Classification of Orbits by Energy

On 19 October 2017, a telescope in Hawaii picked up a faint dot moving too fast. Within days, orbit fitters had a single number that rewrote the object's biography: its orbital energy was positive. Not by a little, not within error bars of zero — decisively positive. That one sign meant 'Oumuamua was not a comet from the Solar System's cold outskirts on a long fall inward; it was a visitor that had never been bound to the Sun at all, would swing through once, and would leave forever. No trajectory integration was needed to reach that verdict. One conserved number, compared against one curve, settled it. That is the power this article develops. Once a central-force problem is reduced to one-dimensional radial motion in the effective potential $V_{\mathrm{eff}}(r)$ , the conserved energy $E$ becomes a horizontal line drawn across a fixed landscape — and where that line sits relative to the curve is a complete classification of the motion. Line tangent at the well bottom: a circle. Line cutting the well twice: a planet or satellite, forever rocking between two radii. Line at the rim: the knife-edge parabola. Line above the rim: a hyperbolic flyby with speed to spare at infinity. For gravity the dictionary is exact and quantitative: the sign of $E$ picks the conic, and the pair $(E, L)$ fixes its eccentricity through one formula. Escape velocities, transfer orbits, comet returns, interstellar visitors — all of them are one energy comparison. Learn to read the line against the curve and most orbit questions answer themselves before any equation of motion is solved.

Key takeaways

Radial motion in a central field is 1-D motion in $V_{\mathrm{eff}}(r) = U(r) + L^2/2\mu r^2$ ; the conserved energy $E$ confines the body to radii where $V_{\mathrm{eff}}(r) \le E$ , with turning points at $E = V_{\mathrm{eff}}(r)$ .
Two finite turning points = bound orbit; one finite turning point with escape to infinity = unbound orbit; a double root at the well bottom = circular orbit, the most bound state for given $L$ .
For the Kepler potential $U = -k/r$ : $E < 0$ ellipse, $E = 0$ parabola, $E > 0$ hyperbola — the sign test works because $U \to 0$ at infinity.
Energy and angular momentum fix the eccentricity: $e = \sqrt{1 + 2EL^2/\mu k^2}$ , so $e<1$ , $e=1$ , $e>1$ mirror $E<0$ , $E=0$ , $E>0$ .
The semimajor axis depends on energy alone, $a = -k/2E$ ; the hyperbolic excess speed is $v_\infty = \sqrt{2E/\mu}$ — the number rocket engineers sell as $C_3 = v_\infty^2$ .
The sign-of- $E$ shortcut assumes the potential vanishes at infinity; in general, compare $E$ to $V_{\mathrm{eff}}(\infty)$ .
Key idea: treat radial motion as 1-D motion in $V_{\mathrm{eff}}(r)$ — the horizontal line $E = V_{\mathrm{eff}}(r)$ cuts the curve at the turning points, and the number and finiteness of the intersections decide whether the orbit is circular, bound, or unbound.

The core insight

Treat radial motion as 1-D motion in $V_{\mathrm{eff}}(r)$ : the horizontal line $E=V_{\mathrm{eff}}(r)$ cuts the curve at the turning points, and the number and finiteness of intersections decide whether the orbit is circular, bound, or unbound.

Where this lives today

The classification is the daily currency of spaceflight. Launch-vehicle performance is quoted as payload versus $C_3 = v_\infty^2 = 2\varepsilon$ — literally "how much positive specific energy can this rocket buy you": $C_3 < 0$ names a bound parking orbit, $C_3 = 0$ is escape, and interplanetary missions shop the curve upward (Mars transfers need $C_3 \approx 10\text{–}15\ \mathrm{km^2/s^2}$ ; New Horizons left for Pluto at a record $C_3 \approx 158\ \mathrm{km^2/s^2}$ ). Every gravity assist is an exercise in the same bookkeeping: the flyby is hyperbolic ( $E > 0$ ) in the planet's frame — the planet cannot capture the spacecraft — yet it redirects $v_\infty$ , and that rotation, viewed in the Sun's frame, changes the heliocentric energy, walking Voyager from bound ellipse to Solar-escape hyperbola one planet at a time. The astronomers' side of the ledger is busier than ever. After 'Oumuamua (2017) and comet 2I/Borisov (2019) came 3I/ATLAS in 2025, and the Vera Rubin Observatory's all-sky survey is expected to turn hyperbolic interlopers from historic events into a steady catalog — each one flagged, sight unseen, by an orbit fit returning $e > 1$ . At the other extreme, the classification generalizes: around black holes the effective potential grows a maximum, and the clean bound/unbound dichotomy gains a third fate — capture — which is exactly the transition LIGO hears when two black holes' mutual orbit crosses from inspiral to plunge.

