Apsidal Angles and Orbital Precession

In 1859 Urbain Le Verrier — the man who had discovered Neptune with a pen — announced a scandal: after subtracting every known planetary tug, Mercury's perihelion still crept forward by a stubborn residue, eventually pinned at 43 arcseconds per century. A whole planet ("Vulcan") was invented to explain it and hunted for fifty years. The real culprit was the force law itself, and the machinery that turns "the ellipse slowly pivots" into a precision instrument is this article's subject: the apsidal angle. The apsides are the turning points of the radial motion — pericentre and apocentre — and the apsidal angle $\Delta\phi$ is the azimuth the orbit sweeps while travelling from one to the other. For the pure Kepler problem $\Delta\phi=\pi$ exactly: perihelion and aphelion sit diametrically opposite, the ellipse closes, and the line of apsides is nailed to the stars. Almost any change to the force — a $1/r^3$ correction from general relativity, the oblateness of the central body, a third body's average pull — shifts $\Delta\phi$ away from $\pi$ , and the orbit responds by rotating its apsidal line at a steady rate: apsidal precession, $\Delta\varpi=2\Delta\phi-2\pi$ per radial period. That fragility is the payoff. Because the unperturbed angle is locked at $\pi$ , any measured drift converts directly into the strength of the perturbation: Mercury's 43″ became the first test of general relativity, satellite apsidal drift measures Earth's equatorial bulge, and binary-pulsar periastron advance weighs neutron stars. This article builds the apsidal-angle integral, the near-circular formula, and the classic perturbation results — ending with Mercury's number computed in full.

Key takeaways

The apsidal angle $\Delta\phi$ is the azimuth swept between successive radial turning points — pericentre to apocentre; successive pericentres are $2\Delta\phi$ apart.
The orbit closes only if $\Delta\phi$ is a rational multiple of $\pi$ ; otherwise the line of apsides rotates by $\Delta\varpi=2\Delta\phi-2\pi$ per radial period — apsidal precession.
Near-circular orbits obey $\Delta\phi=\pi\,[3+rF'/F]^{-1/2}$ ; for power-law forces $F=-c\,r^n$ this is $\pi/\sqrt{3+n}$ — exactly $\pi$ for Kepler ( $n=-2$ ), $\pi/2$ for Hooke ( $n=+1$ ).
A $\beta/r^2$ addition to the Kepler potential shifts the apsidal angle exactly: $\Delta\phi=\pi L/L'$ with $L'^2=L^2+2\mu\beta$ .
General relativity's correction gives $\Delta\varpi_{\mathrm{GR}}=6\pi GM/[ac^2(1-e^2)]$ per orbit — Mercury's 43″/century, on top of 532″/century from planetary tugs.
Key idea: the apsidal angle equals $\pi$ only for the inverse-square law; any other force makes the line of apsides precess by $2(\Delta\phi-\pi)$ per radial period, and the measured rate is a direct probe of the perturbation.

The core insight

The apsidal angle $\Delta\phi=\int_{r_{\min}}^{r_{\max}}\frac{L\,\mathrm{d}r}{r^2\sqrt{2\mu(E-V_{\mathrm{eff}})}}$ equals $\pi$ only for the inverse-square law; any other force makes the line of apsides precess by $\Delta\varpi=2(\Delta\phi-\pi)$ each radial period.

Where this lives today

Apsidal precession is now a workhorse measurement across four decades of orbital scale. Earth satellites: the equatorial bulge (the $J_2$ term, an effective $1/r^3$ force after averaging) rotates the apsidal line of any inclined elliptical orbit — except at the critical inclination $63.4°$ where the effect vanishes, which is exactly where Molniya orbits park their apogees ("frozen" apsides, no station-keeping needed). Binary pulsars: PSR B1913+16's periastron advances $4.2°$ per year — Mercury's formula, thirty thousand times stronger — turning equation (5) into a scale to weigh neutron stars; the double pulsar J0737-3039 advances $16.9°$ per year and tests GR at the 0.01% level. At the galactic centre, the GRAVITY interferometer has measured the Schwarzschild precession of the star S2 around Sagittarius A — about $12'$ per 16-year orbit, prograde as (5) demands — and GRAVITY+ is using the same apsidal instrument to hunt for extended dark mass inside S2's orbit, which would add a retrograde* Newtonian contribution on top of GR's prograde one. Exoplanets close the loop: measured apsidal precession of hot Jupiters probes their tidal deformation.

