Trying this: limit (n!)/(((n)^(n+0.5))(e^-n)) as n->infinity, in Wolfram Alpha gives zero even though it should give $\sqrt(2\pi)$.
I am especially surprised that it wasn't recognised as a form of Stirling's formula.
Any ideas?
Edit: As per the comment below, replacing 0.5 by 1/2 solves the problem, so I suppose that the question now is, why does Wolfram Alpha compute the two seemingly equivalent forms of the same limit to give quite different results?