Here's another proof:
Let's suppose that there are two positive, relatively-prime integers, call them a and b, such that a/b = √5 with a > b.
Multiply both sides by b, so we have
a = b√5
Now square both sides
a2 = 5b2
And note that we can write 5b2 as 4b2 + b2 Now, subtract b2 from both sides....so we have...
a2 - b2 = 4b2
Now, note that the right side is even, and suppose that a and b are both even. But, this is impossible since they are relatively prime and thus, have no factors in common.
And neither can one be odd an the other even, because the left side would be odd, but the right side is even.
The only possibility left is that both are odd. But this can't be so for the following reason: Factor the left side as (a + b) (a - b). And if a and b are odd, then (a + b) is even and so is (a - b) - (remember that a > b, so a - b > 0).
So we have
(a + b) ( a - b) = 4b2 Now divide both sides by 4 ....... and we can write
[(a + b)/2] [(a - b)/2] = b2
Now, if (a-b)/2 is odd, then (a+b)/2 is even ..... and vice-versa. To see this, let (a - b)/2 = 2k + 1. Then a - b = 2(2k + 1) .... Add b to both sides..... a = 2(2k + 1) + b....Now add b again to both sides and we have a + b = 2(2k +1) + 2b.......Now, divide by 2 on both sides, and we have (a+b)/2 = (2k + 1) + b, and both sides are even. Likewise, it can be proved that if (a +b)/ 2 is odd, then (a-b)/2 is even.
But, an even times an odd is always even...... but b2 is odd.
Thus......a and b aren't both even or both odd, and neither are they of different parity. and nothing else can be possible. Thus, our assumption that there exists two positive, relatively-prime integers such that a/b = √5 must be false. And thus, √5 must be irrational.
