerably in the advantage in this position. The threats …Nd4 and …Rg8 are very unpleasant and one can also see from this game that the popular earlier attacking system, 8.a4, is of very questionable value.

<The engine thinks 10.c3 is best, with the following continuation if Black took the pawn: 10.c3 Nxe4 11.d4 bxc3 12.bxc3 0-0 13.Re1 Nf6 14.h3 Bh5 15.Nxe7+ Qxe7 16.g4 Bg6 17.Bg5 Rae8 18.a5! (you’re going to see that move a lot) Qd8 19.Ba4 Qa8 20.Bxf6 ±>

10….gxf6?

<As GoldenBird pointed out, only a gambling man would make this move. White is slightly better after 10....Bxf6, but now he can get more.>

ST: If the bishop captures, Bd5 can be played without Black getting much from the exchange sacrifice …Nd4. Now Bd5 could be answered by …Nd4; then black wins back the sacrificed exchange with advantage, for example 12.Bxa8 Qxa8! 13.c3 Nxf3+ 14.gxf3 Bh3 (threatens mate in a few moves by …Rg8+, …Bg2+, and …Bxf3+); 15.Kh1 Bxf1 16.Qxf1 f5.

GM: If White now wanted to move 11.Bd5, he would come into difficult situations. One might consider I. 11….Nd4 12.Bxa8?? Bxf3 13.gxf3 (forced, since 13.Qe1 Nxc2 would have led to the loss of the Queen) Qc8 14.Kh1 Qh3 15.Rg1 Nxf3 16.Rg2 Kd7!! and White cannot cover the threatened mate. II. 11….Nd4 12.c3 (probably the best) c6! 13.cxd4 cxd5 14.d3 f5 Black is in the advantage.

<Unfortunately, SF shows that Marco’s variation I. is too good to be true. After 11….Nd4 12.Bxa8 Bxf3 13.gxf3 Qc8 14.Kh1 Qh3, White has 15.Bc6+! Kd8 16.Rg1 Nxf3 17.Rg2 Bf8 (so that he can play ….Rg8 without allowing the rook to be taken with check) 18.d3 Rg8 19.Qg1!! How’s that! 19….Bg7 20.Rg3 Nxg1 21.Rxh3 Nxh3 22.Kg2 Bh6+ 23.Kxh3 Rxg1 24.Bxh6 Rxa1.

click for larger view

You’d think that White’s bishops would be stronger than the rook in the ending, but SF gives this position a triple zero, rightly I think.

11….Rg8? 12.Bd5?

<White can get a near-won position with 12.a5!.

click for larger view

The threats of Ba4 and Qb3 (or Qe2) and Qc4, together with White’s ability to advance in the center, make it very difficult to survive. For example:

12….f5? 13.h3! Bxh3 14.Ne1! Bg4 15.f3 Bh3 16.Ba4 Qd7 17.Qb3 Bh4 18.Rf2 Kf8 19.d4 fxe4 20.Qd5 Bxf2+ 21.Kxf2 exf3 22.Qxf3 d5 23.dxe5 Be6 ( 23….Bf5 24.Bxc6 +-) 24.Bh6+ (+5.33, 39 ply)

12….Bd7? 13.d4 bxc3 14.bxc3 Rb8 (+2.97, 39 ply) 15.Re1 (15.d5 is also very strong) Bf8 16.Bc4 Qc8 17.Qd3 Ra8 18.g3 Ne7 19.Bd2 Bc6 20.dxe5 dxe5 21.Nxe5! fxe5 22.Qf3 Qf5! 23.Qxf5 Nxf5 24.exf5

12….Qd7 13.d4 exd4 (13….f5 14.Qd3 (+1.59, 40 ply) fxe4 15.Qxe4 0-0-0 16.d5 Nb8 17.Ne1 Qf5 18.Qxf5+ Bxf5 19.cxb4)>

12….Qd7 13.Kh1 Rb8 14.d3

ST: On the fourteenth move, White finally finds the time to move the Queen’s pawn. Serious errors must have been committed in the opening, for otherwise this sort of thing could not have happened. Black has developed both more quickly and better and has the attack. In particular, the pin on the knight is very troublesome for White.

14….Nd8?! 15.Rg1?!

ST: So as to break the pin with h3 and g4 without having to be resigned to a sacrifice.

<So as not to insert evals and long engine lines after these and the next few moves, what appears to be going on here is that each side should be striving to advance its d-pawn, and to hinder the other side from doing the same. Thus 14….Nd8 is dubious because it ignores the center, 15.Rg1 is questionable because it allows ...c6 with tempo, and 16.Bc4 is a mistake because it allows ...d5 with tempo. 16….Qb7 is a waste of time, because if 16….bxc3 17.bxc3 d5 18.Bxa6??, 18….Ra8 traps the bishop.>

15…..c6 16.Bc4? Qb7? 17.h3 Bh5 18.cxb4?

