A square has a side length of 5 units. Is it possible to cover this square with three squares each with a side length of 4 units?


4 Answers 4



see picture red is 5x5 the other three are 4x4 enter image description here few measurements on the green square: the triangle outside the red square to the left has sides $4,3,5$, therefore the triangle outside the red to the top has sides $1,0.75,1.25$ the crucial thing being $1.25 \ge 5-4$

  • 5
    $\begingroup$ Your diagram looks stretched; none of the squares look square. $\endgroup$ Aug 3, 2020 at 16:42
  • $\begingroup$ @user3294068 If you rotate the screen about 55 degrees to one side, then they all do. $\endgroup$
    – Penguino
    Aug 6, 2020 at 1:58

This is a known problem.

The answer is “yes”.

Because the golden ratio $\varphi=\frac{\sqrt{5}+1}{2}=1.618\dots$ is bigger than $\left(\frac 54\right)^2=1.44$

The following cover of a square of area $\varphi$ by three unit squares was found by Henry E. Dudeney in 1931.

enter image description here

Erich Friedman had a page “Squares Covering Squares” with coverings of the largest known square by $n$ unit squares.


I think the answer is



  • First place one of the $4\times4$ squares within the $5\times5$ square. The remaining area (9 units) should be as compact as possible, so let's shift the $4\times4$ square right up to one corner, leaving an L-shape remaining. (I'm not sure how to prove rigorously that this is optimal.)
  • Now we need to place the other two squares so as to cover that L-shape. Surely the best way to do this is

    diagonally, so that the longest possible part of the $4\times4$ square is along the length of the L on each side.

It seems this should be enough to cover the $5\times5$ square, since


but actually it's not that easy,

because the total $4\sqrt{2}$ length only covers zero width when the square is placed diagonally.

Let's do a quick bit of calculation:


Placed diagonally inside the $4\times4$ square for maximum length, the longest width-$1$ box that can fit is of length $4\sqrt{2}-1=4.66$.

Even allowing for the fact that the two additional $4\times4$ squares will meet at the corner of the L-shape, there's no way we can cover all of the two open edges of the $5\times5$ square, since our coverage of them will only go as far as $4.66$ along before the coverage starts to slope away from the edge.

  • 3
    $\begingroup$ Good effort undone by one little inaccuracy ("Surely the best way to do this is "), still +1 $\endgroup$ Aug 2, 2020 at 23:10
  • 5
    $\begingroup$ Ouch, I feel for Rand. Keep answering puzzles! And welcome Paul, I see you answered a lot recently. $\endgroup$
    – justhalf
    Aug 3, 2020 at 13:20
  • $\begingroup$ Certainly not "rigorous" either, but definitely seems likely that you need to go to a corner because otherwise you will have to cover areas on both sides of the smaller square. With only 3 squares to work with (2 remaining besides the first one not on a corner in this scenario), that seems really hard to do. $\endgroup$
    – Chipster
    Aug 3, 2020 at 23:31

You need to cover the four corners, so one of the smaller squares covers two corners of the bigger one and the other two smalls cover one corner each.

  • $\begingroup$ How can you cover two corners of the bigger square using one of the smaller squares? $\endgroup$ Aug 3, 2020 at 19:41
  • $\begingroup$ Turn it 45 degrees and make each of the two adjacent top edges coincident with one of the top vertices of the bigger square. $\endgroup$
    – AND
    Aug 3, 2020 at 19:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.