One possible improvement on 44..Bf8 is 44...Ra5 for three reasons.

Firstly, the move 44...Bf8 removes the bishop from its defence of the c1-h6 diagonal and so allows White's Knights to come to g5 and to e3.

Secondly, in the event that White's Rook on a1 attacks and pins the Black bishop on a3, Black can break the pin ny the move ....Bb4 as the B will then defend the R on a5 eg 45 h4 Bc1 46 Rd1 Bxa3 47 Ra1 Bb4

Portisch follows Petrosian's play from Santa Monica, 1963. Petrosian showed that, even with the doubled f-pawns, his king was perfectly safe. Fischer deviates from the Gligoric-Petrosian game with 16.Qxd4.

Portisch says he saw the line in which he wins the exchange (with 17....Rd8 and 18....Nd3) but didn't realize that it was actually good for white. "Unfortunately I did not know that the whole line had been published in the May, 1966 issue of Chess Review." So -- a rare secret analysis from the American chess community!

Black can win a whole piece with 19....Nxe1 20.Rxe1 Bf4 21.Qxf6 Rxd2 22.Nxd2 Bxd2 only to be mated immediately with 23.Rd1 B moves 24.Rd3. Portisch defends precisely from move 20 to 25.

Portisch criticizes 27.Kg2 and 28.e5 and says that Fischer should have immediately played Nb3 and seized the queenside dark squares.

Portisch calls the possibility of 32.Rd2 "an astonishing threat" (with black's queen trapped in the middle of the board.

With 31....Rf4, black has the initiative.

Portisch miscalculated with his reasonable-looking plan of 33....Qc6 and 34....c4 - simply missing 36.Ned2 when the c-pawn falls.

38.Nxc4 would lose to 38....Rxf3 39.Rxf3 Rf7. So Fischer finds the pretty move 38.Rd3!