Lesson 3 of 8
The knight's jump: one step and two steps
The hardest piece to visualize, tamed with one exact rule, a free color filter and two mental images.
Practise it now, free
The problem to solve
The knight is THE hardest piece to visualize blindfolded. Unlike the rook (straight lines) or the bishop (straight diagonals), its jump is "circular" and has to be reconstructed mentally. Typical questions, eyes closed: a knight on d4, does it attack f5? And e6? And d6? Which squares does a knight on g1 control?
Players agree: together with the long diagonals, the knight is the biggest obstacle in visualization. Good news: there is a very simple exact rule, an unbeatable color filter, and several mental images that make the 8 squares obvious.
The core: one step of 1 and one step of 2
Nothing in between matters
The knight is the only piece that jumps: what stands between it and its target is irrelevant, and no blindfold player should ever check it. Only the landing square matters, and only for a legal move: it must be on the board and not hold one of your own pieces.
The knight moves in an L: two squares one way, one square perpendicular. Translated into gaps:
The knight test
A knight on X attacks square Y if and only if the two gaps (files, ranks) are 1 and 2, in either order. It is the exact counterpart of the bishop rule: bishop = file gap equals rank gap; knight = one gap of 1 and one gap of 2.
Nd4 attacks f5?
d4 = file 4, rank 4. f5 = file 6, rank 5. File gap 2, rank gap 1. That is 1 and 2, so YES.
Counter-examples from d4
d6: gaps 0 and 2, not 1-and-2, NO. e5 (the adjacent diagonal square): gaps 1 and 1, NO. Classic trap: the diagonal neighbor is never a knight jump.
The phrase to hammer in: "one and two." If the two gaps are not 1 and 2, it is not a knight move.
The color filter: free, and an unbeatable check
The knight changes color on EVERY jump. From a light square it always lands on a dark square, and vice versa. The exact opposite of the bishop (which never leaves its color). Two uses:
- Instant filter: if the target square is the SAME color as the knight, the answer is NO, no calculation needed.
- Error check: if your calculation lands on a same-color square, you made a mistake. Recount.
Prerequisite: reading the color of a square instantly (square colors lesson).
Mental image 1: the 2x3 rectangle
The most popular trick for "seeing" the jump without counting: the knight stands at one corner of a 2 by 3 rectangle and jumps to the diagonally opposite corner. From a central square you can lay this rectangle in 8 ways (one per quadrant, and in each quadrant the long side either up or across), which is exactly the 8 landing squares. Visualizing the small rectangle is much faster than counting "two then one".
Mental image 2: the ring (8-point star)
The 8 landing squares form a ring (a circle, or an 8-point star) around the knight, jumping over the 8 immediate neighbor squares. With training, these 8 squares "pop out" by themselves when you look at a knight, no calculation. That is the end goal: shape recognition replaces arithmetic.
Clock variant: the 8 landings sit at 1, 2, 4, 5, 7, 8, 10 and 11 o'clock. You skip 12, 3, 6 and 9 (the orthogonal neighbors) and the 4 adjacent diagonals.
Mobility by square (bounds the search)
- Center (d4, e4, d5, e5...): 8 jumps, maximum mobility.
- Edge: 3 or 4 jumps (half the ring falls off the board, 3 next to a corner).
- Corner (a1, h8...): only 2 jumps. Hence the saying "a knight on the rim is dim".
Knowing this stops you hallucinating off-board squares and speeds up enumeration near the edge.
Try it: the live board
Move the knight, test squares, and watch the 2x3 rectangle draw itself. Then run the yes/no quiz on a streak.
Why it matters
The test runs both ways
Knight attacks are symmetric: if a knight on d4 hits f5, a knight on f5 hits d4. So to ask "which knight could hit my queen on e6?", build the same ring around e6 and see which of those squares holds an enemy knight. Same work, the other way round.
- Following a game in your head: the knight is the piece that most often makes people "drop" the position. Master it and the rest opens up.
- Seeing forks: most points lost in blitz come from unseen knight forks. Visualizing knight jumps means anticipating forks (king + queen, king + rook, the family fork).
- Calculation: lines with knight jumps cause the most calculation slips; training removes them.
Recommended training progression
- The "one and two" rule calmly, on pairs of squares.
- The color filter: announce the landing color before searching (always the opposite).
- Enumerate the 8 jumps from a central square, then an edge, then a corner. The whole ladder: 8 in the middle 4x4 (c3 to f6), 6 on the ring around it, 4 on the edge and on b2, g2, b7, g7, 3 beside a corner (b1, a2...), 2 in the corner.
- Timed YES/NO test: "does Nd4 attack X?" in a loop.
- Ghost knight / journeys: follow a knight over several jumps without seeing it, reach a target. The invisible-knight exercise is the most effective for mental tracking.
- Interleaving: alternate knight, bishop and color drills.
In-app drill
/train/knight-maze for knight journeys, plus /train/square-colors to keep the filter sharp.
Common mistakes
- Confusing the jump with the adjacent diagonal square. e5 from d4 has two gaps of 1: NOT a jump (you need 1 and 2).
- Forgetting the color change. A knight never stays on its color; use that to verify every result.
- Looking for 8 squares at the edge. Four there, three beside a corner (b1, g1, a2, h2...), two in the corner itself. Do not invent off-board squares.
- Counting "two then one" forever. Move to the 2x3 rectangle or the ring as fast as possible; mechanical counting is a starter crutch.