The physical picture

Picture the effective potential for gravity as a landscape: an infinitely tall centrifugal cliff on the left, a valley, and a long slope rising back to sea level ( $V_{\mathrm{eff}} \to 0$ ) on the right. Now flood it. The water level is the total energy $E$ , and the orbiting body is a boat that can only sail where there is water. Everything about the orbit's type is a statement about the shoreline.

Figure

One well, four fates — where the energy line sits against the effective potential decides the orbit type

Flood the valley only to its lowest point and the boat sits pinned at one radius: a circular orbit, the most tightly bound motion possible for that angular momentum. Raise the water somewhat and the boat sloshes between an inner shoreline and an outer one — perihelion and aphelion — while the angular motion silently carries it around the center: a bound orbit. Raise the level exactly to sea level and something qualitative snaps: the outer shoreline retreats to infinity, and the boat can drift arbitrarily far out, arriving there with nothing left — the parabolic threshold. Any higher and the boat not only escapes but still has headway at infinity: a hyperbolic passage with excess speed $v_\infty$ . Notice what did the classifying: not the force, not the trajectory, just the topology of the flooded region — two walls, one wall, or a single pinned point. That topology argument works for any central potential, however exotic its $V_{\mathrm{eff}}$ : multiple wells mean distinct families of bound orbits, a hump means an unstable circular orbit at its crest and the possibility of capture over the top. For gravity specifically, the picture sharpens into an exact algebraic dictionary — each water level maps to a conic section of definite eccentricity, which is where we go next.

The energy line and the turning points

Everything rests on the reduced energy equation inherited from the radial reduction. The relative motion of the two bodies, described by the reduced mass $\mu$ at separation $r$ , carries conserved energy

E = \tfrac12\mu\dot r^2 + V_{\mathrm{eff}}(r), \qquad V_{\mathrm{eff}}(r) = U(r) + \frac{L^2}{2\mu r^2} \qquad (1)

where: $U(r)$ — the true central potential; $L = \mu r^2\dot\phi$ — the conserved angular momentum; $E$ — the energy of the relative motion in the center-of-mass frame (not the lab kinetic energy — a distinction that matters when the two masses are comparable). Since $\tfrac12\mu\dot r^2 \ge 0$ always, the body can only exist where $V_{\mathrm{eff}}(r) \le E$ . The boundaries of the allowed region are where the radial motion stalls:

E = V_{\mathrm{eff}}(r) \quad\Longleftrightarrow\quad \dot r = 0 \qquad (2)

where: these radii are the turning points — perihelion and aphelion for a bound orbit, the distance of closest approach for an unbound one. At a turning point the body is not at rest: all its speed is angular, $v = L/\mu r$ . Counting the solutions of (2) is the classification: two finite roots enclose a bound orbit; one finite root with an allowed corridor to $r \to \infty$ is a scattering orbit; a double root at the well minimum is the circular orbit. For the Kepler potential $U(r) = -k/r$ (gravity: $k = Gm_1m_2$ ), the well was mapped quantitatively in The effective potential: a single minimum at $r_0 = L^2/\mu k$ of depth $V_{\mathrm{eff}}(r_0) = -\mu k^2/2L^2$ , an infinite centrifugal wall on the left, and a tail rising to zero on the right. That shape makes the sign of $E$ decisive, because the rim of the well sits exactly at $E = 0$ : any negative energy is trapped by the tail; any non-negative energy clears it.

Derivation AThe floor of the energy ladder and the circular orbit

We are establishing the bottom rung: for given angular momentum $L$ in the Kepler potential, the least possible energy is $E_{\min} = -\mu k^2/2L^2$ , attained by the circular orbit at $r_0 = L^2/\mu k$ . The strategy is direct minimization of $V_{\mathrm{eff}}$ — the energy of any orbit is at least the minimum of the curve it rocks in, so the well bottom is simultaneously a bound on energies and an orbit in its own right. Watch the exact factor of $\tfrac12$ appear: the centrifugal refund is half the gravitational debt, a fingerprint of the $1/r$ potential that recurs as the virial theorem.

