In trying to write an alternate and simple proof that at least one leg of a right triangle is a multiple of 4 using Dickson's method of generating triples, I came across quite an interesting observation that ifall the sides of a rectangle is odd, then it's diagonal is irrational. For example, consider a rectangle of length 5 units and breadth 3 units. It's diagonal,by the Pythagorean theorem is 34^0.5, which is irrational. This is the same for lots of rectangles. Is there a general proof for this, or can this whole thing be disproved?

  • $\begingroup$ Basically speaking, I am asking whether the sum of 2 odd perfect squares is an even perfect square... $\endgroup$ – Sandeep Nov 17 '14 at 13:32


This number cannot be a perfect square as its prime factorization includes $2^1$.


If the sides of the rectangle are integers, then the square of the length of the diagonal is also an integer since

$d^2 = l^2 + w^2$

and the integers are closed under multiplication and addition.

The only integers that have rational square square roots are perfect squares. Since

$d = \sqrt{l^2 + w^2}$

the length of the diagonal is irrational unless $d^2 = l^2 + w^2$ is a perfect square.

If an integer is a perfect square, then it is either the square of an even number or an odd number.

If the diagonal is an even number, then $d = 2k$ for some $k \in \mathbb{N}$, so $d^2 = (2k)^2 = 4k^2$.

If the diagonal is an odd number, then $d = 2k - 1$ for some $k \in \mathbb{N}$, so $d^2 = (2k - 1)^2 = 4k^2 - 4k + 1 = 4(k^2 - k) + 1$.

Hence, the square of any integer will have remainder $0$ or $1$ when divided by $4$.

If both the length and width are odd numbers, then $l^2 + w^2$ has remainder $2$ when divided by $4$. To see this, suppose that $l = 2m - 1$ and $w = 2n - 1$, where $m$ and $n$ are positive integers. Then

\begin{align*} d^2 & = l^2 + w^2\\ & = (2m - 1)^2 + (2n - 1)^2\\ & = 4m^2 - 4m + 1 + 4n^2 - 4n + 1\\ & = 4(m^2 - m + n^2 - n) + 2 \end{align*}

Since $d^2 = l^2 + w^2$ has remainder $2$ when divided by $4$, it is not a perfect square. Thus, $d = \sqrt{l^2 + w^2}$ is irrational.

What I have not shown is why the only integers that have rational square roots are perfect squares. It hinges on unique factorization. You can show that $n$ is a perfect square if and only if each prime in its prime factorization appears an even number of times. For example, $$144 = 2^4 \cdot 3^2 = (2^2 \cdot 3)^2 = 12^2$$ while $12 = 2^2 \cdot 3$ is not a perfect square. If $n$ is not a perfect square, then some prime in its factorization must appear an odd number of times. Suppose $n$ is not a perfect square and $$\sqrt{n} = \frac{r}{s}$$ where $r, s$ are integers with $s \neq 0$. Then \begin{align*} n & = \frac{r^2}{s^2}\\ ns^2 & = r^2 \end{align*} If $p$ is a prime that appears in the prime factorization of $n$ an odd number of times, then its appears in the prime factorization of $ns^2$ an odd number of times, while it appears in the prime factorization of $r^2$ an even number of times, a contradiction of unique factorization.

  • $\begingroup$ Could you please elucidate this answer? I feel its right, but simpler things are always understood better... $\endgroup$ – Sandeep Nov 17 '14 at 13:33
  • $\begingroup$ I have edited my answer. Since I do not know your mathematics background, I left the hardest material to the end. In the final paragraph, I outlined a proof of why $\sqrt{n}$ is irrational if $n$ is not a perfect square. To give a complete proof, I would have to employ a number of results from elementary number theory that depend on unique factorization. $\endgroup$ – N. F. Taussig Nov 17 '14 at 20:38

If a and b is odd then $a^2= 1(mod 4)$. Thus, if a,b are odd, $a^2 +b^2 =2 (mod 4)$. If c is even,then $ c^2=0 (mod4)$. If c is odd, $c=1 (mod4)$. Therefore, $a^2 +b^2$ cannot be a perfect square, which implies that its square root is irrational.


Suppose $c^2 = (2a+1)^2+(2b+1)^2=4a^2+4a+1+4b^2+4b+1=4(\ldots)+2$. So it's an even number not divisible by $4$.

But if $c^2$ is divisible by $2$, then $c$ has to be even and therefore $c^2$ is divisible by $4$. This proves it can't be a square of an integer.

If $c$ were rational, but not an integer, then $c^2$ is also a rational number which isn't an integer, so we don't need to worry about that.


I Think there is an answer to this here, since I posted a comment which explains the question in a line: prove that the sum of two odd perfect squares can never be a perfect square.

This one is similar to the one that Yves Daoust posted.

We will obtain a result: 2 * [odd number]. Even if we assume the odd number to be a perfect square, we have a 2 over there.


S Sandeep.


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.