ST: The unfortunate bishop was threatened by …d5; if it had then gone to a2, it would have been shut in by …b3, and if it had gone to b3, the situation would have been still more disagreeable for White.

GM: Again, plainly forced. Bad was 18.Be3 on about of …d5 for 19.Bb3 would now be entirely bad on about of 20….bxc3!; and 19.Ba2 b3 20.Bb1 is plainly not worth pursuing, since the rook on a1 and the bishop on b1 would have been condemned to long-term passivity; but moreover, …Ne6 followed by …d4 threatens totally to entomb the two mentioned pieces.

18….d5 19.Ba2 Bxb4 20.Qe2 Ne6 21.Be3

ST: White decides as late as he can, and then reluctantly, upon the move g4, which to be sure, shuts in the bishop on g6 but also makes the later attack on the g- and h-files possible.

21….d4?

<A positional blunder, to be followed by a tactical one. >

22.Bc1 Be7?

<22.…Bxf3 was forced, with a considerable advantage for White.>

23.g4 Bg6

click for larger view

24.Nh4?

<Instead the simple 24.h4! is winning (+3.37, 38 ply): 24….h6 25.h5 Bh7 26.Bxe6 fxe6 27.Bxh6 Bd6 (27….Qxb2 28.Nd2! (threatening Rab1) Qb6 29.Rab1 Qd8 30.Rxb8 Qxb8 31.Rb1 Qc7 32.Nc4) 28.Nd2 Qe7 29.Nc4, or 24….h5 25.gxh5 Bh7 26.Rxg8+ Bxg8 27.Nh2. But White still has a sizeable advantage after the text move.>

24….Rh8 25.a5

ST: So as at least to create a point of attack on a6. Of course the pawn on a5 itself becomes weak later on.

GM: I have long considered whether Dr. Tarrasch followed his good guardian angel or an evil demon in this move. Obviously the pawn on a6 now appears to be lastingly weak; but in reality its colleague on a5 is much weaker still. One sees that distinctly only forty moves later; but that a5 is not very frightful, is shown on the 26th move by Black, for White must now determine upon Bxg5 (since his pawn on h3 is threatened and Kh2 would be inadequate on account of …h5) and the black bishop on e7 then occupies a dominating position later on b4.

The next moves are readily comprehensible. Black besieges the King’s wing, White seeks to consolidate it, and he also succeeds in that without difficulty. Through Black’s 36th move, the King’s wing is completely barricaded; Janowski now guides his troops to the Queen’s wing again; and after long fumbling, trying, and studying, he finds the way to assault the enemy bulwark.

<Burn recommended 25.Bc4 and b3.

SF10’s top line (+1.56, 37 ply) runs 25.Bc4 Ng5 26.Kg2 Ne6 27.Rb1 Qd7 28.Bh6 Bf8 29.Qd2 Qd6 30.Bxa6 Qb4 31.Qxb4 Rxb4 32.Bd2 Rxa4 33.Ra1 Rxa1 34.Rxa1 Kd7 35.Bc4 Nc7 36.f4 (again taking advantage of Black’s badly-placed bishop) exf4 37.Bxd6 Kxd6 38.Nf3 with an obvious advantage.

25.a5 is the engine’s 8th choice, but as shown by the evaluation (+1.14, 37 ply) there isn’t an enormous difference between it and more favored moves (in addition to 25.Bc4, SF also likes 25.Nf3 Qd7 26.h4 (essentially retracting the error on move 24), and 25.Nf5).>

25....Kd7

ST: Black can carry out many evolutions, and White, in his cramped and threatened position, can do nothing about it.

26.Bc4 Ng5 27.Bxg5?

ST: An advantageous exchange for Black, since with the removal of the doubled pawn, the confinement of the bishop on g6 also comes to an end. But White cannot permanently protect the pawn on h3 with the King in view of the threat of …h5 and …Rbg8.

<A massive positional concession that should be avoided unless it is absolutely forced. And it isn’t. 27.Kg2 Ne6 28.b3 Qc8 29.Qf3 Kc7 30.Nf5 Bxf5 31.exf5 Nd8 32.Qe4 and you don’t need an engine (+1.72, 42 ply) to know White is much better.>

27….fxg5 28.Nf3?

ST: Time pressure played a role here. Nf5 naturally came into consideration, too, and was better, since from now on the knight generally no longer reaches a halfway decent post. Of course, if Black could then have won the b- and f-pawns for the a-pawn, it would nevertheless have been disadvantageous to him on account of the passed pawn generated.

<The knight has a miserable time for the rest of the game. Only here does Black get a slight advantage. 28.Nxg6 as well as 28.Nf5 was better than the text move.>

28….f6 29.b3?

