Let $S_n$ act on $S_n/H=\{ H,t_2 H,...,t_e H\}$ from the left and let $n\geq 5$, $e \geq 3$. Then the kernel of this action is trivial.
This supposedly follows because $e \geq 3$, but I haven't the slightest clue how? Now, the kernel is a normal subgroup of $S_n$, so it must be one of $S_n,A_n,\langle \mathrm{id} \rangle$. I'm actually not sure about the definition of the kernel of an action; Dummit/Foote defines the kernel of this particular action as $$\{ \sigma \in S_n \mid \sigma s_i H = s_i H, \forall s_i \in S_n \} = \cap_{s_i \in S_n} s_i H s_i^{-1}$$
I figure this is equivalent to $$\{ \sigma \in S_n \mid \sigma t_i H = t_i H, \forall t_i \in S_n//H \} = \cap_{t_i \in S_n//H} t_i H t_i^{-1}$$
In other words, those permutations stabilizing all the elements of $S_n/H$. When $e=2$ it can be checked "manually" that the kernel $A_n$, but showing that takes into account odd- and evenness and not the actual number $2$. How do I use the assumption $e\geq 3$ in proving the above assertion?
Hint: Denote that kernel of this action by $K$. Then $K, t_2K,\cdots, t_eK$ are pairwise different.