Little review of Salvio gambit

Salvio (1570-1640) was a great chess player of his time:

https://en.wikipedia.org/wiki/Alessandro_Salvio

This is a link to one of earliest recorded games (Benavides-Saduleto, 1590) featuring Salvio gambit: http://www.chessgames.com/perl/chessgame?gid=1251891

N.B. 5. Ne5 is already the Salvio gambit. However, 5...Qh4+ 6. Kf1 is the unique best continuation. 

There are 4 main historical variations confirmed by engine evaluations.

• Viennese variation 6...Nc6 (ev=-1.6) --> post #3
• Cochrane gambit 6...f3 (ev=-1.1)
• Santa Maria defense 6...Nf6 (ev=-0.8)
• Silberschmidt defense 6...Nh6 (ev=-0.4) --> post #7, #9

It's possible also to play 6...Bg7 (no official name, ev=-0.5).

(to be continued, cf. posts #3, #7, #9)

7 ..... Nh6 seems the best to protect both f7 and g4, but ...for humans only, i guess 

Viennese variation 6...Nc6 is the best chance to refute the Salvio gambit. We have only 1 game in the local database: Weiss-Lenner (1934, 1-0)

https://www.chess.com/games/view/25076

SF9+ suggests the following possibilities: 

• 7. Nxf7 (ev=-1.2)
• 7. Bxf7+ (ev=-1.2)
• 7. d4 (ev=-1.5)

Let's choose 7. Bxf7+ as the most straightforward attack, it gives this variation

evaluated at about -1.6.

U are right, my friend: Silberschmidt variation 7...Nh6 is the most popular and appears in almost all historical games. 

ty i swear not to have used any help ! juste petites cellules grises !! the computers create some lines just close to  Magnus approaches, and the vulgum pecus (me) is just like the kulak before the knout! ty

topic .... 8 ....d7 open line for the Bishop/ Nf6 protect and attack ! / Bishop b4 threat e1 // all candidates,

                                why not ...Nc6xNe5, double pawns, less pressure ??? 

The most popular Silberschmidt defense 6...Nh6 actually gives white good equalizing chances. They are represented by 7. d4 (ev=-0.2), traditionally followed by so-called Anderssen counterattack 7...d6 (ev=-0.3):

An other good option would be delayed Cochrane gambit 7...f3 (ev=-0.3).

Hahaha ... simply giving away the knight ... U are kidding, bro!  Your McDonnell gambit 5. Nc3 is great too  but as risky as Salvio imho.  SF9+ evaluates it at about -1.1. Btw, in the McDonnell-d'Arblay game above, 6. Qxf3 (ev=-1.2) would be substantially better than that "unrisky" Muzio-like 6. O-O (ev=-2.2). 

The Silberschmidt defense: Anderssen counterattack normally leads to the following position:

There are the following good options for white:

• 9. Nf4 (ev=+0.0, d=30)
• 9. Nc3 (ev=-0.1, d=30)
• 9. g3 (ev=-0.3, d=30)

In historical games white usually played 9. g3:

Benavides-Saduleto (1590) http://www.chessgames.com/perl/chessgame?gid=1251891

Greco-NN (1620) http://www.chessgames.com/perl/chessgame?gid=1272506

Roussereau-NN (1680) http://www.chessgames.com/perl/chessgame?gid=1336254

for Benavides and Roussereau, I don't see clearly why white win, it seems that black have some moves to use to make it longer than 21 moves, (but i 'm weak), ty

Hmm ... indeed, there is no white win in Benavides-Saduleto.  On the contrary, the engine evaluation is something like -1. Maybe it was a time-out !? 

And 21...Nc4 leads to a draw in Roussereau-NN game. -

All right, Skye. I'll add some info on Santa Maria and Cochrane variations and then put everything in the first post. 

Yes, there were several interesting historical games with 5...Nh6 (ev=+0.3, d=32). However, this move is a mistake since 5...Qh4+ is substantially better (ev=-1.6, d=32). So, I analyze only 5...Qh4+ in this thread.