A King is not an Ace

Roman Keycard Blackwood (RKCB) is one of the world's most adopted conventions. It is at the end of an evolutionary branch, tracing an ancestor in Blackwood which asked for aces, through Roman Blackwood which identified aces, to keycards, the four aces and the trump king. Where 'ordinary' Blackwood aims to keep you out of a slam missing two aces, a keycard ask additionally avoids reaching the six-level missing an ace and the king of trumps.

Edgar Kaplan was a critic of RKCB noting that crucially, you had to know what were trumps and, targeting both the linguistic sloppiness of "five aces" and the semantics, that however much one might wish the king of trumps to be an ace, it remains that it isn't one. You cannot for example, securely bid a slam prepared to take a trump finesse. As proof, this was one of those 'slam hands' from the last county match, both vulnerable:

  • KQ108632
  • Q7
  • 10843
  • ---
  • J754
  • AK5
  • Q
  • AKJ73

South passed and West opened three spades. East took stock and ventured 4NT, RKCB. Learning of just one keycard opposite he nevertheless bid six spades, hoping it was the ace of trumps. A long time ago (I'd been turned away from a full cinema screening the first Superman film with Christopher Reeve) in a pub I made six diamonds in an 11-card fit missing king and queen. No such luck here; kings are not aces.

Matters at another table were more complicated by West opening four spades; now the prospect of East holding ace-queen-to-eight was too attractive. Both sequences are a salutary reminder that asking for aces – or keycards – is a brake on bidding, designed to reduce risk.

Back to the three-opener: many pairs use the cheapest minor – so here four clubs, but four diamonds over a three-club opener – to ask for aces. Those who are 'wide range' (undisciplined) with their pre-empts employ the first step as a "I refuse to tell you, my hand is that bad". This saves a level and leaves a direct 4NT for aces, in case responder has his own suit and that's what he wants to know about.

Alternative keycard asks can be useful but the day before my partner and I perpetrated this cautionary tale; none vulnerable:

  • 10
  • QJ542
  • 86
  • AK742
  • KQJ7652
  • A1093
  • A
  • 10
All Pass
  1. Good hand with heart support, intending to follow-up with 4NT
  2. Now that 4NT has been passed, a substitute keycard ask
  3. One or four keycards but…
  4. Interpreting response as zero or three…

My 5NT bid had been discussed as an idea and, when we gave explanations at the termination of the auction, I was pleased it had been recognised. But immediately it transpired we differed as to the responses – how? Because of other keycard asks, we play different responses to 4NT than to others. I carried across the same scheme to 5NT when it displaced 4NT. The Great Shuffler spared some of our blushes: after losing trick one to the spade ace, the heart finesse (into the pre-emptor!) lost as well, down two but the small slam was failing too.


I received correspondence about this topic before the newspaper went to print. That mainly concerned the wisdom of having separate response-schemes for different keycard asks. At the heart of this is the division between using '1430' responses and '3041'. It is undeniable that "the more you have, the safer are higher responses" is sound, why then use 5 as one keycard and 5 as none?

The answer lies in the utility of those responses, especially when hearts are trumps. The critical case is when a 4NT ask elicits a one keycard answer. If that reply is 5 then if asker knows there is one keycard missing and cannot see the trump queen, he will not be able to do anything about it. If instead the one-keycard answer is 5 he can proceed with 5 asking for the trump (heart) queen.

So to best use space, the '1430' scheme is good – for the majors at least (1430 is the score for a making slam in a major). For the minors, RKCB is pretty much off limits anyway because the 5 and 5 response both show two keycards. That means to ask in a minor you must hold at least two keycards and the trump queen (lest a 5 response leave you missing 1½ keycards). Lastly note that with hearts as trumps, you cannot ask for keycards with one keycard plus the trump queen.

Which brings us to having more economical keycard asks. My partnerships make use of Deadwood, where 4-minor sets that suit as trumps and the first step is "my hand is not good for slam". I wrote about this in three articles in 2015. Asking in four of the trump suit affords a luxurious five steps before committing to slam – compare the four steps over 4NT for spades and three for hearts. Thus there's no need to accommodate utility and 'cheaper = less' responses afford a sign-off in 4NT.

I am pleased to say that this 'two schemes' arrangement caused us no grief until the 5NT accident above. Before we leave that I confess that 5NT was 'aggressive', maybe reckless.

What would I have done over 6 (2 keycards + Q)? I would have to pass that or punt 7 (in similar fashion to hand 1). But at the time I really thought it was a 'six or seven' decision, not whether to stay low. Note that if 5NT – 6 shows one keycard, there is no ask for the trump queen, caught again in the 1430/3041 cleft. That still seems pertinent when you're a level higher but I certainly didn't reason that way at the table, I simply transferred the 4NT agreement, without regard to the wording.

Perhaps most prudent would have been to pass 5 (clearly forcing after my 4) and see if partner bid on (didn't double) after his non-forward-going 4.

Published Saturday 11.May.2019