I know that we have for a function $\Gamma: (-\varepsilon,\varepsilon)^2 \rightarrow M$ we have (at least I think I know that this is true)
$$\nabla_{\frac{\partial \Gamma}{\partial s}} \frac{\partial \Gamma(s,t)}{\partial t} = \nabla_{\frac{\partial \Gamma}{\partial t}} \frac{\partial \Gamma(s,t)}{\partial s}.$$
I am not looking for a proof of this one!
Now I found a proof on page 2 of this reference (first line) click me where I was wondering what exactly happens between the second and third term.
I mean it looks pretty much like this, but I was wondering at which points you have to evaluate the function and in what way they commute with each other.
So there it says $$ \nabla_{\frac{\partial \Gamma}{\partial t}} \frac{\partial}{\partial s}|_{s=0} \Gamma = \nabla_{\frac{\partial \Gamma}{\partial s}} \frac{\partial}{\partial t}|_{s=0} \Gamma .$$
My problem is that the $t$ evaluation is completely missing here and I don't know whether I have to commute $t$ and $s$ evaluation, too. Thus, I don't know where the covariant derivative should be evaluated.
So if anybody could make this step more precise, this would totally answer my question.
In Riemannian Geometry book of Manfredo perdigão do carmo you can see one proof for your question in pag 68 in the symmetry lemma.
Sorry for my English.
you can see 1: https://i.stack.imgur.com/9DSDx.png
ps: The lemma 4.1 in image is the symmery lemma.