Is it possible to partition any rectangle into congruent isosceles triangles?


No. Note that the acute angle of your triangle must divide $\pi/2$ (look at a corner), so there are countably many such triangles (up to similarity), and hence you get only a countable set of possible ratios of sides.

  • $\begingroup$ Brilliant! $\mbox{}$ $\endgroup$ – Joseph O'Rourke Nov 6 '10 at 21:03
  • $\begingroup$ You seem to be saying that there is a one-to-one correspondence between classes of similar triangles and classes of rectangles of sides of a given ratio. Perhaps I'm misunderstanding your argument, but if I'm not, I don't see why that has to be true. Can you elaborate a bit? Many thanks! $\endgroup$ – John Iskra Nov 8 '10 at 20:43
  • $\begingroup$ Not one-to-one, but if we fix triangle (with sides a and b), then both sides of rectangle are linear combinations of a,b with integer coefficients. So, there are at most countably many of them for fixed a,b. $\endgroup$ – Fedor Petrov Nov 8 '10 at 21:56
  • $\begingroup$ Got it. Thank you! That is a really nice proof. $\endgroup$ – John Iskra Nov 9 '10 at 2:31

If the length divided by the width is rational, then yes. Just partition the rectangle into congruent squares and cut each square along a diagonal.

  • 3
    $\begingroup$ Is this a necessary condition as well? $\endgroup$ – Jeremy West Nov 6 '10 at 0:22
  • 4
    $\begingroup$ The argument for necessity would run something like this: 1. By looking at a corner, convince yourself that the triangles must be right (i.e. a half-square). (Some angle on the triangle must divide $90^\circ$; then analyze by cases which angle it can be.) 2. Using irrationally of $\sqrt2$, argue that for any rectangle tiled with half-squares, all short (say) sides must be parallel/perpendicular to each other (as opposed to at $45^\circ$). 3. Whether all short sides are parallel to the sides of the rectangle or perpendicular, you still win. $\endgroup$ – Theo Johnson-Freyd Nov 6 '10 at 1:04
  • 1
    $\begingroup$ Yeah. Thank you to those who have posted so far: I did know, though, that the answer is 'yes' in the cases listed above. In fact, relaxing any of the hypotheses seems to give a positive answer. That is, if the triangles need not be isosceles, if they need not be congruent etc. I was asked this question without the congruency condition by a sophomore student I have. It seems like, potentially, a good research project, but I didn't know whether it was 1. known 2. trivial or 3. too hard. This is outside my area of specialization, so I thought I'd post here to get some sense of things. $\endgroup$ – John Iskra Nov 6 '10 at 1:37
  • 4
    $\begingroup$ If you knew all that, you could have saved people some work by mentioning it in your original statement. $\endgroup$ – Gerry Myerson Nov 6 '10 at 1:53
  • 2
    $\begingroup$ @Gerry: some was equal to about epsilon here, so no worries as far as I am concerned. @unknown: There is nothing wrong with extremely long questions on MO. One reason why I like MO is that I can learn a lot from just reading questions, and in this sense longer questions are better than short ones. Concerning your points {1,2,3}: I am not sure whether the answer is known, but combining Theo's comment with my answer gives a complete characterization. If one can thrash out the details of Theo's comment, I think it would be neither trivial (2) or too hard (3). Probably 2.4. $\endgroup$ – Tony Huynh Nov 6 '10 at 12:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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