Step 1. Write the effective potential for gravity: $V_{\mathrm{eff}}(r) = -\dfrac{k}{r} + \dfrac{L^2}{2\mu r^2}$ . Why a single well is guaranteed: the barrier wins as $r\to0$ ( $1/r^2$ beats $1/r$ ), the attraction wins as $r\to\infty$ (it dies slower), so the curve descends from $+\infty$ , crosses zero, bottoms out, and rises back to $0^-$ — exactly one minimum.

Step 2. Find the extremum:

V_{\mathrm{eff}}'(r) = \frac{k}{r^2} - \frac{L^2}{\mu r^3} = 0 \quad\Longrightarrow\quad r_0 = \frac{L^2}{\mu k}

Why this radius is special beyond being the well bottom: it will reappear as the semi-latus rectum of every conic with this angular momentum — the geometric size scale that $L$ alone fixes.

Step 3. Evaluate the depth:

E_{\min} = V_{\mathrm{eff}}(r_0) = -\frac{k}{r_0} + \frac{L^2}{2\mu r_0^2} = -\frac{\mu k^2}{L^2} + \frac{\mu k^2}{2L^2} = -\frac{\mu k^2}{2L^2}

where: the two contributions are the full gravitational term and the half-sized centrifugal term. No orbit with this $L$ can have $E < E_{\min}$ : the energy line would lie entirely below the curve, leaving no allowed radius at all.

Step 4. Confirm the well bottom is a genuine minimum: $V_{\mathrm{eff}}''(r_0) = -\dfrac{2k}{r_0^3} + \dfrac{3L^2}{\mu r_0^4} = \dfrac{\mu^3k^4}{L^6} > 0$ . Why this matters for the classification: a positive curvature means energies just above the floor produce small radial oscillations — the ellipses of low eccentricity that the next derivation quantifies — rather than some runaway. The floor is stable footing. *Verified: matches Goldstein §3.3; the result is consistent with the energy–eccentricity relation below, which gives exactly $e = 0$ at $E = E_{\min}$ .*

From energy to eccentricity: the exact dictionary

The qualitative "two walls versus one wall" story becomes an exact formula through the turning points. For gravity, equation (2) is a quadratic in $1/r$ , and its discriminant is — almost embarrassingly directly — the orbit's eccentricity. This is the algebraic bridge between the energy method and the geometry of conic-section orbits.

Go deeper · Derivation B · 2 minThe energy–eccentricity relation from the turning-point quadratic

Derivation BThe energy–eccentricity relation from the turning-point quadratic

We are deriving the master dictionary entry, $e = \sqrt{1 + 2EL^2/\mu k^2}$ , and reading the orbit classification straight off it. The strategy: write the turning-point condition $E = V_{\mathrm{eff}}$ for the Kepler potential, substitute $u = 1/r$ to make it a quadratic, and extract the apsidal radii from the two roots. The eccentricity then emerges as the normalized spread of the roots — geometrically, how lopsided the orbit is between its nearest and farthest points. The step to watch is the root count as $E$ changes sign: the classification is literally the algebra of a quadratic gaining and losing positive roots.

Step 1. Set $\dot r = 0$ in the energy equation with $U = -k/r$ and substitute $u = 1/r$ :

E = -ku + \frac{L^2}{2\mu}u^2 \quad\Longrightarrow\quad \frac{L^2}{2\mu}u^2 - ku - E = 0

Why the substitution: in $r$ the equation has an awkward mixed form; in $u$ it is a clean quadratic whose two roots are precisely $1/r_{\min}$ and $1/r_{\max}$ .

Step 2. Solve:

u_\pm = \frac{k \pm \sqrt{k^2 + 2EL^2/\mu}}{L^2/\mu} = \frac{\mu k}{L^2}\left(1 \pm \sqrt{1 + \frac{2EL^2}{\mu k^2}}\right)

where: the prefactor $\mu k/L^2 = 1/r_0$ is the inverse of the well-bottom radius, and the radical is a dimensionless measure of how far $E$ sits above the floor. Call that radical $e$ :

e = \sqrt{1 + \frac{2EL^2}{\mu k^2}} \qquad (3)

Step 3. Read the apsides from the roots: $u_\pm = (1 \pm e)/r_0$ , i.e.

r_{\min} = \frac{r_0}{1+e}, \qquad r_{\max} = \frac{r_0}{1-e} \qquad (4)

Why $e$ deserves the name eccentricity: forming the lopsidedness ratio gives $(r_{\max}-r_{\min})/(r_{\max}+r_{\min}) = e$ exactly — and comparing with the conic $r(\phi) = r_0/(1+e\cos\phi)$ derived from Binet's equation identifies $e$ with the geometric eccentricity and $r_0$ with the semi-latus rectum $p$ .

