[guided]**Enumerate divisors.** The positive divisors of $15$ are $1, 3, 5, 15$; all are odd (as required, since $m$ must be odd).
**Tabulate.** For each $m$, we recover $s = (m-1)/2$ and then $k = s^2 + s + 1$:
\begin{align*}
m = 1: \quad s = 0, \quad k = 0 + 0 + 1 = 1, \\
m = 3: \quad s = 1, \quad k = 1 + 1 + 1 = 3, \\
m = 5: \quad s = 2, \quad k = 4 + 2 + 1 = 7, \\
m = 15: \quad s = 7, \quad k = 49 + 7 + 1 = 57.
\end{align*}
**Discard $k = 1$.** A $1$-regular graph is a disjoint union of edges; the only connected $1$-regular graph is $K_2$, which has diameter $1$. So $k = 1$ does not yield a Moore graph, and we exclude this case.
**Include $k = 2$.** The $5$-cycle $C_5$ is a Moore graph of degree $2$: it is $2$-regular, has girth $5$, and has diameter $2$ (the "opposite" vertex on the cycle is at distance $2$). Thus $k = 2$ is genuinely allowed — we handled it in the earlier degenerate case, where the fraction in $(\star)$ vanished because $k - 2 = 0$.
**Conclusion.** The only possible degrees are
\begin{align*}
k \in \{2, 3, 7, 57\}.
\end{align*}
This completes the proof.
(Historical note. Moore graphs actually exist for $k = 2$ (the pentagon $C_5$), $k = 3$ (the Petersen graph), and $k = 7$ (the Hoffman–Singleton graph). Whether a Moore graph of degree $57$ exists is a famous open problem in algebraic graph theory; this theorem says only that no other degree is possible.)[/guided]