DrOnline said:
Yes, I am aware of that, but if we call \sqrt{x} = u
Then \int \frac {1} {u} = ln(u), or: ln(\sqrt{x})
Bu that is false, which to me seems to violate the rule.
Edit: fixed some faulty reasoning. Now my question is valid.
You are doing the method of substitution incorrectly.
By your argument, you could just as well have let \displaystyle u=\frac{1}{\sqrt{x}} and come up with \displaystyle \int u\,, \ which would give a very different, and also incorrect, answer.
You may notice that in you textbook, there is
always a differential, dx in the case of your original integral, included with the integration symbol.
So you must do more than substituting \displaystyle u\ \ \text{for}\ \ \frac{1}{\sqrt{x}} \ . You must also substitute for dx.
In the case of \displaystyle u=\sqrt{x}\,,\ \ du=\frac{1}{2\sqrt{x}}dx\ .
Your integral becomes:
\displaystyle \int \frac{1}{\sqrt{x}}\,dx=\int \frac{2}{2\sqrt{x}}dx=\int 2\,du\ .
Added in
Edit: I see RGV beat me to it. We both agree, but expressed ourselves enough differently so that I'll leave my post here rather than deleting it. (I'm pretty slow at typing this stuff up.)