Blackwood, and Roman Key Card

What it is actually for

Blackwood is not a slam try. It does not ask "shall we bid a slam", and it must never be used to find out.

By the time you bid 4NT you should already have decided that twelve tricks are there — provided you are not missing two cashable aces.

That is the whole job: stopping you bidding a slam off the top two tricks. Whether the hands fit, whether the suits break, whether the tricks are there — settle all of that before you ask, because the answer speaks to none of it.

When not to ask — the part that matters

Almost every Blackwood disaster comes from asking on a hand where the answer cannot help. Four cases:

• You have a void. Blackwood counts aces and a void is as good as one — but you cannot tell partner about it, and worse, you might be counting an ace opposite your own void, which takes no trick at all. Cue-bid instead.
• You have two quick losers in an unbid suit. Partner can hold two aces and you can still be off ♣ A K. Blackwood counts aces; it does not count losers.
• No trump suit is agreed. Then 4NT is almost certainly quantitative — see below, and partner will pass it.
• You cannot stand the answer. If a reply of one ace leaves you at the five level in a contract you did not want, do not ask. This is the discipline nobody teaches: bidding 4NT commits you to at least the five level whatever comes back.

Roman Key Card Blackwood

Ordinary Blackwood has an obvious hole. The king of trumps is worth about as much as an ace, the queen of trumps often decides a slam, and neither is mentioned.

Roman Key Card fixes both. It counts five key cards — the four aces plus the trump king, and folds the trump queen into the answer.

The common version is called 1430, after its first two responses:

• 5♣ — 1 or 4 key cards.
• 5♦ — 0 or 3 key cards.
• 5♥ — 2 or 5, without the trump queen.
• 5♠ — 2 or 5, with the trump queen.

The pairs, 1 or 4, 0 or 3, are never genuinely ambiguous: the asker has enough information to know which, because a partner who could hold four key cards would have done something other than raise quietly.

The other version, 0314, swaps the first two responses so 5♣ is 0 or 3. Both are widely played, and they are exactly opposite.

This is the most dangerous disagreement in bridge. Two players on different versions bid a grand slam missing an ace, and the auction looks sensible to both of them until dummy comes down.

After the answer

• 5NT — asks for kings, and by bidding it you have promised the partnership holds all the key cards. It is a grand slam try, so do not bid it to satisfy curiosity.
• The cheapest bid that is not the trump suit — the queen ask, when the answer did not settle it. Responder returns to the trump suit without the queen and bids something else with it.
• Five of the trump suit — the signoff. It means the partnership is missing too much, and it is the reason you should not have asked unless you could stand this answer.

Gerber: the same question over notrump

4♣ asks for aces, and it is used only directly over a natural notrump bid — because everywhere else 4♣ is needed as a genuine bid or a cue.

• 4♦ = 0 or 4 aces
• 4♥ = 1
• 4♠ = 2
• 4NT = 3

It exists because the ace-ask has to stay below game. Asking at 4♣ leaves room to stop in 4NT; asking at 4NT does not.

The universal rule is "Gerber only over notrump". The hands that go wrong are the ones where somebody bid it over a suit and partner read it as clubs.

What good players do instead

Strong partnerships use Blackwood less than beginners expect.

The alternative is cue-bidding. Once trumps are agreed and the partnership is past game, a new suit shows a control there — an ace or void first time round, a king or singleton later.

The two hands then trade information about where their strength is rather than how much of it there is.

That solves what Blackwood cannot: it locates the losers, and it handles voids. Blackwood is the last step, once only the ace question remains.