Here, I examine the misattribution of the so-called “Karstedt method” in the Philidor position, as presented in the posthumous editions of Dvoretsky’s Endgame Manual (DEM, diagram 9-17a in the 5th edition and 9-18 in the 6th edition).

D. 1



Unlike the first editions of Dvoretsky’s Endgame Manual, the 5th and 6th editions (see DEM, 6th ed., diagram 9-18, position after 1.Kf6) describe the move 1…Re1! as “Karstedt’s draw”: “But in 1897 Max Karstedt discovered ‘Karstedt’s draw.’ We call it the second defensive method in the Philidor position: an attack from the rear allows Black to hold.”

This description, although not unique in chess literature, is historically inaccurate. Another historical error appears in Jesús de la Villa’s well-known book 100 Endgames You Should Know (2008, pp. 129-130; now in its sixth edition), where both the method and the specific move 1…Re1 are attributed to Kling and Horwitz. De la Willa writes after 1(2)...Re1: “ This is the K&H 'trademark'. The rook stands behind the pawn to prevent its advance”.

In fact, the move 1…Re1! in position D. 1 - the pure “attack from the rear” - was discovered independently by Orrin Frink and Ilya Rabinovich in 1926, almost thirty years after Karstedt.

Philidor’s second variation

The position of D. 2 originates in Philidor’s 1777 analysis (Analyse du jeu des Échecs, 2nd ed., p. 271):

D. 2



Philidor believed that Black loses after 1.e5 Ra1 2.Kf6 Rf1+ (?!) 3.Ke6 Kf8 4.Rh8+ Kg7 5.Re8 Re1 (?!) 6.Kd7 Kf7(?) 7.e6+ Kg7 8.Ke7(?) Re2(?) 9.Rd8 Re1 10.Rd2 Re3 11.Rg2+ Kh7 12.Kf7 Rf3+ 13.Ke8 Re3 14.e7 Rd3 15.Rc2 Kg7 16.Rc7 Rd2 17.Rd7 Rb2 18.Rd1 Rb8+ 19.Kd7 Rb7+ 20.Ke6 Rb6+ 21.Rd6 Rb8 22.Rd8 (all signs are mine).
For more than a century this was considered proof that the defender must never leave the 6th rank before the pawn advances.

Karstedt’s real contribution

1.e5 Ra1 2.Kf6 Rf1+ 

Interestingly, Karstedt briefly remarks that he would have preferred 2...Re1. But his actual idea had nothing to do with this move; it centered on 5…Ra1!.
 
3.Ke6 Kf8 4.Rh8+ Kg7 5.Re8

D. 2a



In 1897, Max Karstedt demonstrated that in the second variation of the Philidor position, leaving the 6th rank too early does not lead to a loss (Deutsches Wochenschach, no. 18, 2.05.1897, p. 145-146). After 5.Re8, Black can secure a draw through a combined defensive strategy. With 5…Ra1!, Black creates the threat of flank checks, forcing the white rook to leave the e8-square, where it was defending its pawn. Taking advantage of this, Black returns the rook to the pawn’s file and, by attacking the pawn from the rear, once again prevents its advance.This back-and-forth maneuver of the rook characterizes the Karstedt method. In other words, Karstedt’s method combines a flank attack and a rear attack, but not the direct move 1…Re1! from the start.

5...Ra1 6.Rd8 Re1 7.Rd5 

7.Kd6 Kf7=.

7...Kf8 8.Kd7 

8.Kd6 Ke8=.

8...Kf7=.

It is worth noting that Karstedt, and later Berger (1922, p. 280), believed that in Diagram 2a 5…Ra1! is the only drawing move and that Philidor’s move 5...Re1 loses. In 1906, Tarrasch correctly observed that 5...Re1 does not lose, but he uncritically continued with Philidor’s second variation (6.Kd7 Kf7? 7.e6+ Kg7 8.Ke7?) and only after the erroneous 8.Ke7? demonstrated how Black could draw (Tarrasch, 1906, p. 269-271). The correct drawing method after 5...Re1 6.Kd7 was established by Frink in 1926 - by means of 6…Ra1 or 6…Rd1+.

Frink and Rabinovich, 1926 - the true discovery of 1…Re1!

Back to D. 1

Even after the discovery of M. Karstedt's method, position of D. 1 was considered lost for Black until 1926. In the second edition of his Theorie und Praxis der Endspiele (1922, p. 280) Berger attempted to prove that Karstedt's method saves Black only because of the poor positioning of the white rook on h7, which gives Black the opportunity to win a tempo after 4.Rh8+ Kg7. Only in 1926 both Orrin Frink and Ilya Rabinovich independently demonstrated that even after 1.e5 Rb6 2.Ra7 Rb1?! 3.Kf6, Black holds by 3…Re1! It is precisely this technique that constitutes the essence of the attack-from-the-rear method.

Frink published his analysis in The Chess Amateur (July 1926, p. 297-298, August 1926, p. 329). Rabinovich presented it in May 1926 at the Advanced Courses for Chess Teachers (Leningrad) and included it in the first edition of his textbook, Endgame (Leningrad, 1927, pp. 256–257). Remarkably, Rabinovich assumed that the move Re1 had already been discovered before him. He wrote, "Unfortunately, we do not know who was the first to discover this move in this position." A year later, Grigoriev (64, no. 1, 1927, p. 16) and Chéron (Traité complet d’échecs,  1927, no. 147, p. 106) confirmed that 3...Re1! draws. 

The position and its defensive logic

1...Re1!