Step 4. Classify by counting positive roots. For $E_{\min} < E < 0$ : $0 < e < 1$ , both roots positive — two finite apsides, a bound ellipse. At $E = 0$ : $e = 1$ , the root $u_- = 0$ means $r_{\max} = \infty$ — one finite apsis, marginal escape, the parabola. For $E > 0$ : $e > 1$ , $u_-$ is negative and unphysical — only a perihelion survives, the hyperbola. At $E = E_{\min}$ : $e = 0$ , both roots merge at $r_0$ — the circle. Why the boundaries are razor-thin: $e$ depends continuously on $E$ , so the parabola at exactly $E = 0$ is a set of measure zero — real objects are always slightly elliptic or slightly hyperbolic, which is why orbit fitters report $e = 1.0006$ rather than "parabola".

Step 5. Cross-check with the semimajor axis. Define $a$ through $r_{\min} + r_{\max} = 2a$ (bound case): $2a = r_0\left(\tfrac{1}{1+e} + \tfrac{1}{1-e}\right) = \tfrac{2r_0}{1-e^2}$ , and substituting $r_0 = L^2/\mu k$ with (3) gives

a = \frac{r_0}{1-e^2} = -\frac{k}{2E} \qquad (5)

where: all $L$ -dependence has cancelled — the size of a bound orbit is fixed by energy alone, while $L$ (through $e$ ) only decides its shape. For $E>0$ the same formula yields a negative $a$ , the standard sign convention for hyperbolas. Verified: reproduces Landau and Lifshitz §15 and Goldstein §3.7; limits check — $E = E_{\min}$ gives $e = 0$ and $a = r_0$ ; $E \to 0^-$ gives $e \to 1$ and $a \to \infty$ .

The dictionary in full, for the record — sign of energy, eccentricity, geometry:

e = \sqrt{1 + \frac{2EL^2}{\mu k^2}}: \qquad E<0 \leftrightarrow e<1 \text{ (ellipse)}, \quad E=0 \leftrightarrow e=1 \text{ (parabola)}, \quad E>0 \leftrightarrow e>1 \text{ (hyperbola)}

where: the correspondence is two-way — astronomers measure a conic and infer the energy; mission designers budget an energy and know the conic. Two immediate corollaries carry most of the practical traffic. For unbound orbits, evaluating (1) at $r \to \infty$ (where $V_{\mathrm{eff}} \to 0$ and all motion is radial) gives the hyperbolic excess speed

v_\infty = \sqrt{\frac{2E}{\mu}} \qquad (6)

where: this is the speed the visitor keeps after climbing all the way out of the well — the launch industry's characteristic energy is exactly $C_3 = v_\infty^2 = 2E/\mu$ . And for any radius on any Kepler orbit, solving $E = \tfrac12\mu v^2 - k/r$ with (5) gives the vis-viva equation (per unit mass, $GM = k/\mu$ for a light orbiter):

v^2 = GM\left(\frac{2}{r} - \frac{1}{a}\right) \qquad (7)

where: feed it any $r$ and the orbit's $a$ , and it returns the speed there — the single most-used formula in orbital mechanics. Its three faces are the classification again: for an ellipse ( $a > 0$ ) the bracket vanishes at $r = 2a$ , an unreachable ceiling that caps the orbit; for the parabola ( $a \to \infty$ ) it reduces to the local escape speed $v = \sqrt{2GM/r}$ ; for a hyperbola ( $a < 0$ ) the $-1/a$ term keeps $v^2 > 0$ all the way out, leaving exactly $v_\infty$ at infinity.

