MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I've been studying some proofs of the four-square theorem. Some of them are pretty clear. However, I came across a statement that the four-square theorem can be easily derived from Gauss-Legendre three-square theorem. Hard as I tried, I couldn't find out how to do it.

I was hoping someone could give me some idea or point out some article.

Thank you!

share|cite|improve this question
0 and 1 are both squares... – zeb Mar 16 '13 at 10:31 This gives a short and elementary proof of three square theorem. – i707107 Mar 16 '13 at 16:54
I see that this question has just been edited to change "four squares theorem" to "four-square theorem". The literature seems to be divided on the issue. MathSciNet lists 9 papers with "four squares theorem" in the title, 7 with "four-square theorem", 4 with "four square theorem", and 3 with "four-squares theorem". – Gerry Myerson Apr 17 at 22:44

The three squares theorem tells you that a positive integer $n$ can be represented as the sum of 3 squares if and only if it is not of the form $n = 4^a(8b+7)$, see e.g. Thus it suffices to show that for every integer of the form $n = 4^a(8b+7)$, there is an $m \in \mathbb{N}$ such that $n - m^2$ is not of this form. However taking $m := 2^a$, we get $n - m^2 = 4^a(8b+6)$, and we are done.

share|cite|improve this answer
The page you linked to just proves the easy part of the three squares theorem. – Marc van Leeuwen Mar 16 '13 at 13:42
@Marc: The OP asked for a proof of the implication "three squares theorem => four squares theorem" rather than for a proof of the three squares theorem. – Stefan Kohl Mar 16 '13 at 14:06
That's really neat. Thank you:) – user32252 Mar 17 '13 at 9:09

Also, the Gauß/Legendre theorem in question implies the following (improved) version of the four-squares theorem:

Every positive integer is the sum of four POSITIVE squares unless it belongs to the set

$A \cup B$


$A$ $=$ {$1, 3, 5, 9, 11, 17, 29, 41$}


$B$ $=$ {$2\cdot 4^{m}: m \in \mathbb{Z}^{+}$} $\cup$ {$6\cdot 4^{m}: m \in \mathbb{Z}^{+}$} $\cup$ {$14\cdot 4^{m}: m \in \mathbb{Z}^{+}$}.

Proof (d'après Prof. J. H. Conway in "The sensual quadratic form" [page 140]). By Gauß/Legendre and the well-known result on numbers that can be written as a sum of two squares it follows that every natural number of the form $8k+3$ (or $8k+6$) is the sum of three positive squares. Multiplying by $4$, we see that the same conclusion applies to natural numbers of the forms $32k+12$ and $32k+24$. Then, one can show that any integer $>49$ which is not a multiple of $8$ is the sum of four positive integers by subtracting an square so as to obtain a number of one of the aforementioned forms (for instance, from a number of the form $8k+2$ subtract $2^{2}$ in order to obtain a number of the form $8k+6$...). The proof is completed by checking the numbers up to $49$ and verifying that a number $n$ divisible by $8$ is the sum of four positive squares only if $\frac{n}{4}$ is. QED.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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