Jerome Gambit: Ghosts of Moves Unplayed

Playing a 3-minute game often requires both players to put aside any thoughts of what moves they could have played, or should have mande, but didn't. With more time to reflect, such musings might help in forming future plans, but in blitz, they can haunt ongoing play.

BoomBoomTNT-YT - tyronev1
3 0 blitz, Chess.com, 2020

1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+

4...Kxf7 5.Nxe5+ Nxe5 6.Qh5+ g6 7.Qxe5 

Now, Black can choose Blackburne's Defense, 7...d6, or Whistler's Defense, 7...Qe7 - both are complicated and dangerous - or amongst a handful of lesser lines, e.g. 7...Bd6.

7...Qe7 

8.Qd5+

We have recently taken a look at this infrequently- played move. See "Jerome Gambit: Largely Overlooked by History".

Recent alternatives:

8.Qxe7+ Kxe7 9.b3 a6 10.Bb2 Nf6 11.O-O Kf7 12.Nc3 Bd4 13.Rae1 Bxc3 14.Bxc3 d6 15.e5 dxe5 16.Bxe5 Rg8 17.Bxc7 Bg4 18.f3 Bh5 19.Bd6 Rad8 20.Bb4 g5 21.Re7+ Kg6 22.Rxb7 Rg7 23.Rb6 Rd4 24.Bc3 Rd8 25.Rxf6 checkmate, Ghigi,N - NN, 5 1 blitz, 2020;

8.Qxh8 Ke6 (8...Qxe4+ 9.Kd1 Qxg2 10.Qxh7+ Kf8 11.c3 Qxh1+ 12.Kc2 Qe4+ 13.d3 Qa4+ 14.b3 Qg4 15.Bh6+ Nxh6 16.Qxh6+ Kf7 17.Qh7+ Kf6 18.Nd2 Bxf2 19.Ne4+ Kf5 20.Qf7+ Ke5 21.Qf6+ Kd5 22.c4 checkmate, tomkoolen1996 - ThomasBovenkerk, 10 5 blitz, lichess.org, .com, 2020; or 8...Nf6 9.d3 b6 10.Bg5 Bd4 11.c3 Bb7 12.Bxf6 Bxf6 13.Qxh7+ Kf8 14.Qxe7+ Black resigned, Anonymous -Anonymous, 5 0 blitz, lichess.org, 2020) 9.Qxg8+ Kd6 10.Qd5 checkmate, decidueyemainbtw -MarcosEngenheiro, Chess.com 2020; and

8.Qf4+ Qf6 9.Qg3 Bd6 10.f4 Bxf4 11.Qb3+ Qe6 12.O-O Qxb3 13.Rxf4+ Kg7 14.axb3 d6 15.d4 Ne7 16.Nc3 Bd7 17.Be3 a6 18.Raf1 Rhe8 19.Rf7+ Kh8 20.Bg5 Nc6 21.Bf6+ Kg8 22.Rxd7 Black resigned, Wall,B - Guest744598, PlayChess.com, 2020.

8...Kg7

The text move looks a little bit better than 8...Kf8, which might place the King in the way of a Rook that wants to be developed, e.g. eronald - ayushsankar1006, 5 0 blitz, lichess.org, 2020 (0-1, 23). 

9.d4 Bb4+

Not quite as strong as 9...Bb6, levigun - obviously, 5d + 2d GameKnot.com, 2004, (0-1, 11) 

10.c3 Nf6 11.Qe5 

White has to be careful, with only one piece developed vs three, even with his strong pawn center. Fortunately, Black helps him out.

11...Qxe5 

Exchanging Queens in the Jerome Gambit if often a help to the defender. In this case - a 3-minute blitz game - Black has not looked deeply enough, however.

12.dxe5 Nxe4 13.cxb4 Re8 

14.Nc3 

A bit stronger was 14.0-0 Rxe5 15.Bf4 Re6 16.Bxc7. This possibility seems to haunt both player's thoughts during the next few moves, perhaps unconsciously.

14...Rxe5 15.O-O 

Of course, White can not now play 15.Bf4 Re6 16.Bxc7 because of 16...Nxc3+.

15...d5 16.Be3 

This is a solid move, although, again, 16.Bf4 was on.

16...c6

Black suddenly realizes that the 16.Bf4 skewer could win a pawn, so he moves that pawn to safety. But he overlooks something.

17.Bd4 

Black resigned

The Rook is a goner.