The physical picture

A bound orbit is a radial oscillation riding on a rotation: the radius breathes between $r_{\min}$ and $r_{\max}$ while the azimuth advances steadily. The apsidal angle measures how far the rotation gets during one half-breath. If it is exactly $\pi$ , aphelion sits directly opposite perihelion and the orbit is a closed, frozen ellipse. Shift it slightly and each radial cycle plants the perihelion at a new angle: the ellipse survives as a shape but slowly pivots into a rosette:

Figure

A pinned ellipse versus a pivoting one — precession is the apsidal line rotating a little each radial period

The geometry of one half-breath, drawn precisely: the orbit leaves pericentre, spirals outward between the two turning circles, and reaches apocentre after sweeping the apsidal angle — here $120°$ , the value for a force $F\propto r^{-3/4}$ :

Figure

One half radial period — the orbit climbs from the inner turning circle to the outer one while sweeping the apsidal angle

Two reference values anchor all intuition: Kepler's $\Delta\phi=\pi$ (one breath per revolution, closed focal ellipse) and Hooke's $\pi/2$ (two breaths, closed centred ellipse). Bertrand's theorem says these are the only force laws holding a rational apsidal angle at every amplitude — everything else precesses, which is exactly what makes the apsidal angle an instrument. The deep reason the Kepler value is protected is the conserved Laplace–Runge–Lenz vector; precession is literally the rotation rate of that would-be-frozen vector once the potential is perturbed.

The apsidal angle

Energy conservation gives the radial speed, angular momentum the angular speed, and their ratio integrates from one turning point to the other:

\Delta\phi=\int_{r_{\min}}^{r_{\max}}\frac{(L/r^2)\,\mathrm{d}r}{\sqrt{2\mu\left[E-V(r)\right]-L^2/r^2}} \qquad (1)

where: $r_{\min},r_{\max}$ — pericentre and apocentre, roots of the radicand; $L,E$ — the conserved angular momentum and energy; $\mu$ — the reduced mass. Time has been eliminated: this is pure geometry of the path. By the orbit's mirror symmetry about any apse, apocentre-to-pericentre sweeps the same angle, so pericentre-to-pericentre is $2\Delta\phi$ . Closure and precession now read off directly:

\text{closure}\iff\frac{\Delta\phi}{\pi}=\frac{p}{q}\in\mathbb{Q},\qquad \Delta\varpi=2\Delta\phi-2\pi\ \text{per radial period} \qquad (2)

where: $p,q$ — coprime integers (closure after $q$ radial periods, $p$ revolutions); $\Delta\varpi$ — rotation of the apsidal line per radial cycle. Sign convention: $\Delta\phi>\pi$ is prograde precession (Mercury's case); $\Delta\phi<\pi$ is regression. For nearly circular orbits the integral collapses to a frequency ratio — Derivation A gives:

\Delta\phi=\pi\,\frac{\omega_\phi}{\omega_r}=\pi\left[\,3+\frac{r_0F'(r_0)}{F(r_0)}\,\right]^{-1/2}\;\xrightarrow{\;F=-c\,r^n\;}\;\frac{\pi}{\sqrt{3+n}} \qquad (3)

where: $r_0$ — the circular radius; $n$ — the force exponent (Kepler $n=-2$ gives $\pi$ ; Hooke $n=+1$ gives $\pi/2$ ; the flat-rotation-curve force $n=-1$ gives $\pi/\sqrt2$ ). Stability — the bracket positive, $n>-3$ — is a prerequisite, the circular-orbit stability article's subject. Convention note: with the potential exponent $n_V=n+1$ this is the hidden-symmetry article's $\pi/\sqrt{n_V+2}$ .

