Wall Square & 3-4-5 Diagonal
Diagonals for squaring a layout, and why the 3-4-5 triangle everyone uses is far too small to square a real wall.
Check diagonals I've already measured
Bigger triangle, better corner
The 3-4-5 is too small
Everyone knows the method; almost nobody knows how badly it scales. Read the diagonal ⅛″ out on a 3-4-5 in feet and the corner is 0.249° off — which throws the far end of a 40 ft wall out by more than two inches. Run the same ⅛″ error on a 12-16-20 and the wall is out by half an inch. Your tape doesn't get more accurate as the triangle grows; the triangle gets better at ignoring it.
The relationship is exact. The angular error is the hypotenuse times the measurement error, divided by the product of the legs — so scaling everything by k divides the error by k. Doubling the triangle exactly halves the error. Use the largest multiple the space allows, every time, and there's no reason to stop at 3-4-5 in a room that would take a 15-20-25.
And equal diagonals only prove a rectangle if the opposite sides already match. An isosceles trapezoid has two equal diagonals and is nothing like square — parallel sides of 8 and 14 with a height of 10 give diagonals of 14.866 apiece. Measure the sides in pairs first, then the diagonals. Checking diagonals alone on a frame nobody measured proves nothing at all.
- Move half the difference. If the diagonals differ by an inch, shift one corner half an inch toward the longer one. Moving the full amount sends the error the other way and starts a chase that never converges.
- The diagonals agreeing with each other matters more than either agreeing with the calculation, because a consistent error in both side measurements cancels out of the comparison.
- Keep the tape at the same tension every time. Hooking on a corner and pulling slack changes the reading by more than the accuracy the method is capable of.
- A tape run over an obstruction reads long by an amount nobody can estimate. On long diagonals, two people and a taut line beat one person and an optimistic reach.
How to use
- Enter the two sides to get the diagonal they should measure.
- Use the largest 3-4-5 multiple that fits — the table shows what each one buys.
- Measure the opposite sides in pairs before trusting any diagonal.
- If the diagonals differ, move one corner by HALF the difference.
Frequently asked questions
How does the 3-4-5 method work?
Measure three units along one wall, four along the other, and the diagonal between those two marks should be exactly five. Any consistent unit works, and any multiple of the triple works — 6-8-10, 9-12-15, 12-16-20. It is the oldest practical way to lay out a right angle, and it needs nothing but a tape.
Is the 3-4-5 triangle accurate enough?
Not for a wall of any length. Read the diagonal an eighth of an inch out on a 3-4-5 measured in feet and the corner is 0.249 degrees off, which throws the far end of a forty foot wall out by more than two inches. The method is sound; the size people use it at is the problem.
Why is a bigger triangle better?
Because the angular error is the hypotenuse times the measurement error, divided by the product of the legs — so scaling the whole triangle by a factor divides the error by that same factor. Doubling it exactly halves the error. Your tape does not become more accurate as the triangle grows; the triangle becomes better at ignoring the error you were always making.
What multiple should I use?
The largest that fits the space, always. A room that will take a 12-16-20 has no reason to be squared with a 3-4-5, and the larger triangle is four times more forgiving of the same misread. It costs nothing but a longer tape pull, and it is the single easiest accuracy improvement available on a layout.
Can I just check that the diagonals are equal?
Only after checking that the opposite sides match. Equal diagonals alone do not prove a rectangle — an isosceles trapezoid has two equal diagonals and is nothing like square. Parallel sides of eight and fourteen with a height of ten give diagonals of 14.866 apiece. Measure the sides in pairs first, then the diagonals.
My diagonals differ. How much do I move?
Half the difference, toward the longer diagonal. If they differ by an inch, shift one corner half an inch along the wall. Moving the full difference is the common mistake: it overshoots and sends the error the other way, which starts a chase back and forth that never quite converges.
Should the diagonal match my calculation exactly?
The two diagonals agreeing with each OTHER matters more than either agreeing with the calculated figure. A consistent error in both side measurements shifts the calculated diagonal but cancels out of the comparison between the two — so a frame whose diagonals match is square even if both differ slightly from the arithmetic.
Does tape tension matter?
More than the method’s own accuracy, which is why it is worth being deliberate about. Hooking a tape on a corner and pulling it slack changes the reading by more than the precision the triangle is capable of delivering, and a tape run over an obstruction reads long by an amount nobody can estimate. Keep the same tension on every measurement.
Does this work for a deck or a foundation?
Yes — it is the same geometry at any scale, and the case for a large triangle is stronger the larger the structure. Foundations and decks are exactly where a two inch runout matters, and they are also where there is plenty of room for a 15-20-25. Batter boards and a taut string beat a tape held at arm’s length over those distances.
What if the space is too small for a big triangle?
Use the largest that fits and measure more carefully to compensate — two people, a taut tape, and the same tension each time. Reading to a sixteenth rather than an eighth halves the error just as doubling the triangle would, so on a small job accuracy of measurement substitutes directly for size of triangle.
🔒 This tool runs entirely in your browser. Nothing you enter is uploaded, logged, or stored.