Jerome Gambit: Largely Overlooked by History

I wanted to share another eronald (of lichess.org) game for a number of reasons. He faced one of the more challenging defenses to the Jerome Gambit, he selected a line of play that was recommended over 140 years ago - and which has been scarcely played at all. This game also allows me to tuck in a note from my research that comes from I-do-not-know-where. 

eronald - ayushsankar1006
5 0 blitz, lichess.org, 2020

1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+ 

4...Kxf7 5.Nxe5+ Nxe5 6.Qh5+ g6 

This can lead to either the Blackburne Defense, 7...d6, or Whistler's Defense, 7...Qe7. Both are complicated and each can be dangerous for the unwary. 

My preference to play, and not to face, is Whistler's. See "More (Update): Whistler's Defense" for a discussion.

7.Qxe5 Qe7 8.Qd5+ 

The highly dangerous (for White) 8.Qxh8 appeared in Jerome - Norton, D., correspondence, 1876 (1/2 - 1/2, 20) and Jerome - Whistler, correspondence, 1876 (0-1, 15). Jerome was fortunate to gain a half point from the two games. The December, 1876 issue of  American Chess Journal, commenting on the Whistler game, recommended 8.Qd5+ without analysis.

Then 8.Qd5+ practically disappeared from the face of the earth. Perhaps it should be referred to as Jerome Gambit Secrets #12 ?

I have in my notes something from a 2020 1 0 bullet game at
lichess.org - the players are not named

If played correctly 7.Qxe5 Qe7 8.Qd5 + Kg7 9.d4 Bb4 + 10.c3 Nf6 11.Qe5 Bd6 12.Qxe7 + Bxe7 13.f3 White plays one piece, and for the second they have more space and two pawns - you can still play very much (position on the interactive whiteboard), both on the one and the other side. Although, objectively speaking, Black has a win, but you need to make fairly accurate moves, let's recall the game of the unforgettable Mikhail Tal - how many could hold their position after the Tal victims, who turned out to be objectively won in the home analysis?

It is not every day that you see the Jerome Gambit and the Magician from Riga mentioned in the same paragraph! 

Of course, Tal has already been mentioned on the blog: see "The Evans-Jerome Gambit Returns (Part 1)" and "Correctness".

8...Kf8 

If, instead, 8...Kg7, the game continued 9.d4 Bb6 10. Bg5 Nf6 11.e5 Nxd5 White resigned, levigun - obviously, GameKnot.com, 2004. This is the only other 8.Qd5+ game in The Database.

9.O-O c6 10.Qc4 d5 11.exd5 cxd5 

12.Qf4+ 

White declines the pawn (12.Qxd5), as Black could then develop his Bishop or Knight, attacking the Queen with tempo. 

12...Nf6 13.d4 Bd6 14.Qd2 Qe4 

This looks a bit odd, although Black retains his advantage.

Given that this was a 5-minute blitz game, and taking into consideration Black's previous move, perhaps he was planning to set up the Bishop + Queen battery, but at the last second, noticed that 14...Qe5 would drop Her Majesty?

15.Re1 Qh4 16.Qh6+ Qxh6 17. Bxh6+ Kf7 18. Nc3 Ng4 

ayushsankar1006 continues to press his attack, even with Queens off of the board. Both players now ignore the pawn at h2 for a short while.

19.Bg5 Be6 20.Nb5  

Perhaps planning to allow the h-pawn capture, and then trap the Bishop with g2-g3, while limiting where the prelate could otherwise retreat to? Or, was the clock ticking? In any event, 20.h3 was probably the move to make. 

20...Bxh2+ 21.Kh1 a6 

The fly in the ointment. The Knight is invited to leave.

22.Nc7

Likely the clock.

22...Bxc7 23.f3 White resigned

BSJG: Why?

The Blackburne Shilling Gambit - 1.e4 e5 2.Nf3 Nc6 3.Bc4 Nd4!? - has one really risky response from White - 4.Nxe5?! - but several very acceptable responses for the first player: 4.Nxd4, 4.d3, 4.0-0 and 4.c3. So, why engage with 4.Bxf7!?, the Blackburne Shilling Jerome Gambit? As we have seen recently, the move can be a surprising, unsettling challenge to Black, who had high hopes to be the one attacking.