Perturbing the Kepler problem

The instrument becomes quantitative when we perturb the inverse-square law and watch $\Delta\phi$ move. One case is exactly solvable: add a $1/r^2$ term to the potential (a $1/r^3$ force). Derivation B shows it simply renormalizes the centrifugal term, giving an exact apsidal angle:

V=-\frac{k}{r}+\frac{\beta}{r^2}:\qquad \Delta\phi=\pi\,\frac{L}{L'},\qquad L'^2=L^2+2\mu\beta \qquad (4)

where: $\beta$ — the perturbation strength, either sign; $L'$ — an effective angular momentum absorbing the perturbation. For small $\beta$ , $\Delta\varpi\approx-2\pi\mu\beta/L^2$ per orbit: an attractive extra term ( $\beta<0$ ) gives prograde precession. This one formula covers a lot of real physics, because both the leading GR correction and a planet's oblateness produce effective $1/r^3$ force terms after averaging. General relativity's version carries its own famous coefficient. The Schwarzschild correction to the effective potential is $\delta V=-GML^2/(\mu c^2r^3)$ , and first-order perturbation theory over the Kepler ellipse (Derivation C) gives the per-orbit advance:

\Delta\varpi_{\mathrm{GR}}=\frac{6\pi GM}{a\,c^2\,(1-e^2)} \qquad (5)

where: $M$ — the central mass; $a,e$ — semi-major axis and eccentricity; $c$ — the speed of light. Read the structure: the advance grows for tight orbits (small $a$ ) and eccentric ones (the $1-e^2$ ), is independent of the orbiting mass, and is always prograde. Mercury — closest in, most eccentric of the classical planets — shows it best: $43''$ per century, computed to the arcsecond in the worked example.

Derivation AThe apsidal angle for nearly circular orbits

The goal is the working formula (3), $\Delta\phi=\pi[3+r_0F'/F]^{-1/2}$ , which converts a force law into an apsidal angle in one line. The strategy: a near-circular orbit is a small radial oscillation about the minimum of the effective potential, so we need only two frequencies — the radial wobble set by the well's curvature, and the revolution; the apsidal angle is $\pi$ times their ratio, because pericentre-to-apocentre is half a radial period. Watch the two uses of the circular-orbit condition to eliminate $L$ .

Step 1. Balance at the circular radius. $V_{\mathrm{eff}}'(r_0)=-F(r_0)-L^2/\mu r_0^3=0$ gives $L^2=-\mu r_0^3F(r_0)$ , positive for attraction. Why: the circular orbit is the equilibrium of the radial coordinate; this relation converts $L$ into force-law data.

Step 2. Curvature of the well. $V_{\mathrm{eff}}''(r_0)=-F'(r_0)+3L^2/\mu r_0^4=-F'(r_0)-3F(r_0)/r_0$ , using Step 1. Why: the curvature is the spring constant of the radial wobble: $\omega_r^2=V_{\mathrm{eff}}''(r_0)/\mu$ .

Step 3. The two frequencies. $\omega_\phi^2=L^2/\mu^2r_0^4=-F(r_0)/\mu r_0$ , so $\dfrac{\omega_r^2}{\omega_\phi^2}=\dfrac{-F'-3F/r_0}{-F/r_0}=3+\dfrac{r_0F'(r_0)}{F(r_0)}$ . Why: everything dimensional cancels — the ratio depends only on the local logarithmic slope of the force.

Step 4. Half a radial period of angle. Pericentre to apocentre takes $t=\pi/\omega_r$ , during which the azimuth advances $\Delta\phi=\omega_\phi\,t=\pi\,\omega_\phi/\omega_r$ — formula (3). For $F=-c\,r^n$ the slope is $n$ and $\Delta\phi=\pi/\sqrt{3+n}$ . Why: the apsidal angle literally measures the angle clock during half a tick of the radius clock; a power law makes it radius-independent. *Verified: $n=-2$ gives $\pi$ , $n=+1$ gives $\pi/2$ , both matching the exact closed solutions; agrees with Goldstein §3.6 and Landau §14.*

Go deeper · Derivation B · 1 minExact apsidal angle for an inverse-cube perturbation

Derivation BExact apsidal angle for an inverse-cube perturbation

The goal is the exact result (4): adding $\beta/r^2$ to the Kepler potential changes the apsidal angle to $\pi L/L'$ with $L'^2=L^2+2\mu\beta$ — no expansion in $\beta$ required. The strategy is an absorption trick: the perturbation has the same $1/r^2$ shape as the centrifugal term, so it is swallowed into an effective angular momentum, reducing the problem to an exactly Keplerian one. Watch the final step — the dynamics uses $L'$ , but the angle is swept at the true rate $\dot\phi=L/\mu r^2$ , and the mismatch is the whole effect.

Step 1. Absorb the perturbation. $V_{\mathrm{eff}}=-\dfrac{k}{r}+\dfrac{L^2}{2\mu r^2}+\dfrac{\beta}{r^2}=-\dfrac{k}{r}+\dfrac{L'^2}{2\mu r^2}$ with $L'^2\equiv L^2+2\mu\beta$ . Why: the perturbation and the centrifugal barrier have identical radial shape — a coincidence special to $1/r^2$ potentials that makes this case exactly solvable.

Step 2. Solve the radial problem. The radial motion is now exactly Keplerian with angular momentum $L'$ : the orbit oscillates between turning points of a pure Kepler problem, and one full radial cycle corresponds to $2\pi$ of the auxiliary angle $\Phi$ defined by $\mathrm{d}\Phi=(L'/\mu r^2)\,\mathrm{d}t$ . Why: with $L'$ in the centrifugal seat, every radial formula of the Kepler problem applies verbatim; the auxiliary angle is the one in which the effective orbit is a closed conic.

Step 3. Convert back to the true angle. The physical azimuth accumulates as $\mathrm{d}\phi=(L/\mu r^2)\,\mathrm{d}t=(L/L')\,\mathrm{d}\Phi$ . Over one radial half-period ( $\Delta\Phi=\pi$ ): $\Delta\phi=\pi L/L'$ — equation (4). Why: the same $r(t)$ is being watched through two angular clocks; the true clock runs slower than the auxiliary one by exactly $L/L'$ .

Step 4. Expand for small perturbations. $\Delta\phi=\pi(1+2\mu\beta/L^2)^{-1/2}\approx\pi(1-\mu\beta/L^2)$ , so $\Delta\varpi=2(\Delta\phi-\pi)\approx-2\pi\mu\beta/L^2$ per orbit. Why: attractive extra term, $\beta<0$ , gives $\Delta\phi>\pi$ — prograde advance, the sign Mercury shows; the exact form also covers strong perturbations where the expansion fails. *Verified: $\beta=0$ recovers $\Delta\phi=\pi$ ; the result matches the Landau–Lifshitz §15 problem on the $1/r^2$ perturbation.*

Go deeper · Derivation C · 1 minMercury's relativistic perihelion advance

Derivation CMercury's relativistic perihelion advance

The goal is the flagship number: equation (5), evaluating to $43''$ per century for Mercury. The strategy is first-order perturbation theory around the Kepler ellipse: general relativity adds $\delta V=-GML^2/(\mu c^2r^3)$ to the effective potential, and the standard apsidal-shift formula converts an orbit-averaged $\delta V$ into a perihelion advance; the Kepler orbit equation makes the average elementary.

Step 1. State the perturbation. The Schwarzschild radial equation matches Newton's with an extra term: $\delta V=-\dfrac{GML^2}{\mu c^2 r^3}$ . Why: the leading post-Newtonian correction for a test mass; the $L^2$ factor couples orbital motion to spacetime curvature.

Step 2. Apply the first-order apsidal-shift formula. For small $\delta V$ , the advance per orbit is $\Delta\varpi=\dfrac{\partial}{\partial L}\left[\dfrac{2\mu}{L}\displaystyle\int_0^{\pi}r^2\,\delta V\,\mathrm{d}\phi\right]$ on the unperturbed ellipse. Why: the classical-perturbation workhorse (Landau §15) — the apsidal angle is the $L$ -derivative of the radial action's correction, and $\mathrm{d}t=(\mu r^2/L)\mathrm{d}\phi$ converts time average to angle average.

Step 3. Average over the Kepler ellipse. With $r^2\delta V=-GML^2/(\mu c^2 r)$ and the conic $\dfrac1r=\dfrac{1+e\cos\phi}{a(1-e^2)}$ : $\displaystyle\int_0^\pi\frac{\mathrm{d}\phi}{r}=\frac{\pi}{a(1-e^2)}$ . Why: the $r^2$ weight turns the $1/r^3$ potential into a $1/r$ average, and only the constant part of the conic survives the integral.

Step 4. Differentiate and collect. Carrying the $L$ -dependence through $p=a(1-e^2)=L^2/\mu k$ , the derivative gives $\Delta\varpi=\dfrac{6\pi GM}{ac^2(1-e^2)}$ — equation (5). Why: the factor 6 assembles from the derivative acting on both explicit and implicit $L$ -dependence; mass independence signals the equivalence principle.

Step 5. Insert Mercury. $GM_\odot=1.327\times10^{20}$ , $a=5.79\times10^{10}$ m, $e=0.206$ : $\Delta\varpi=5.02\times10^{-7}$ rad/orbit; with $415.2$ orbits per century, $5.02\times10^{-7}\times415.2\times206265''=\boxed{43.0''\ \text{per century}}$ . Why: a tenth of an arcsecond per orbit, but relentless accumulation lifts it into 19th-century measurability. *Verified: the standard result (Einstein 1915); $43.0''$ matches the observed residual after Newtonian planetary perturbations.*

Limiting cases

Case	Result	Meaning
Kepler, $n=-2$	$\Delta\phi=\pi$ , $\Delta\varpi=0$	Closed focal ellipse; apsidal line frozen.
Hooke, $n=+1$	$\Delta\phi=\pi/2$	Closed centred ellipse; the other Bertrand survivor.
Power law near-circular	$\Delta\phi=\pi/\sqrt{3+n}$ , needs $n>-3$	Generic: irrational multiple of $\pi$ , steady precession.
Small $\beta/r^2$ perturbation	$\Delta\varpi\approx-2\pi\mu\beta/L^2$ per orbit	Exact in $\beta$ via $L'$ ; sign of advance = sign of extra attraction.
GR correction	$\Delta\varpi=6\pi GM/[ac^2(1-e^2)]$	Always prograde; largest for tight, eccentric orbits.

Worked examples

1 · Reading a force law off the precession (confidence builder)

A near-circular orbit is observed whose apsides are $120°$ apart — pericentre to apocentre. What force law is acting, and how does the apsidal line move?

Step 1. Invert equation (3): $\Delta\phi=\pi/\sqrt{3+n}$ with $\Delta\phi=120°=2\pi/3$ gives $\sqrt{3+n}=180/120=1.5$ , so $3+n=2.25$ and $\boxed{n=-0.75}$ : the force is $F\propto r^{-3/4}$ , falling off much more gently than gravity.

Step 2. Precession per radial cycle from (2): $\Delta\varpi=2\times120°-360°=\boxed{-120°}$ — the apsidal line regresses by a third of a turn every radial period ( $\Delta\phi<180°$ , as expected for a force shallower than inverse-square).

Step 3. Closure check: $\Delta\phi/\pi=2/3$ is rational, so this near-circular orbit closes after 3 radial cycles and 2 revolutions — a three-petal pattern. (By Bertrand's theorem the closure is an accident of the near-circular limit: at finite eccentricity $\Delta\phi$ drifts off $120°$ and the petals precess.)

2 · Mercury's forty-three arcseconds, in full (exam level)

Compute the general-relativistic perihelion advance of Mercury and place it in the observed budget. Data: $GM_\odot=1.327\times10^{20}\ \mathrm{m^3/s^2}$ , $a=5.79\times10^{10}$ m, $e=0.206$ , orbital period $87.97$ d.

Step 1. Per orbit, equation (5): $\Delta\varpi=\dfrac{6\pi\times1.327\times10^{20}}{5.79\times10^{10}\times(2.998\times10^8)^2\times(1-0.206^2)}=\boxed{5.02\times10^{-7}\ \mathrm{rad}}$ — about $0.104''$ per orbit.

Step 2. Orbits per century: $36525/87.97=415.2$ .

Step 3. Accumulate: $5.02\times10^{-7}\times415.2=2.08\times10^{-4}$ rad/century $\times\,206265''=\boxed{43.0''\ \mathrm{per\ century}}$ .

Step 4. The budget. Relative to a fixed reference frame, Mercury's perihelion advances about $575''$ per century: $\approx532''$ from the Newtonian pulls of the other planets plus the $43''$ of GR. Le Verrier's scandal was that the planetary $532''$ fell short of observation — and the missing $43''$ , unexplainable by any Vulcan, became general relativity's first confirmation. (Quoted against the precessing equinox the raw figure is larger by the $\sim5025''$ of Earth's axial precession — always check the frame.)

Origins · 1 minIsaac Newton and how this was discovered

Isaac Newton

1642–1727 · Cambridge

Newton owns this article's central instrument, though the fact is little known: Book I of the Principia (1687) contains the theorem of revolving orbits, showing that adding an inverse-cube force to any orbit rotates it rigidly without changing its radial motion — precisely the absorption of Derivation B, two centuries early. He used it as a measurement device exactly as we do: from the near-stillness of planetary apsides he argued the force exponent must be extremely close to $-2$ , and he analyzed the glaring exception — the Moon's apsidal line, rotating a full turn in under nine years under the Sun's perturbation — coming within a factor of two of the right rate, a discrepancy resolved decades later by Clairaut's more careful perturbation theory.

History

1687

— Newton's revolving-orbits theorem ties apsidal motion to the force law; the Moon's apse problem opens perturbation theory.

1749

— Clairaut resolves the lunar apsidal-rate puzzle, vindicating inverse-square gravity at second order.

1859

— Le Verrier announces Mercury's anomalous perihelion advance; the phantom planet Vulcan is born.

1915

— Einstein derives the missing 43 arcseconds per century from general relativity — its first quantitative success.

1975

— Hulse and Taylor's binary pulsar shows a periastron advance of 4.2 degrees per year, GR precession magnified thirty-thousand-fold.

Going further

References. Goldstein, Poole & Safko, Classical Mechanics, 3rd ed., §3.6 (the differential orbit equation), §3.6 Bertrand's theorem · Landau & Lifshitz, Mechanics, 3rd ed., §14, §15 (Kepler problem), problems on apsidal angles · Morin, Introduction to Classical Mechanics, Ch. 7 (central forces and orbit precession) · Taylor, Classical Mechanics, Ch. 8 (two-body central-force problems) · Arnold, Mathematical Methods of Classical Mechanics, §8 (Bertrand's theorem)

Provenance. Equations and Bertrand's theorem from Goldstein §3.6 and Arnold §8; apsidal-angle integral and $1/r^2$ perturbation from Landau-Lifshitz Mechanics §14-15. A verifier should recompute: (1) the nearly circular formula $\Delta\phi=\pi[3+rF'/F]^{-1/2}$ reducing to $\pi$ and $\pi/2$ for $n=-2,+1$ ; (2) the GR result $6\pi GM/[ac^2(1-e^2)]$ ; (3) Mercury's $43''$ /century using $a=5.79\times10^{10}$ m, $e=0.206$ , $GM_\odot=1.327\times10^{20}$ m $^3$ s $^{-2}$ , $c=2.998\times10^8$ m/s, and ~415 orbits/century (orbital period 88 days). Worked-example numbers recomputed for this upgrade: $5.02\times10^{-7}$ rad/orbit, $415.2$ orbits/century, $43.0''$ ; inverse problem $\Delta\phi=120°\Rightarrow n=-0.75$ , regression $120°$ /cycle; budget $532''+43''\approx575''$ /century (inertial frame). PSR B1913+16 $4.2°$ /yr and J0737-3039 $16.9°$ /yr are standard values; S2 precession $\approx12'$ /orbit (GRAVITY 2020).

Builds on: The Effective Potential · The Orbit Equation and Binet's Equation · Reduction to the Radial Equation · Angular Momentum Conservation in Central Fields · Conic-Section Orbits Leads to: The Kepler Problem · Classification of Orbits by Energy · Kepler's First Law: Elliptical Orbits

Is the apsidal angle a property of the force law or of the particular orbit?

Both, and keeping the dependencies straight prevents most confusion. For a given force law, $\Delta\phi$ generally depends on which orbit you fly — $E$ and $L$ enter the integral (1) through the turning points. The near-circular limit is special: there $\Delta\phi=\pi[3+rF'/F]^{-1/2}$ depends only on the local slope of the force, and for power laws not even on the radius. Only for the two Bertrand force laws is the apsidal angle a pure constant at every amplitude ( $\pi$ and $\pi/2$ ). That is why quoting "the" apsidal angle of a generic potential carries a silent near-circular qualifier — and why eccentric orbits need the full integral.

Why is the angle from apocentre back to pericentre exactly the same as from pericentre to apocentre?

Time-reversal symmetry about an apse. At a turning point the radial velocity vanishes, so running the motion backward from that instant produces a valid orbit retracing the same radial history; geometrically, the orbit is mirror-symmetric about the line from the force centre through any apse. The outbound half-cycle is the mirror image of the inbound one, so each sweeps the same azimuth. This symmetry is also why precession is one number per radial period: every pericentre-to-pericentre stretch is congruent to every other, rotated by $\Delta\varpi$ .

How can Derivation B be exact when perturbation theory is usually approximate?

Because for the special shape $\beta/r^2$ the "perturbation" is not treated perturbatively at all — it is absorbed. The centrifugal term is also $\propto1/r^2$ , so the sum is again a pure Kepler effective potential with shifted constant, $L'^2=L^2+2\mu\beta$ , and the radial motion is exactly Keplerian. The only trace of the perturbation is that the physical angle accumulates at the rate set by the true $L$ , not $L'$ , giving the exact factor $\Delta\phi=\pi L/L'$ . Newton's revolving-orbits theorem is the same statement: an inverse-cube force rotates an orbit rigidly. Any other shape breaks the coincidence, and first-order averaging (Derivation C) becomes the honest tool.

If GR's correction is attractive, why does it make the apsides advance rather than fall behind?

Track the frequencies. Extra attraction concentrated at small radius steepens the force law, lowering $\beta^2=3+rF'/F$ below the Kepler value of 1. A smaller $\beta$ means the radial oscillation is slow relative to the revolution: the planet completes more than half a turn before reaching apocentre, $\Delta\phi=\pi/\beta>\pi$ . Each radial cycle plants the perihelion a bit ahead in the direction of motion — prograde. The logic reverses for forces shallower than inverse-square, or for a spread-out mass distribution: $\Delta\phi<\pi$ and the apsides regress — exactly the discriminating signature GRAVITY+ exploits at the galactic centre.

Why is Mercury the planet that shows the GR effect best?

Equation (5) says the advance scales as $1/[a(1-e^2)]$ : you want to be close to the Sun and eccentric. Mercury is both — smallest $a$ , largest $e$ of the classical planets — and its 88-day year piles up 415 orbits per century, converting $0.104''$ per orbit into an observable $43''$ . Venus is nearly circular ( $e=0.007$ ): its GR advance per orbit is not much smaller, but the apsidal line of an almost-circle is nearly unmeasurable. Earth's advance is $3.8''$ /century. The scaling also explains the modern champions: binary pulsars with $a$ of a solar radius and $e\sim0.6$ turn the same formula into degrees per year.

Do the 532 arcseconds from planetary perturbations also come from this article's machinery?

Conceptually yes: each planet's time-averaged pull adds a small non-inverse-square component to the force on Mercury, shifting its apsidal angle off $\pi$ ; summing the contributions (Venus $\approx278''$ , Jupiter $\approx154''$ , Earth $\approx90''$ , the rest $\approx10''$ ) gives the $\approx532''$ Le Verrier computed by hand. In practice the calculation is organized as Laplace–Lagrange secular perturbation theory, but the observable is the same $\Delta\varpi$ and the logic identical: perturbation in, apsidal drift out. The drama was that this careful Newtonian sum failed by $43''$ — and the failure was the signal: the leftover belonged not to an unseen planet but to the force law itself.