The chessboard that gains a square

2026-05-08

Take an 8 × 8 chessboard - 64 cells. Cut it into four pieces along three straight lines. Rearrange the pieces into a 5 × 13 rectangle. Count the cells: 65. The pieces did not change. One extra cell appeared from nowhere.

Press play. The four pieces lift off the square and slot into the rectangle - same shapes, same areas, no trickery you can see. The readout climbs from 64 to 65 as they land. Press play again to send them back.

Where the extra cell hides

The pieces are honest. The cell is too. The lie is in the diagonal. Tap ? to draw it: a thin parallelogram running corner to corner of the rectangle, exactly one cell of area, between the triangles' hypotenuses and the trapezoids' slanted edges.

The triangles' hypotenuses have slope 3⁄8 = 0.375. The trapezoids' slanted sides have slope 2⁄5 = 0.400. The rectangle's true diagonal has slope 5⁄13 ≈ 0.385. Three different lines, all close enough that the eye reads them as one. The "diagonal" of the rectangle is not a line at all - it is a sliver, and the missing cell lives inside it.

Why these particular numbers

The puzzle does not work for arbitrary numbers. It works for 3, 5, 8, 13 - and only because those four numbers obey one identity:

5 · 13 − 8²  =  65 − 64  =  +1

Three, five, eight, thirteen. Each one is the sum of the two before. They are consecutive Fibonacci numbers, and that identity is no coincidence: it is true for every three Fibonacci neighbours. Cassini, who noticed it in 1680, wrote it like this:

F_n−1 · F_n+1  −  F_n²  =  (−1)ⁿ

The product of two Fibonacci neighbours is always one off the square of the Fibonacci between them. Sometimes one too many, sometimes one too few - the sign alternates. With 3, 5, 8 the square is one too many (the 5 × 8 rectangle would lose a cell). With 5, 8, 13 the square is one too few. The missing cell is built in.

So the puzzle is not a trick of geometry. It is a trick of Fibonacci. The slopes 3⁄8 and 2⁄5 are close because the ratio of consecutive Fibonacci numbers is creeping toward φ from both sides at once - and "close" is exactly what you need to fool a reader into believing the diagonal is straight.

Four pieces. One identity. The cell that wasn't there.