What Is a Proof Without Words? Seeing Why Something Must Be True

A proof without words is a picture arranged so that the truth of a statement follows from what the reader sees, generally by showing that two quantities are equal because they count or fill the same thing. The classic case establishes that the sum of the first several odd numbers equals a square, by arranging the odd numbers as nested L-shapes that assemble into a square, at which point the result is not merely illustrated but demonstrated, since the arrangement works for any size and the reader can see why. Similar diagrams establish results about triangular numbers, about sums of series, about inequalities between different kinds of average, and famously about the relationship between the sides of a right-angled triangle, for which dozens of distinct diagrammatic proofs exist.

The effectiveness comes from properties of diagrams that text lacks:

• •A diagram presents relationships simultaneously rather than in sequence, so the structure is available at once
• •Spatial reasoning is fast and reliable for questions about area, fit and counting
• •A well-chosen figure removes the need to track symbols, which is where errors in algebraic manipulation occur
• •Generality can be visible, since a construction that evidently extends to any size proves the general case
• •The reason is exposed rather than only the conclusion, which is what distinguishes understanding from verification
• •They are memorable, which matters more than it sounds, since a proof that can be reconstructed is worth more than one that can only be looked up

Diagrams can mislead and the failures are instructive. Figures drawn to a particular case may not generalise, and a diagram assuming a configuration that does not always hold proves nothing about the cases it excludes. Famous fallacious proofs exploit exactly this, including a well-known figure that appears to show all triangles are isosceles, which works only because the diagram is drawn inaccurately and the argument fails when the construction is done correctly. Dissection puzzles that appear to show a square rearranged into a rectangle of different area rely on imperceptible gaps. The general point is that a diagram is an argument with unstated assumptions about configuration, and checking that those assumptions hold in every case is the work that a rigorous proof does explicitly and that a picture can hide.

The genre has a substantial literature and the classic examples are worth knowing. Summing consecutive whole numbers becomes obvious when two copies of a triangular arrangement are fitted into a rectangle. The sum of an infinite geometric series with a ratio of one half is shown by repeatedly halving a square until the whole is filled. The relationship between the arithmetic and geometric means follows from a construction in a semicircle. Sums of cubes equalling the square of a sum appear as a nested arrangement of squares. Several of these were known in ancient Greek, Indian, Chinese and Islamic mathematics independently, which is unsurprising given that the reasoning is visual rather than notational and does not depend on a particular algebraic language. Collections of such figures have been published as books, and the genre has a regular presence in mathematical magazines.

The existence of these diagrams sits inside a long-running question about what mathematical proof is supposed to accomplish. On one account a proof establishes truth, and any correct argument serves equally regardless of whether anyone understands why. On another, a proof should explain, showing not only that a statement holds but why, and on that view a diagram exposing the structure is superior to a symbolic manipulation that verifies without illuminating. Mathematicians in practice value both and distinguish them openly, describing some proofs as merely correct and others as explaining what is going on, which is a judgement about mathematical practice rather than about logic. Computer-assisted proofs sharpened the question, since a verification too long for anyone to read establishes truth while supplying no understanding at all, and that has been argued about since the first such result.

A diagram arranged so that the reason is visible, most famously showing that consecutive odd numbers assemble into a square. Simultaneous presentation and reliable spatial reasoning are why they work, and visible generality is what makes them proofs rather than illustrations. They mislead where a figure assumes a configuration that does not always hold, which is exactly what rigorous proof states explicitly.