The following game, played online at GameKnot.com, and recently retrieved, is a brutal example of White quickly wresting the game from his opponent.


xuam - maiden_coritiba
League division D2
GameKnot.com, 2014

1.e4 e5 2.Nf3 Nc6 3.Bc4 Nd4 4.Bxf7+ 

I have long referred to this as the Blackburne Shilling Jerome Gambit, with a knowing smile. Although the Bishop sacrifice - often followed by a Queen move - is typical Jerome play, there is actually no evidence that Alonzo Wheeler Jerome actually played it. On the other hand, there is no evidence (beyond assumption) that Joseph Henry Blackburne played "his" Shilling Gambit, as well.

4...Kxf7 5.Nxe5+ Ke6 6.Qh5 

White shows an interest in finishing off Black's King, quickly, even if the thematic move is risky.

It is interesting to note that "book" is 6.c3 Kxe5 7.cxd4+ Ke6 
with a balanced game, although Black's King has to be feeling uncomfortable. Indeed, White has scored 12 - 3 in games in The Database. Stockfish 9 recommends the followup 8.O-O Kf7 9.Qf3+ Qf6 10.Qb3+ Qe6 11.Qf3+ Qf6 with a draw by repetition.

6...Nxc2+ 

It is tempting to go after White's Rook at a1, and this move has been a bit more successful for the second player, who has scored 19 - 17 - 1 in The Database. The "boring" but upright alternative is 6...Nf6!?.

7.Kd1 Nxa1 

Black has the Rook. One indication that he may have been hasty in doing so (a "deep sacrifice" by White, that often occurs in the BSJG) is that The Database shows Black's results dipping a bit, to 12 - 15 - 1.

Another indication is that Stockfish 9 now rates White as being almost a Rook ahead.

(Take a second to ponder the difference between club play, as represented by The Database, and "objective reality", as represented by the computer.)

8.Qf5+ 

Oh, dear. Given that Black's "best" response is now 8...Kd6 9.Nf7+, giving up his Queen, White has been amply rewared for his aggression.  

8...Ke7 

Find the checkmate.

9.Qf7+ Kd6 10.Nc4+ Kc5 11.b4+ 

Or 11.Qd5+ Kb4 12.a3+ Ka4 13.b3+ Nxb3 14.Nc3#

11...Kb5 12.Nc3+ Kxb4 13.Ba3 checkmate

Jerome Gambit: GameKnot Visit

Every once in a while I stop by the online site GameKnot.com, to see what Jerome Gambit games I can pick up. Here are a couple of quick ones.

gocubs2004 - maleeniets
chrusage's mini-tournament IIII
GameKnot.com, 2005

1.e4 e5 2.Nf3 Nc6 3.Bc4 Nd4 

The Blackburne Shilling Gambit. Tricky - but White does well as long as he avoids the temptation of 4.Nxe5?!.

4.Bxf7+

The Blackburne Shilling Jerome Gambit. It is quite effective when it comes as a surprise.

4...Kxf7 5.Nxe5+ Ke6 6.c3 Nb5

This is a bit unusual ("book" is 6...Kxe5), but there are 16 previous examples in The Database, where White scores 13 - 3. Since Black's position remains "objectively" better, I suspect the lopsided score in favor of the first player is due to the second players' unfamiliarity with the BSJG.

I think White, in this game, suspects his opponent is adrift, and goes for the jugular.

7.d4 Nd6 8.d5+ Kxe5 9.Qd4 checkmate

Ouch.

byzantine1453 - x-filez
Let's Play Chess
GameKnot.com, 2018

1.e4 e5 2.Nf3 Nc6 3.Bc4 Nd4 4.Bxf7+ 

Here we go again: the Blackburne Shilling Jerome Gambit.

4...Kxf7 5.Nxe5+ Ke6 6.Nc4

Again, this is not main line, but there are 167 games with it in The Database, with White scoring 54%.

6...Nf6 7.c3 Nc6 8.d4 d5 9.e5 dxc4 10.exf6 Kxf6 

Black's King looks a bit exposed, but, at this point, it is on White's shoulders to prove compensation for the sacrificed material.

11.Qf3+ Kg6 12.O-O Bd6 13.Nd2 Rf8 14.Qd5 

The Queen seeks sancturary from the attack by the Rook, but this move brings her into even greater danger.

14...Bxh2+ 15.Kxh2 Qxd5 White resigned

Double ouch.

Jerome Gambit Secrets #4

One of my favorite Jerome Gambit "secrets" has actually been solved, but the story is always a good one to tell. And tell again.

Let's look at a line.

1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+

4...Kxf7 5.Nxe5+ Kf8

About 4 1/2 years ago I posted about this move

As early as his first article with analysis (Dubuque Chess Journal 4/1874), Alonzo Wheeler Jerome considered the possibility that Black might refuse to capture the second piece, and play for King safety instead with 5...Kf8.  