29….h5

click for larger view

<Now Black’s advantage is obvious; he has attacking chances on both sides, while White has no prospects.>

30.Kg2 Bb4 31.Qa2?

<The queen is needed to help defend the kingside.>

31….Ke7 32.Rh1 Rh7 33.Rh2 Rbh8 34.Rah1?

ST: This passive defensive position is always the punishment for the hara-kiri move g4.

<Curiously, this move makes it harder to defend g4.>

click for larger view

34….Be8?

<Stronger is 34….Qc8 35.Kg3 Be8 36.Ng1 Bd7 37.f3 Qc7 38.Kg2 Qxa5 39.Qxa5 Bxa5 winning a pawn. White doesn’t have time for 40.Bxa6 because of 40….Ra8 41.Bc4 Bd2 42.Ne2 Ra2 43.Kf1 h4 and he’s hopelessly trussed up.>

35.Ng1

ST: So as to be able to defend the pawn on g4 with the f-pawn in addition in case of need. At the same time, the knight should be played via e2 to g3.

35….c5

ST: Black wants to move the bishop on e8 to b5, so that the pawn on a6 will not have to be permanently defended by the Queen. The move’s only drawback is that it entirely shuts out the bishop on b4. But if he played it to d6 and then played …c5, it would have been entirely shut in. Thus it was only a question of being shut in or out.

36.Ne2 h4

ST: Thereby Black renounces every possibility of attack along the h-file. But otherwise the knight would have come to g3, and from there, after Black’s …Bb5, to f5.

GM: Black first disables the white knight.

<36….Bd7 first, forcing the weakening 37.f3 (37.Kg3? hxg4 38.hxg4 Rxh2 39.Rxh2 Rxh2 40.Kxh2 Bxg4) would have been stronger. But Tarrasch later plays f2-f3 gratuitously.>

37.Ra1

ST: Now White has little more to fear, and the game should have been drawn.

37….Bb5 38.Rhh1 Rf8

ST: Threatens to advance the f-pawn.

GM: Only a trial balloon. Perhaps White overlooked …f5! One need ascribe no deeper significance to the now following troop movements by the two heroes from the 40th to the 51st move by White. Dr. Tarrasch tacks because he cannot undertake anything at all; Janowski takes time, on the contrary, because his opponent cannot escape. Only the 51st move by Black (…Rd8) reveals a small part of the dangers to which White is exposed.

39.Kh2 Bc6

ST: Black still threatens to reinvigorate, through the sacrifice of the exchange, an attack that has come to a standstill.

40.Qc2 Qc8 41.Rhf1 Rg7 42.f3?

ST: A very doubtful move. The King must be protected on the side by one pawn or another, the f-pawn or the g-pawn, otherwise it is exposed later without protection to all attacks, as here appears.

click for larger view

42….Kd6 43.Ra2 Ra7

ST: Something can happen only on the a- and b-files (after ...Bb5 along with ...Bxc4 and ...Bc3).

44.Raa1 Qd7 45.Kg2 Rb8?

click for larger view

46.Rfb1?

<It was easy to miss, but Black’s clumsy array allowed White to break free with 46.f4!! gxf4 47.Qc1 Kc7 48.Nxf4! exf4 49.Qxf4+ Kb7 (49….Qd6? 50.Qxd6+ Kxd6 51.Rxf6+ Kc7 52.g5 Raa8 52.g6 Rg8 53.g7 and Black loses) 50.Qxf6 Raa8 51.Bd5! Rf8 (51….Bxd5?? 52.Qb6+) 52.Qh6 and 52….Qc7 53.Rxf8 Qg3+ leads to a perpetual.>

46….Bb5 47.Ra2 Rab7 48.Bd5 Ra7 49.Kf1 Qd8 50.Rba1 Qc7 51.Kg2 Rd8 52.Kf1

GM: Janowski has composed a problem of the highest elegance. His next move renders his opponent defenseless.

52....Ke7

GM: “If you throw a pea into the Mediterranean Sea, a sharper eye than ours could observe the effect of it on the Chinese coast.” — says G. Chr. Lichtenberg somewhere. Here we can see that he is right. The position on the 25th move was the Mediterranean Sea and a5 was the pea, which Dr. Tarrasch threw into the sea. Now we are standing on the Chinese coast and can see the dreadful effects of that act. White’s rooks are crippled because the pawn on a5 is weak; the Queen must defend the pawn on d3, the knight on e2 dare not stir, since after ...Bc3 the pawn on a5 would be immediately lost. Can an army that is so very much impaired in its freedom of movement be capable of defense?

53.Bc4

ST: Black threatened to sacrifice the exchange and then renew the attack on the king with ...Qxd5 and ...f5.

GM: The beginning of the end; there was, in general, no salvation from the threat of ...Rxd5, e.g.

