The unlikeliest move part 3 – the minor pieces

Last time we covered Queens, Rooks and Kings – and the unlikeliest possible move for each of them. We saw the superfluous three-queens position of “Qa1b2#”, the beautiful discovery of “Rbb2#” and the kingly “Kxa1#”. Three very possible checkmates, the first two of which have probably never been played in the billions of chess games in history.

Now it’s the turn of the lowly Pawns, Knights and Bishops. Strap yourself in, this is a long read.


“Pawns are the soul of chess” said the original chess professional Philidor. In this particular puzzle, they are a headache. They can move in so many more ways than pieces; firstly their captures are different to their normal moves (!?), even more strangely they can transform into any other piece (known as promotion), and finally due to a tweaking of the rules of chess around 500 years ago, the outrageous en passant had to be invented to stop pawns leaping past each other (note: I say outrageous as, when I taught somebody chess, forgot to mention en passant and then played it against them, they fell off their chair laughing at me for just making up rules on the spot). So we have many things to consider here – and in fact they all tie in together in a sort of logical conundrum – hence the headache.

Let’s start with capturing – and to increase unlikeliness, we may as well throw in a checkmate: dxe6#. Good start. One note here – writing “dxe6” means you captured a piece. If you write “de” then you have captured a pawn. Because you write “d” to denote a d-pawn means that you specify the pawn – or do you? What about doubled pawns? If you had, say a pawn on d5 and d2, both able to capture pawns on the c-file, what would you write? “d2c” seems a logical way to do it (see later discussion).

Okay, so how about promotion? Well if we go for this, we lose the ability to include multiple pawns –something like either “a8=B#” or with a capture “exd1=N#” can only ever be delivered by one pawn. Nevertheless, these are still very unlikely – and I’d suggest the unlikeliest would be:



Under-promotion with mate! (editors note – A1 being the top right square)

The rest of this section is more an argument about chess notation itself – so skip to Knights if you’re not a pedant like me!

Although I can find no official rules about this – it seems that whenever there is any ambiguity over which pawn has moved, you specify more about the square it ends up on, rather than its starting square. If this is true, then any multiple pawn implications are ruined. If not, then there is much more to pawns that we thought.

Consider “c2d#”. A c-pawn on the 2nd rank captures a d-pawn with checkmate. This implies another pawn on the c-file could have captured a d-file pawn with mate (else you would just write “cd#”).


One position where “c2d#” is possible (editors note – A1 is the bottom left corner)

Now – how about en passant? There is no specific notation for it – you just simply write the capture, for example “dc”. There could be multiple ways to take it though – for example you have pawns on d4 and d5, and your opponent plays c5 (starting from c7). What now? You have to specify which capture – so if it is en passant, it may simply be “dxc6”, and if it is the standard capture, “dxc5”. But now we reach a real nuance (as if this whole series wasn’t already!); would you actually write “dc6”, rather than “dxc6”? I can’t find anywhere which clarifies this; but all the chess programs opt for including the ‘x’. This not only makes this a dead end for our puzzle (as “dxc6” could also easily be a capture of a piece, and therefore doesn’t imply multiple pawns anymore) – but it is also fairly illogical. And we know how much this angers chess players. To me, it seems as though you should write “dc6#” for a pawn capture where another pawn could also capture from d-to-c-file with mate, which is possible like this:


One position where “dc6#” is possible (editors note – a8 is the top left corner)


The ‘Teaser’ in part 1 of this article used the ‘tricky beasts’ as its subject. “Nfxh8#” was the move, which implies a lot about the position. However, there is a slightly different option for knights – as with the ‘three-queens’ position, you can imply three knights as well (and they all have to be able to give checkmate) with: “Na3xc4#”


All three knights can give mate here with the same move (editors note – A1 is the bottom left  corner)

Fairly clearly, this means three knights on the board – which means at least one under-promotion. The debate over the unlikeliness of this is messy; in a serious game under-promotion is rarely better than queen-promotion, but in a casual game players may do this to be showy. We could get into a debate about whether you would ever reach a checkmate opportunity or whether players would resign beforehand, or whether notation would be used in casual games at all – but I don’t think that would help my headache…

I’ll leave it open to comment, but I am resisting the aesthetic pleasure of the discovered mates on h8, and choosing the more pragmatic “Na3xc4#”.


We have already discussed moves that imply promotions in Queens and Knights – but with Bishops it is surely more unlikely to have multiple bishops of the same colour square? By now you’ll be used to the themes, so I’ll get straight to the two options: Three bishops able to capture something with mate, or two bishops able to move to a square with (discovered) mate. Three cannot deliver discovered mate (believe me I tried).

In a recent local derby, my opponent and long-time Bristol player and coach Jerry came bounding up to me with the solution of three bishops. And indeed, this is a good contender. After some discussion we arrived at the particular: “Bc6xb7#”


Three bishops!? A strange game indeed (editors note – a8 is the top left corner)

The other option is two bishops from the same file able to give discovered mate: “B2xh3#”


This one requires discovered mate – also a very strange game (editors note – h1 is the bottom right corner)

Note – this should be more unlikely than something like “Bgh3#” because the bishops can be on more possible squares. With our solution, they must be on g2 and g4 – otherwise you would specify the file, not the rank.

Let’s recap on what this implies: Bishops on g2 and g4, the enemy king on g3, an enemy piece on h3 and rooks/queens able to deliver mate by discovery.

Candidate moves – recap

Curiously, it is queens and rooks that do not involve a capture – but for different reasons (with rooks we can imply discovered checks, whilst queens would have implied less specific squares for the king). All checkmates though – because why not?

“Kxa1#”, “Qa1b2#”, “Rbb2#”, “B2xh3#”, “Na3xc4#”, “bxa1=B#”.


The winner

Well – my winner anyway. I’m sure you have your own. After a long think, I went for “B2xh3#” as the unlikeliest move possible. Reasons below:

  • It implies an earlier under-promotion to a bishop (and – the difference in unlikeliness between 2 bishops on the same colour and 3 is much less than the difference between 1 and 2, I would argue)
  • It implies two possible discovered checkmates. It is not possible for the bishops themselves to deliver mate – because their move to h3 would not threaten any new squares than they were before.
  • The enemy king must be on exactly g3
  • An enemy piece must be on h3
  • There are still 5 squares around the king to cover (f2, f3, f4, h2 and h4).

The problem of unlikeliness

As mentioned – calculating the likelihood of having to notate each of the moves in this article series is both mind-boggling and also unprovable. The snag is that the question itself has human elements to it. You would have to weigh up questions of whether two players would ‘collude’ to play a strange game with under-promotions, or with mazy king-runs, eschewing easier checkmates to go for the fanciful finish. Maybe I can be satisfied with an unknowable problem – one which proves neither humans nor machines are not omniscient – rather than beating the machines outright. In any case, I will be happier knowing I can stop thinking about unlikely moves, recover from my headache and reboot for our last match of the season – where some fairly standard moves are likely to win the day (editors note – amen to that!).


Mike Harris

Mike is a regular pretender in Bristol’s top division and can also be seen propping up local tournament ladders. He writes a regular column for the Bristol Chess Times and plays a solid 20 openings a season.

Leave a Reply

%d bloggers like this: