Newtonian Shirokov Effect: Epicyclic Frequency Splitting from Mass Multipoles
We analyze small oscillations of nearly circular orbits in an axisymmetric Newtonian potential expanded in mass multipoles, as the classical counterpart of the relativistic Shirokov effect. Computing the full Hessian of the effective potential at the true (possibly tilted) equilibrium and solving the coupled two-mode oscillator exactly, we obtain a complete picture. (i) A quadrupole splits the radial and vertical epicyclic frequencies, $Ω_θ^2-Ω_r^2=-3GQ/r_0^5=6GMJ_2R^2/r_0^5$, at first order in $J_2$; the Newtonian analogue of the Shirokov splitting, equivalent to the classical statement that an oblate body's apsidal and nodal rates differ. (ii) A gravitational dipole produces no splitting: it equals $M r_{\rm CM}$, is removable by re-centering at the center of mass, and cannot appear in any coordinate independent frequency; the apparent first order coupling cancels at the true tilted equilibrium, any residual absorbed by the induced quadrupole of the shifted source, confirmed by direct orbit integration. (iii) A genuine octupole does split the frequencies, $ω_+^2-ω_-^2\approx6G|O|/r_0^6$. The selection rule is thus not even/odd parity: every multipole splits the frequencies except the dipole. These yield two complementary probes of an axisymmetric source: the frequency splitting measures the oblateness $J_2$, while the orbital plane tilt, $δθ_0\simeq-r_{\rm CM}/r_0$, measures the center of mass offset $r_{\rm CM}$, an orbital geometry observable rather than a frequency one. We give solar system estimates for both. Carried through to Shirokov's original observable -- the secular transverse drift after $n$ orbits -- the quadrupole effect gives $ξ^θ=ξ_0^θ\,πn\,(6J_2R^2/r_0^2)$, of order $10^{-8}$ cm at $1$ au and $\sim10^{-6}$ cm near $0.1$ au, comparable to Shirokov's Schwarzschild estimate.