I. 53.Qb1 Rxd5 54.exd5 e4! 55.fxe4 Qh2 56.Ng1 Qg3 and wins.

II. 53.Kg2 Rxd5 54.exd5 Qd6 55.Qd1 Qxd5; the position of Black is so superior that the defense of White appears to be without prospects. 53....Rb7 54.Kf2 Bxc4?

GM: The deciding deed is accomplished. Janowski now dominates the open b-file and soon penetrates into the enemy king.

<SF doesn’t like this move, though its preferred lines seem to amount to a lot of shuffling without genuine progress. It seems that Black has to play …Bxc4 sooner or later, but it may just be impossible for Black to win from here against accurate defense.>

55.bxc4?

GM: Qxc4 permits longer resistance, although the pawn on a5 would soon be lost after 55....Rb5. But White still could, for all that, give his opponent many hard nuts to crack with 56.Ra4.

<After 55.Qxc4 Rb5 56.Kg2 Rxa5(?), it’s dead even after 57.Rxa5 Bxa5 58.Qxa6. Although SF’s evaluation hovers around -1.5 after 55.Qxc4, it struggles to come up with any constructive plan for Black. Basically, it’s very hard for Black to win White’s a-pawn without losing either his own a-pawn or c-pawn. After 55.bxc4, though, Black does seem to be able to force his way into White’s camp.>

55....Rdb8 56.Kg2

ST: White is fully prepared against the single threat of …Bc3.

56….Qd8? 57.Rb1

ST: Parries the threat to penetrate with the rook to b3 after ...Bc3; Nxc3, dxc3; Qxc3. Upon 58....Bxa5, White would win the pawn back by Qa4 after the exchange of rooks.

57....Ke6 58.Qa4

ST: A completely unfounded change in the formation. White should move the King, preferably to h2, after which Black could accomplish something only with difficulty.

58....Qc7 59.Rab2?

<White could have retracted his last move with 59.Qc2, since 59….Bxa5? 60.Rxb7 Rxb7 61.Qa4 is dead equal. The idea behind 59.Rab2 is that 59….Bxa5 allows Qe8+ after the exchange of all the rooks. But Janowski’s simple reply ends that threat, after which the a-pawn falls.>

59….Ke7 60.Kf2?? Bd2?

<Both Tarrasch and Marco note that Black could have won the exchange with 60....Be1+, though Marco suggests that Janowski may have wished “to show that a much more elegant winning possibility exists.”

White probably still could have drawn with 60.Nc1! Bxa5 61.Rxb7 Rxb7 62.Rxb7 Qxb7 63.Nb3! (63.Qxa5? Qb2+ is a hopeless queen ending) Bb4 64.Kf1 and it’s hard to see how Black can break through White’s light-square blockade. After 60.Kf2 Bd2, we can finally say that White is lost.>

61.Rxb7 Rxb7 62.Qa2 Be3+ 63.Kf1 Qb8 64.Rxb7+ Qxb7

ST: Now Black is master of the game. He can penetrate the enemy game via the open file that he dominates; the pawn on a5 must fall and the knight plays a pitiful role opposite the bishop.

65.Qc2 Qb4 66.Qa2 Kd6

GM: The stadium of agony.

67.Kg2 <67.Qa1 Qb3> 67….Kc7 <Janowski could have played …Qe1 at once, but it’s really just a matter of how quickly Black wants to win here> 68.Kf1 Kc8 69.Kg2 Qe1 70.Qb2 Qf2+ 71.Kh1 Qxf3+ <of course 71….Bf4 could be played here, but grabbing the pawn first doesn’t spoil anything> 72.Kh2 Qf2+ 73.Kh1 Bf4 White resigns.

GM: A true tournament game of the kind that is found less and less frequently found in modern tournaments. Once again we see a systematic siege, a narrower and narrower encircling accomplished by Janowski with incomparable mastery. Dr. Tarrasch is completely blockaded and his army is finally lost to zugswang. Of course many thoroughly ineffective moves and maneuvers appear; but one does not merely believe that they are completely pointless. Much is achieved in war, if one tires his opponent, leads him astray until new constellations of the grouped forces permit direct attack. One can then easily overpower the confused, worn-down opponent.

The king doesn't need to be safe from checks. Or, rather, there's another path to safety: 67....Qe1 68.Qb2 Qf2+ 69.Kh1 Qxf3+ 70.Kh2 Ke6 71.Qb6+ Kf7 72.Qb7+ Kg6 and White's queen has to come back to guard the knight.

<The white pieces are caught in a cage. They can do nothing except uselessly moving to and fro. So the king can calmly walk to c8 without any risk. Janowski prefered 67. ... Kc7 and 68. ... Kc8 because he simply did not want his king to be molested by checks. This shows a rational thinking in order to find the best way.>