\section{Conventions: visibility, noise and the energy \texorpdfstring{$\EG$}{E\_G}} \label{sec:setting} \paragraph{The measured quantity.} A rigid body of mass $M$ is prepared in a superposition of two static positions a distance $d$ apart. Both branches carry the same mass. The source is inertial and enters as an external c-number density. The two paths are recombined and the fringe visibility $|z(t)|$ is recorded as a function of the time $t$ spent apart, where $z$ is twice the off-diagonal element of the body's density matrix between the two branches. A rate is a constant $\Gamma$ with $|z(t)|\propto e^{-\Gamma t}$ at late times. \paragraph{The rate from noise.} Let the mass density couple to a potential operator, $H_{\rm int}=\int\rho\,\Phi\,d^3x$. The branches differ through the which-path variable $B(t)=\int\Delta\rho(\mathbf{x})\,\Phi(\mathbf{x},t)\,d^3x$. With $\delta B=B-\langle B\rangle$, define its symmetrized noise and its dissipative (absorptive) response, \begin{equation} S_B(\omega)=\frac12\int\!d\tau\,e^{i\omega\tau}\bigl\langle\{\delta B(\tau),\delta B(0)\}\bigr\rangle,\qquad \chi''_B(\omega)=\frac{1}{2\hbar}\int\!d\tau\,e^{i\omega\tau}\bigl\langle[B(\tau),B(0)]\bigr\rangle . \label{eq:SB-chiB} \end{equation} For Gaussian fluctuations and pure dephasing, \begin{equation} -\ln|z(t)|=\frac{1}{2\pi\hbar^2}\int_{-\infty}^{\infty}\!d\omega\;S_B(\omega)\, \frac{1-\cos\omega t}{\omega^2}. \label{eq:dephasing} \end{equation} Since $\int(1-\cos\omega t)\,\omega^{-2}d\omega=\pi t$, the late-time slope is \begin{equation} \Gamma=\lim_{t\to\infty}\frac{-\ln|z(t)|}{t}=\frac{S_B(0)}{2\hbar^2}. \label{eq:rate-floor} \end{equation} A rate requires a zero-frequency floor. If $S_B(\omega)=\order{\omega^2}$ as $\omega\to0$, the integrand of Eq.~\eqref{eq:dephasing} is bounded at small $\omega$ and the dephasing exponent $-\ln|z|$ saturates. \paragraph{Di\'{o}si--Penrose noise.} Classical white potential noise with \begin{equation} \langle\delta\Phi(\mathbf{x},t)\,\delta\Phi(\mathbf{y},s)\rangle= \frac{C\hbar G}{|\mathbf{x}-\mathbf{y}|}\,\delta(t-s),\qquad S_\Phi(\mathbf{k},\omega)=\frac{4\pi C\hbar G}{k^2}\ \ \text{for all }\omega, \label{eq:dp-noise} \end{equation} gives $S_B(0)=C\hbar G\iint\Delta\rho\,\Delta\rho\,|\mathbf{x}-\mathbf{y}|^{-1}=2C\hbar\EG$, so Eq.~\eqref{eq:rate-floor} reproduces Eq.~\eqref{eq:dp-rate}. At $C=1$ this is the noise behind Di\'{o}si's master equation $\dot{\hat\rho}=-(G/2\hbar)\iint[\hat M(\mathbf{x}),[\hat M(\mathbf{y}),\hat\rho]]\, |\mathbf{x}-\mathbf{y}|^{-1}$~\cite{Diosi1987,Diosi1989}, with $\hat\rho$ the density matrix and $\hat M(\mathbf{x})$ the mass-density operator. We measure $C$ in these Di\'{o}si units throughout. The normalization used with Penrose's prescription in Refs.~\cite{Donadi2021,Arnquist2022} is larger by $8\pi$, so it corresponds to $C=8\pi$; Appendix~\ref{app:norm} gives the master equation of each cited bound and derives the factor. \paragraph{The energy.} $\EG$ in Eq.~\eqref{eq:dp-rate} contains the same-branch terms; they make $\EG\geq0$. For uniform spheres of radius $a$ at centre separation $d$, with $s=d/a$, \begin{equation} \EG=\frac{GM^2}{a}\,f(s),\qquad f(s)=\begin{cases}\dfrac{s^2}{2}-\dfrac{3s^3}{16}+\dfrac{s^5}{160}, & s<2,\\[2mm] \dfrac65-\dfrac1s, & s\geq2,\end{cases} \label{eq:EG-sphere} \end{equation} (Appendix~\ref{app:energy}); the two branches agree with their first three derivatives at contact, $s=2$. The energy is tidal, $\EG\simeq GM^2d^2/(2a^3)$, for $d\ll a$ and saturates at $(6/5)GM^2/a$ for $d\gg a$. The point-mass expression $GM^2/d$ is the magnitude of the cross term alone; it is not a limit of Eq.~\eqref{eq:dp-rate} for any mass distribution.