The idea of the attack-from-the-rear method is to prevent the pawn’s advance from the fifth to the sixth rank. For this purpose, the rook attacks the pawn along the file on which it stands, while the king - after giving check from the eighth rank - takes up a position on the seventh rank (in the case of a bishop’s pawn, only on the short side of the board). As soon as the white king vacates the square in front of his pawn, creating the threat of advancing it to the sixth rank, Black places his king on the square adjacent to the pawn’s file. By controlling the critical square of the sixth rank, Black stops the pawn’s advance.

Unlike in Philidor’s second variation (with the white rook on the short side - h7, after 1...Ra1 2.Kf6), in D. 1 the move 1…Re1! is strictly the only one that holds. J. Berger considered only 1...Rf1+?, which loses after 2.Ke6: 2...Kf8 3.Ra8+! Kg7 4.Ke7 (also possible are 4.Kd6 or 4.Kd7) Rf7+ 5.Kd6 Rb7 6.e6! Rb6+ 7.Kd7 Rb7+ 8.Kc6 Re7 9.Kd6 Rb7 13.e7.

The position arising after 1...Rf1+? 2.Ke6 is generally known in chess literature as Lasker’s position, as it appeared in his Lehrbuch des Schachspiels (1925, p. 159; p. 181 of the 1947 English edition). Lasker demonstrated only that Black also loses after 2...Kd8. 

Back to D. 1 after 1…Re1!

2.Ke6

As Frink noted, 2.Ra8+ Kd7 gives White nothing. 

D. 1a



Now that Black's rook has occupied an ideal position on the e-file, retreating the king to the long side - 2…Kd8 (II) - is also a viable option, although retreating the king to the short side - 2…Kf8 (I) - is simpler. The weaker side only has this option with a central pawn, but not with a bishop pawn: the short side with a central pawn (3 files) is not as short as with a bishop pawn (2 files), and thanks to this, Black can achieve a draw through a flank attack. If Black's rook occupies a position behind the pawn, then the position of the white rook on the long or short side is irrelevant.

I. 2…Kf8 3.Ra8+ Kg7

D. 1b



Because of the rook's position on the e-file, White, even after forcing the black king off the pawn's promotion square, is unable to begin its further advance. After 4.Kd6 follows 4…Kf7! - and Black holds the critical square, maintaining balance.

4.Re8

The only way to strengthen White's position is to transfer the rook to the pawn file to facilitate its advance. Here, unlike variation II (2…Kd8?!), where the black king is on the long side (d. 2g), after 4.Re8 White has no immediate threats. Black has several defenses, but the most consistent is the Karstedt method, which, as we have already seen, does not allow the pawn to advance to the sixth rank at all.

4…Ra1

Now we have a transposition to the D. 2a, Karstedt’s draw. 

A month after publishing his analysis, O. Frink proved that with the weaker king on the short side, Karstedt's back-and-forth maneuver is not the only defense (The Chess Amateur, August 1926, p. 329). After 5.Rd8, Black can also immediately resort to a flank attack.

5…Ra6+?!

O. Frink cites this move as one of many continuations leading to a draw. Indeed, any other rook move on the a-file results in a draw. As soon as the pawn advances to e6, the famous drawn Tarrasch position of 1906 arises. However, White may not rush with the e5-e6 move.

6.Rd6 Ra8 7.Ke7 Ra7+ 8.Rd7

Frink considered only this move. White has the more dangerous 8.Ke8, after which Black is required to play accurately։ 8…Ra8+! 9.Rd8 Ra6! (Chekhover, in 64, 5.02.1937, p. 3, referring to Rovner) 10.Rd7+ Kg8 11.Ke7 Kg7 12.e6=. 9…Kg6? loses to 10.Rd6+! Kf5 11.e6 Ke5 12.Rb6! Nf6 13.e7+ Kg7 14.Re6. Instead of 12.Rb6! Chekhover and Levenfish suggested 11.Rd7? Ra8+? 12.Kf7 with a win (Shakhmaty v SSSR, n. 2, 1937, p. 51). However, after 11…Ra1 (or Ra2, Ra3, Ra4) 12.e7 Ke6! the position is drawn.

8…Ra1 9.e6 Kg6=.

Thanks to the king's favorable position on the short side, Black has after 4.Re8 Ra1 5.Rd8 numerous other draw possibilities. Besides 4…Ra1, any other rook move on the a-file (except, naturally, Ra7 or Ra8) or on the first rank (except, naturally, d1) yields a draw. Black loses only with 4…Kg6 and 4…Kh6, since after 5.Rg8, Black's king is cut off on the g-file. If Black doesn't play 5…Re1, White manages to advance his pawn to the sixth rank, but thanks to his opportunity to occupy the a-file, Black achieves a draw. As already noted, the key here is the Tarrasch position. Despite the abundance of drawing possibilities, it is also clear that 5…Re1 is the most preferable, as it deprives White of any significant practical winning chances - White is unable to advance his pawn.

II. 2...Kd8?!

In 1926–1927, I. Rabinovich focused exclusively on this defensive response by Black. In his article in Shakhmaty v SSSR (no. 11, 1936, p. 330) and in the second edition of his textbook (1938, p. 321; English edition, 2012, p. 364), he strongly preferred 2...Kf8, even marking it with an exclamation point. He argued that, methodically, it is correct for the king to occupy the short side rather than the long side. 

D. 1c



2.Ra8+ Kc7 3.Re8

If 3.Kf6, then 3…Kd7! Having protected the pawn with 3.Re8, White now threatens 4.Kf7. A waiting strategy after 3.Re8 would fail. However, since it is now Black’s turn, he can initiate an effective flank attack from the short side.