Given identical 30-60-90 triangles, what is the convex polygon with the highest number of sides that I can build from them?

This seems a very easy task by first look, but I’m totally stuck right now. The only thing I’m pretty sure of is that the polygon can at most have 12 sides, because a 150 degrees angle is the maximum convex angle I can find combining triangles. Obviously this does not guarantee that a dodecagon can be filled with 30-60-90 triangles.

Any suggestion? I was tempted to post it in Math, so if you think it is not pertinent I will shut the question down.

  • $\begingroup$ Did you mean "regular" polygon? It would be trivial to assemble an irregular polygon $\endgroup$
    – user46002
    Commented Oct 23, 2018 at 22:14
  • 2
    $\begingroup$ The regular polygon with the highest number of sides that I can build is an hexagon indeed. I mean a general convex (I specified it in the title but didn’t add it in the text, just edited) polygon. $\endgroup$ Commented Oct 23, 2018 at 22:17
  • $\begingroup$ If say six — I can't think of anything better. $\endgroup$
    – user46002
    Commented Oct 23, 2018 at 23:07
  • 3
    $\begingroup$ "equal-sided 30-60-90 triangles" is an impossibility unless you allow non-zero curvature. If you mean "identical 30-60-90 triangles" you should say that. $\endgroup$
    – Nij
    Commented Oct 24, 2018 at 5:56
  • 1
    $\begingroup$ At the end of second paragraph, is "decagon" a typo for "dodegacon" ? $\endgroup$
    – Evargalo
    Commented Oct 24, 2018 at 7:02

3 Answers 3


Here is a convex dodecagon made of $50$ of those triangles.

enter image description here

Can it be done with fewer?

  • $\begingroup$ I didn’t even thought that it could be done with such a small number of triangles! Great work! $\endgroup$ Commented Oct 24, 2018 at 6:31

It is possible to do better than a hexagon, if an irregular polygon is acceptable.

One can construct an irregular dodecagon with $180$ identical 30-60-90 triangles as shown below. The dodecagon would have sides of alternating lengths, with ratio $4:\sqrt{12}$; as a specific size for the 30-60-90 triangle is not given.

In the diagram below, the larger white triangles are not unit 30-60-90 triangles — they are actually made up of four, smaller 30-60-90 triangles, which is demonstrated in the top-left (shaded in pink). The image also includes two regular hexagons (of different sizes), a bunch of equilateral triangles, and some rectangles (all constructed from 30-60-90 triangles). It is not possible to get more than 12 sides because the minimum external angle that can be created from 30-60-90 triangles is 210 degrees.
dodecagon constructed from 30-60-90 triangles

It is also possible to construct an equilateral triangle or a hexagon. On reflection (and thanks to @Hugh's comment) you probably can't make a square — but can get relatively close to a square. By taking $\sqrt{3}$ by $1$ squares, each made from two triangles, you could stack them $n$ by $\frac{n}{\sqrt{3}}$ which, for sufficiently large $n$, would approach $1:1$.

  • 1
    $\begingroup$ Well done, @Penguino, I didn't think of making a dodecagon that big! You say that it's possible to make a square, but I don't think that's a perfect square...? Please correct me if I'm wrong — regardless of you using two different sizes of triangle, all four sides contain a small "long leg" and a small "short leg". Other than that, the bottom and top edges contain: 2 big "hypotenusen" and 1 big "long leg". The left and right sides contain: 2 big "long legs" and 1 big "hypotenuse". Which, doesn't add up: $2n + 2n + n\sqrt{3} \neq 2n + n\sqrt{3} + n\sqrt{3}$. $\endgroup$
    – user46002
    Commented Oct 24, 2018 at 2:51
  • $\begingroup$ Yep - you are right - I sketched to soon :) $\endgroup$
    – Penguino
    Commented Oct 24, 2018 at 2:55
  • $\begingroup$ I've just suggested an edit, it corrects some grammar and adds math formatting. :) $\endgroup$
    – user46002
    Commented Oct 24, 2018 at 3:21
  • 1
    $\begingroup$ Feel free to edit Hugh. $\endgroup$
    – Penguino
    Commented Oct 24, 2018 at 3:26
  • 1
    $\begingroup$ Great answer indeed! Did you find a ‘filling algorithm’ while building the figures, or just tried casual combinations? Btw, regarding the square, there are two triangles in the bottom left of the picture that are hypotenuse/long cathetus-sided, which is not allowed :) $\endgroup$ Commented Oct 24, 2018 at 6:29

The answer is that

a convex dodecagon can indeed be made out of 30-60-90 tiles; you can see one rather complicated construction here.

[EDITED to add:] The original version of this answer made a claim stronger than was correct; thanks to Jaap Scherphuis for pointing out my error.

  • 4
    $\begingroup$ That is not quite regular. If the hypotenuse of the triangle has length $1$, some sides have of the dodecahedron have length $7$, others have length $4\sqrt{3}\approx6.928$. $\endgroup$ Commented Oct 24, 2018 at 3:04
  • $\begingroup$ Whoops, of course you're right. But it's definitely convex. $\endgroup$
    – Gareth McCaughan
    Commented Oct 24, 2018 at 10:35

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.