This was, in fact, the defense that Jerome, himself, credited to G. J. Dougherty, ("a strong amateur, against whom I first played the opening") of Mineola, New York, in a yet unfound game; that O.A. Brownson, editor of the Dubuque Chess Journal, played against Jerome in an 1875 game (Dubuque Chess Journal3/1875); that magazine editor William Hallock used against D.P. Norton in an 1876 correspondence game played “by special request” to test the gambit (American Chess Journal 2/1877); that William Carrington tried in his 1876 match vs Mexican Champion Andres Clemente Vazquez (Algunas Partidas de Ajedrez Jugadas en Mexico, 1879); and which Lt. Soren Anton Sorensen recommended as “more solid and easier to manage” in his seminal Jerome Gambit essay (Nordisk Skaktidende 5/1877). 

It is interesting that early in Jerome's Gambit's life, there were players willing to accept one "gift" but who were skeptical of accepting two "gifts".

6.Qh5

This move shows up in 38 games in The Database, with White scoring 49%.

As I noted

White also has the option of playing 6.Qh5, the Banks Variation, as in Banks - Rees, Halesowen, 2003, when Black can transpose with 6…Nxe5  as recommended by the American Chess Journal, (3/1877) - "The continuation adopted by Jerome, Qh5 looks promising." 

Pete Banks ("blackburne" online), a stalwart member of the Jerome GambitGemeinde (and still the strongest player I know who has played the Jerome regularly over-the-board in rated contests), brought international attention to Alonzo Wheeler Jerome's invention by writing to International Master Gary Lane, who commented at length on the opening, and on a couple of Banks' games, in his March ("The Good Old Days") and April ("Chess Made Easy") 2008 "Opening Lanes" columns at ChessCafe.com. IM Lane also mentioned one of Banks' games in his The Greatest Ever chess tricks and traps (2008), which reprised some of the earlier material. 

It is humorous to note that in his "Opening Lanes" column Lane wrote, after 5.Nxe5+, "I think anyone with good manners playing Black would now kindly ask their opponent if they wanted to take their move back" while in his book he changed this to "I think anyone with good manners playing Black would now go to another room to carry on laughing." 

Apropos the Banks Variation itself (i.e. playing 6.Qh5 in response to 5...Kf8), IM Lane noted in "The Good Old Days" that "6...Qe7 is a good alternative [to 6...Qf6 of Banks - Rees], because it stops the checkmate and protects the bishop on c5." 

A few months later, 6...Qe7 was tested successfully in a GameKnot.com game, splott - mika76, 2008: 1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+ Kxf7 5.Nxe5+ Kf8 6.Qh5 Qe7 7.Ng6+ hxg6 8.Qxh8 Qxe4+ 9.Kf1 Qd4 10.Ke1 Qxf2+ 11.Kd1 d6 12.h3 Qxg2 13.Re1 Qf3+ 14.Re2 Bf2 15.d3 Nd4 16.Nc3 Qh1+ 17.Kd2 Nf3 checkmate. Clearly White, the very-slightly-higher rated player, was taken aback by the move. I asked mika76 if he had been influenced by IM Lane's recommendation, but he said he had come up with the move himself.

Jerome Gambit: A Mockery of Common Sense

I really enjoyed the game in the previous post, so I looked in The Database for some more games by the Jerome Gambiteer obviously. The next game is even more wild, as White takes even greater risks - and wins.

obviously - grobnic
GameKnot.com, 2004

1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+

4...Kxf7 5.Nxe5+ Nxe5 6.Qh5+ g6 

This can lead to the infamous Blackburne Defense (7...d6), or the much-less-known but much-more-dangerous Whistler's Defense (7...Qe7).

7.Qxe5 Qe7 8.Qxh8

The Database is notoriously lean when it comes to either early Jerome Gambits, or over-the-board games. Coverage of Whistler's Defense is a good example.

There are 3 early correspondence games by Alonzo Wheeler Jerome (1876, 1879) where he scored 0-2-1.

Then, over 100 years later, there are then a handful of online games at FICS: 1 in 1999 (0-1); 1 in 2001 (1-0); 3 in 2003 (1-1-1); and then 10 in 2004 (4-5-1) - 7 of which involved obviously.

So, White's 8th move in the current game (capturing the Rook) probably deserves the same annotation it garnered in The American Chess Journal, June 15, 1876, when it showed up Jerome - Norton, correspondence, 1876 -

Played experimentally; and analysis will probably show the capture to be unsound. 

