and whole number arithmetic went their separate ways
It's not hard to see that one geometric square can be cut apart into four pieces and reassembled as two identical smaller squares (so it's natural to say the larger square is twice one of the smaller ones): draw the two diagonals of the larger square, then cut along them; you'll get four pieces (identical right isosceles triangles); take two and glue them together along their hypotenuses, then do the same with the other two. Voila! Two identical smaller squares from a larger one from a larger one!
Now, if you are a atomist, it is natural for you to hope that the squares are actually little square arrays of tiny atoms:
oo
oo
ooo
ooo
ooo
oooo
oooo
oooo
oooo
And since one geometric square can be twice another, the number of atoms in the bigger square should be twice the number of atoms in a smaller square. So you start looking for a square number that's twice another square number, and then youhave a lot of "close but no cigar" moments:
Two of these
oo
oo
make up one of these
ooo
ooo
oo
Oops! There's an atom missing!
Two of these
ooooo
ooooo
ooooo
ooooo
ooooo
make up one of these
ooooooo
ooooooo
ooooooo
ooooooo
ooooooo
ooooooo
oooooooo
Oops! There's an atom too many!
There's actually an endless supply of near misses like that, with an atom too few or an atom too many. Here are the first few near misses:
(3 x 3) - 1 = 2 x (2 x 2)
(7 x 7) + 1 = 2 x (5 x 5)
(17 x 17) - 1 = 2 x (12 x 12)
(41 x 41) + 1 = 2 x (29 x 29)
(99 x 99) - 1 = 2 x (70 x 70)
(239 x 239) + 1 = 2 x (169 x 169)
&c &c
But surely we can find a square number that's twice another square number, can't we?
Um ... sorry! Nope!
Here's an easy way to see there's no square number that's twice another square number. Suppose there were whole numbers A > 0 and B > 0 with A x A = 2 x B x B. Then there's a smallest such A > 0: we could find it by trying
A = 1 (nope!)
A = 2 (nope!)
A = 3 (nope!)
...
and so on until we found the very first one that worked: A x A = 2 x B x B for some B > 0. Notice that B < A. Now A is either even or odd, and if A is odd then A x A is odd and equal to the even number 2 x B x B. Since no odd number is even, we see that A must be even: say A = 2 x C. Thus 2 x C x 2 x C = A x A = 2 x B x B. Dividing by 2, we get B x B = 2 x C x C. So B is another a square number that's twice another square number. But A was supposed to be the very smallest example, and yet B < A! That's ridiculous, so there's no square number that's twice another square number
Hippasus was not popular with the atomists