Counting squares: the pawn endings I can now solve on the tram

1.Why pawn endings first

Every ending eventually turns into a pawn ending or is decided by the threat of turning into one. That is the argument, and I resisted it for two years on the grounds that these positions almost never appear on the board with the pieces gone. They do not need to. The evaluation of a middlegame trade, the decision about whether to swap the last rook, the choice of which pawn to push: all of it rests on whether the position two trades away is won, drawn or lost.

The second reason is more practical. Pawn endings are the only part of the game I can study without a board in front of me, because the positions are small enough to hold in the head. Twenty minutes on the tram twice a day is not a training programme, but over a winter it adds up to more hours than I would ever sit down and give it deliberately. A pocket set and a notebook fit in a coat pocket. A folder of opening files does not.

2.The square, and when it lies to you

The first thing anybody learns is the square: draw a box with the pawn and its promotion square as one side, and if the defending king can step inside it, the pawn is caught. It is a genuinely useful shortcut and it saves calculation under a clock. It also lies in three situations, and I got caught by all three before I wrote them on the inside cover of the notebook. The shortcut is a starting point for a count, not a substitute for one.

The first lie is the pawn still on its second rank, which can move two squares and therefore has a bigger box than it appears to have. The second is anything on the a-file or h-file, where the defending king can reach the corner and the attacking king cannot force it out. The third, and the one that cost me an actual half point, is a defending king inside the box but standing on a square where it has to move out of the way of its own king or blocked by its own pawn.

3.Corresponding squares without the jargon

The idea that took me longest is the one with the worst name. In a blocked position where both kings are shuffling, certain pairs of squares belong together: if my king stands on one of a pair and it is my opponent's move, he loses ground. Learning to see the pairs is a matter of working backwards from the squares that obviously matter, which are the ones next to the pawns. Once two or three pairs are mapped, the rest of the position tends to fall out of them.

In practice I do not map anything at the board. What I do is recognise that the position is one of the shuffling kind, count whose move it needs to be for the key square to fall, and then arrange to lose a move if I need to. Losing a move on purpose with a lone king is the whole technique, and it took a month of tram rides before it stopped feeling like a trick. A reader wrote to correct my wording on this section twice, and the second correction was right.

• Ask first whose move the key square needs.
• Work backwards from the squares beside the pawns.
• A rook-file pawn is a separate case, not an awkward version of this one.
• If the count says it is close, do not trust the shortcut.
• Write the answer in the notebook before moving the pieces.

4.Six positions at the back of the notebook

The back page has six positions on it, drawn by hand, and I can now set all six up from memory. Two are the basic king-and-pawn wins with the king in front and behind; one is the rook-file draw; one is the position where the defender has to choose which way to step and only one choice holds; one is a blocked shuffling position; and the last is a race where both sides promote and the resulting queen ending is not a draw. Six is the number that fits.

Whether this has helped is hard to separate from everything else I have changed, but one Wednesday in February I reached the choose-a-direction position with three minutes left, recognised it immediately, and drew a game I would have lost in 2024. That single half point is not evidence of much. What I can say with more confidence is that trades no longer frighten me the way they used to, because I can see one step further past them than I could a year ago.