\subsection{A finite Gaussian envelope} The matched Gaussian pair has means $(n/2,n/2)$, marginal variances $n/4$, and correlation $\rho=e^{-2t}$, as do the exact counts. Its log-clock information is $g(t)$ from \eqref{eq:g}. The proof uses the posterior parity recurrence to construct three latent polynomials with zero conditional mean; their Gram matrix bounds the conditional second Krawtchouk score. It includes every endpoint cell and uses no Gaussian approximation. \begin{theorem}[A finite Gaussian information envelope] \label{thm:finite-gaussian} In the stationary Ehrenfest experiment of Proposition~\ref{prop:exact}, let $R_n$ be the retained information fraction \eqref{eq:R}, $B_n$ the quadratic reference \eqref{eq:finite-endpoint-bounds} and $g$ the matched Gaussian information \eqref{eq:g}. For every integer $n\ge1$ and finite $t>0$, with $\lambda=nt$, \begin{equation}\label{eq:finite-gaussian-envelope} B_n(t)\le R_n(\lambda)<\frac{g(t)}{\lambda}. \end{equation} In particular, \begin{equation}\label{eq:finite-gaussian-relative} 1\le\frac{R_n(\lambda)}{B_n(t)} <1+\frac{2}{n(e^{4t}-1)} \le1+\frac{1}{2\lambda}. \end{equation} Thus $\lambda\ge5$ suffices for a relative interval of width at most ten percent around $B_n$, at every population size and sampling time. For each fixed $t>0$, the envelope is asymptotically sharp: \begin{equation}\label{eq:finite-gaussian-limit} ntR_n(nt)\longrightarrow g(t),\qquad n\longrightarrow\infty. \end{equation} \end{theorem} The complete proof, including the singular populations $n=1,2,3$ and an explicit strict correction for $n\ge4$, is in Appendix~\ref{app:finite-gaussian}.