Analysis has since shown the capture of the Rook to be unsound - but, then again, analysis, even back in the 1870s, showed the Jerome Gambit itself to be unsound...

We must remember, however, that "all is new that has been forgotten", and that both players in the current game probably had little information on either the Whistler or the Jerome to guide them. That said, obviously obviously had some knowledge of the opening, as he scored 4-0-1 with the Black pieces, and 2-0 with the White pieces.

For that matter, The Database contains 59 games with the Whistler Defense, and White scores 57% - an indication that things are complicated enough that the first player will have his chances, at least in club play.

8...Qxe4+

This is the way to show that White's King is in more danger than Black's.

Bill Wall, who seems to be able to get away with just about anything in the Jerome Gambit, once experienced 8...Qf6 9.Qxh7+ Kf8 10.O-O Black resigned, Wall,B - Sepoli, Chess.com, 2010.

9.Kd1

This move is about as strong as 9.Kf1, e.g. 9...Nf6 10.d3 (10.Nc3 Qxc2 11.h4 b6 12.Qd8 Ba6+ White resigned, grobnic - obviously, GameKnot.com, 2004) 10...Qd4 (10...Qf5 11.f3 Bf8 12.Nc3 d5 13.h4 Qd7 14.h5 Bg7 15.hxg6+ Kxg6 16.Rh6+ Bxh6 17.Bxh6 b6 18.Ne2 Qf7 19.Qf8 Qxf8 20.Bxf8 drawn Jerome,A - Norton,D, correspondence, 1876) 11.Be3 Qxb2 12.Bxc5 Qxa1?! (12...d6 13.Ke2 Qxc2+ 14.Nd2 Bg4+ White resigned, flatchio - obviously, GameKnot.com, 2004) 13.Qf8+ Ke6 14.Qe7+ Kf5 15.Ke2 Qe5+ 16.Qxe5+ Kxe5 17.Re1 d6 18.Kf3+ Kf5 19.Bd4 Kg5 20.h4+ Kf5 21.Re7 Nd5 22.g4 checkmate, Wall,B - Neilson,C, Melbourne, FL 2017.

9...Qg4+

This move is good, but perhaps not quite as strong as 9...Qxg2, e.g. 10.Re1 (10.Qxh7+ Kf8 11.Re1 d5 12.d4 Bg4+ 13.Kd2 Qxf2+ 14.Re2 Qxe2+ 15.Kc3 Qc4+ 16.Kd2 Qxd4+ 17.Ke1 Qd1 checkmate, blackburne - perrypawnpusher, Jerome Gambit thematic, ChessWorld.net, 2008) 10...Qf3+ (going for a repetition in a winning position) 11.Re2 Qh1+ 12.Re1 Qf3+ 13.Re2 Qh1+ 14.Re1 Qf3+ drawn, Wall,B - Mathieubuntu, 40 0, FICS, 2011.

Black risked most of his advantage with the reasonable precaution 9...Nf6, although he was still able to outplay his opponent: 10.f3 Qh4 11.Nc3 b5 12.Qd8 Qg5 13.Ne4 Nxe4 14.Qxg5 Nxg5 15.d4 Be7 16.d5 h5 17.h4 Nh7 18.Bf4 d6 19.Ke2 Nf6 20.Rhd1 Bb7 21.Bg5 Nxd5 22.Bxe7 Nxe7 23.g4 hxg4 24.fxg4 Rh8 25.Rf1+ Ke8 26.h5 gxh5 27.g5 Rg8 28.Rg1 Bc8 29.Kd2 Bg4 30.Rg3 Rxg5 31.Re1 Kd7 32.Rh1 a5 33.Re1 h4 34.Rge3 Nd5 35.Re4 h3 36.R4e3 Nxe3 White resigned, vallabhan - obviously, GameKnot.com, 2004

10.Ke1

White improves on the historical 10.f3 Qxg2 11.Qxh7+ Kf8 12.Re1 d5 13.Qh4 Qxf3+ 14.Re2 Bg4 15.Nc3 Bf2 White resigned, Jerome,A - Whistler,G, correspondence, 1876.

10...Qxg2 11.Qxh7+ Kf8 12.Rf1 Qe4+

Again, good, but not best (12...d5). It is still hard to see how White will survive, but he does have a material advantage...

13.Kd1 d5

Chess can be incredibly cruel.

White's defense here has in good part consisted of shuttling his King back and forth. Black now plays the move recommended in the last note - and it turns over the advantage to his opponent.

