Okay, well, I have to show that, if $G$ is a finite group, such that, $|G|= 2k$, where $k$ is odd, then, translation by its element of order 2 is an odd permutation and $G$ must have a normal subgroup of order $k$. Here is what I've done so far:
Well, it is easy to show that $G$ must have an element of order 2. Let us call this, $a$.
It seemed to me that $G$ must be of the form $ [e, b_1, b_2,...b_k, a, ab_1, ab_2,...ab_k]$, where the $b_i$ are elements of odd order, but I needed to show this.
So, note that, if $ab= b^{-1}a$, then, assuming $ ab^{i-1} = b^{-i+1}a$, we have,
$ab^i = (ab^{i-1})b = b^{-i+1}(ab) = b^{-i+1} b^{-1}a = b^{-i}a = (ab^i)^{-1}$.
So, by induction, all such $ab^i$ have an order of 2.
As we can associate a $b^i$ to every $ab^i$, there are equal numbers of each, and so, there are $k$ of each.
This justifies the statement that $G = [e, b_1, b_2,...b_k, a, ab_1, ab_2,...ab_k].$
Translating $G$ by $a$, let this be $T_a$, I note is simply equal to a product of $k$ transpositions of the form $ [b_i, ab_i]$, and so, $\epsilon(T_a) = (-1)^k = -1$ as k is odd.
Now, I know that the tentative bit is the assumption that the base case for induction works. So, how could I show that $ab = b^{-1}a$?
Also, I know that the normal group is the set of elements with odd order, namely $[e, b_1, ...b_k]$, but how exactly could I show that it is closed?
Would this work?
Suppose that $b_ib_j = a b_r$, now, as each $b_i$ has generators which have an exponent of $k$, the left-side also has an exponent of k, whereas the right does not. Therefore, the $b_i b_j$ must remain in the group of elements with odd order.
It seems a little shaky to me, anyhow. Also, as a final query, are all groups of the form $2k$ with k odd, isomorphic to the k-dihedral group?
Thank You very much!