1.d4 d5 2.c4 c6 3.Nf3 Nf6 4.Nc3 e6 5.Bg5 dxc4 6.e4 b5 7.e5 h6 8.Bh4 g5 9.Nxg5 hxg5 10.Bxg5 Nbd7 11.exf6 Bb7 12.g3 Qb6 13.Bg2
There are various different ways of reaching this much-played and complicated position, and more than one way of proceeding from it: but if we take the advice of Steffen Pedersen (The Botvinnik Semi-Slav, Gambit, 2000, p.99-100) then Black should castle queenside before advancing the c-pawn. This is because, after 13...c5, White doesn't have to play 14.d5 but has a better move in 14.dxc5!. Pedersen explains:
After 13...c5 White can of course transpose to the main line, Chapters 2-4, with 14.d5 O-O-O 15.O-O b4, but with (sic) 14.dxc5! Bxc5 15.O-O is a good alternative.Now Pedersen considers both 15...Bxg2 16.Kxg2 and 15...O-O-O 16.Qe2 good for White. More specifically, he describes the latter as "a favourable version of Line B22", which variant crops up on the following page and goes 13...O-O-O 14.O-O c5 15.dxc5 Nxc5 16.Qe2. Oddly, though, in a note at Black's fifteenth he actually gives a game which proceeded 15...Bxc5 16.Qe2.
Now as far as I can see, 13...O-O-O 14.O-O c5 15.dxc5 Bxc5 16.Qe2 gives precisely the same position - with precisely the same last move - as 13...c5 14.dxc5 Bxc5 15.O-O O-O-O 16.Qe2. It's just that in the second case the castling comes after the captures, rather than vice versa. Which makes it not so much a "version" of another line but a note within it. This is a little clumsy. Given that page 100 and page 101 are facing one another, it means in effect that Pedersen is giving a line on one side of the fold that he doesn't realise he's also giving on the other.
Well, you may say, never mind, we can still follow what he's saying and moreover, Wells (The Complete Semi-Slav, Batsford, 1994) concurs, going so far as to mark 13...c5?! as dubious, and commenting:
14.dxc5! really looks promising, as Black cannot follow up...Nxc5 with ....O-O-O for legal reasons.So it's all well and good, and next time, I shall know this: 14.dxc5 is the move to play and 13...c5 is not.
But it does beg the question - why, if 13...c5 is inexact, does Pedersen, at the start of the aforementioned chapters 2-4, give the standard move order as 1.d4 d5 2.c4 c6 3.Nf3 Nf6 4.Nc3 e6 5.Bg5 dxc4 6.e4 b5 7.e5 h6 8.Bh4 g5 9.Nxg5 hxg5 10.Bxg5 Nbd7 11.exf6 Bb7 12.g3 Qb6 13.Bg2 c5 14.O-O - thereby missing, on several prominent occasions, precisely the crucial detail he is careful to point out later?