It appears he needed to play 13...Qf3+ first. What difference does the check make? White shows with his next move.

14.f3 

In what appears to be a mockery of common sense, Stockfish 8 now claims that White has a clear advantage, giving 14...Qf4 15.Nc3 c6 16.Ne2 Qf6 17.d3 Bf5 18.Ng3 Re8 19.Nxf5 Qxf5 20.Bd2 Bd4 21.Kc1 Re7 22.Qh4 c5 23.a4 Re2 24.Kd1 Rg2 25.Qd8+ Kg7 26.Kc1 Bxb2+ 27.Kxb2 Rxd2 28.Qc7+ Kh8 29.Qxc5 Qxd3 30.Rf2.

14...Bg4 

It is difficult to explain this move. It is tempting to suggest a mouse slip, but the time control for this game was 5 days per move.

15.fxg4+ Nf6 Black resigned

After 16.Rxf6+ Ke8 17.Nc3!? White's pieces will be able to protect his King from too many Queen checks, while Black's King will be open to great dangers.

Jerome Gambit: Thematic and Aggressive

The following game shows how gambiteers would love the Jerome Gambit to proceed. White's play is thematic and aggressive, while Black's errors tend to be reasonable-looking moves that do not stand up to the situation they are played in. (I should point out that the time limit for the game was several days per move.)

obviously - vallabhan
GameKnot.com, 2004

1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+

4...Kxf7 5.Nxe5+ Nxe5 6.Qh5+ Ke6 7.Qf5+ Kd6 8.f4 Ng6

Elsewhere I have referred to this move as "somewhat inexact", as it saves one threatened piece while allowing another piece to go by the board. See "Reliable".

Stronger alternatives are the proactive 8...Kc6 and the counterattacking 8...Qh4+.

9.Qd5+ Ke7 10.Qxc5+ d6 11.Qf2

Putting the Queen on the f-file, where it can be backed up by a Rook on f1, and threaten the enemy King. It also increases the tactical complexity of the position by not guarding the pawn on e4.

11...Nf6 12.f5

I am amazed that this move hasn't been repeated, according to The (55,560 games) Database, as it is the essence of Jerome-ness.

The alternate move 12.d3 has 4 wins and 3 losses.

12...Ne5

If Black plays the logical 12...Nxe4, then Stockfish 8's main line recommendation is 13.Qf3 Ng5 14.Qg3 Ne4 15.Qf3 with a draw by repetition - which may make sense to a computer, but probably not to a human.

I would be more inclined to answer 12...Nxe4 with 13.Qe2, when after 13...Bxf5 14.O-O Qd7 15.d3 Rae8 16.dxe4 Kd8 17.Nc3 Be6 the game is even, but not over.

13.d4 Nc6

This removes the Knight from danger, and puts pressure on White's d4 pawn. Nonetheless, cranky Stockfish 8 prefers 13...Nf7. (That move seems a bit defensive, and I am not sure that a human would be thinking "defense" a piece up - but maybe he should.)

14.Nc3 Re8 15.Bg5 Kf8 16. O-O-O h6 17. Bh4 Bd7

White's moves flow, and the position screams "pawn storm on the Kingside". It doesn't happen, but only because White breaks up Black's Kingside instead.

18.Rhe1 Qc8

Getting out of the pin of the Knight on f6, and getting off the possibly soon-to-be hot d-file; but Black needed to do something drastic, like 18...Kg8 19.Nd5 Rf8 20.e5 Nxd5!? 21.Bxd8 Rxf5 22.Qg3 Nxd8 23.exd6 cxd6 24.Qxd6 Bc6, when he would have three pieces for his Queen, but White should still be better. (Yes, I had help in figuring this out.) 

19.Bxf6 gxf6 20.Qh4 Kg7 21.Nd5 Qd8

Black is in a lot of trouble on the Kingside. The better way of defending his f-pawn, 21...Rf8, still leads to problems, for example 22.Qg4+ Kh8 23.Qg6 Bxf5!? 24.exf5 Qd7 25.Qxh6+ Qh7. Returning the extra piece has allowed Black's Queen to participate in the defense of his King, but White has two extra pawns - and his Rooks are bound to cause trouble as they move up and over to the g- and h-files.

22.Re3 Bxf5

Yes, White's e-pawn cannot capture the Bishop without giving up the Rook, but that was never the  plan.

23.Rg3+ Kh8 24.Qxh6+ Bh7 25.Qg7 checkmate