Figure

Same angular momentum, rising energy — the conic family at fixed L shares its semi-latus rectum and fattens from circle to hyperbola

Which tool answers which question

you want	use	notes
orbit type from $E$	sign of $E$ (Kepler)	only valid because $U(\infty) = 0$
orbit type, any potential	count roots of $E = V_{\mathrm{eff}}(r)$ , eq. (2)	two walls bound, one wall unbound
shape from $(E, L)$	$e = \sqrt{1 + 2EL^2/\mu k^2}$ , eq. (3)	the exact dictionary
apsidal radii	$r_{\min,\max} = a(1 \mp e)$ , eq. (4)–(5)	or roots of the turning-point quadratic
size from energy	$a = -k/2E$ , eq. (5)	$L$ -independent — energy alone
leftover speed at infinity	$v_\infty = \sqrt{2E/\mu}$ , eq. (6)	$C_3 = v_\infty^2$ in launch specs
speed anywhere	vis-viva, eq. (7)	needs only $r$ and $a$

The radial clock: periods and closure

Classification says where the body may go; one more integral says how long the radial round trip takes. Separating variables in (1) gives the time between turning points, and doubling it, the radial period. The same separation with $\mathrm{d}\phi = (L/\mu r^2)\,\mathrm{d}t$ gives the angle swept per radial cycle — the quantity that decides whether the orbit closes.

Go deeper · Derivation C · 2 minThe turning-point integral and the apsidal angle

Derivation CThe turning-point integral and the apsidal angle

The goal is the pair of quadratures that convert the classification into clocks: the radial period $T_r$ and the apsidal angle $\Delta\phi$ swept between perihelion and aphelion — valid for any central potential, not just Kepler. The strategy is direct separation of variables in the energy equation; no ingenuity, just bookkeeping done carefully once. Watch the closure criterion at the end: it is the doorway to Bertrand's theorem and to precession as a precision probe of new physics.

Step 1. Solve (1) for the radial speed: $\dot r = \pm\sqrt{\tfrac{2}{\mu}\left(E - V_{\mathrm{eff}}(r)\right)}$ , sign flipping at each turning point. Why the flip is automatic: at a wall-touch $\dot r$ passes through zero and the body re-enters the allowed region the only way it can — back the way it came in $r$ .

Step 2. Separate and integrate over one inbound-to-outbound sweep, then double:

T_r = 2\int_{r_{\min}}^{r_{\max}} \frac{\mathrm{d}r}{\sqrt{\tfrac{2}{\mu}\left(E - V_{\mathrm{eff}}(r)\right)}} \qquad (8)

where: the integrand diverges at both limits (the speed vanishes there) but integrably, like $1/\sqrt{x}$ — turning points cost finite time. Why this defines a period at all: the radial motion is a genuine 1-D oscillation in the well, so it repeats with period $T_r$ regardless of what the angle is doing.

Step 3. Convert the same integral to angle using $\mathrm{d}\phi/\mathrm{d}r = \dot\phi/\dot r = \dfrac{L/\mu r^2}{\dot r}$ :

\Delta\phi = 2\int_{r_{\min}}^{r_{\max}} \frac{(L/r^2)\,\mathrm{d}r}{\sqrt{2\mu\left(E - V_{\mathrm{eff}}(r)\right)}} \qquad (9)

where: $\Delta\phi$ — the apsidal angle, the angle between successive perihelion and aphelion directions doubled to a full radial cycle. It is the ratio of the two clocks, angular to radial.

Step 4. State the closure criterion: the orbit is a closed curve if and only if $\Delta\phi$ is a rational multiple of $2\pi$ — after finitely many radial cycles the apsides return to their starting direction. For Kepler, evaluating (9) gives $\Delta\phi = \pi$ per half-cycle: perihelion to aphelion spans exactly half a turn, the orbit closes after one radial period, and the ellipse is frozen. Why this is fragile and precious: Bertrand's theorem says only $-k/r$ and the isotropic spring close all bound orbits; any perturbing physics detunes $\Delta\phi$ from $\pi$ and shows up as apsidal precession — which is precisely how Mercury's 43 arcseconds per century betrayed general relativity. *Verified: standard quadratures, matching Goldstein §3.5–3.6; Kepler closure $\Delta\phi = \pi$ consistent with the closed ellipses of the explicit solution.*

Limiting cases

Case	Result	Meaning
$E = E_{\min} = -\mu k^2/2L^2$	double root at $r_0 = L^2/\mu k$ , $e = 0$	circular orbit — deepest state at given $L$
$E_{\min} < E < 0$	two turning points, $0 < e < 1$ , $a = -k/2E$	bound ellipse rocking between apsides
$E = 0$	$r_{\max} \to \infty$ , $e = 1$ , $v_\infty = 0$	parabola — escapes with nothing to spare
$E > 0$	one turning point, $e > 1$ , $v_\infty = \sqrt{2E/\mu}$	hyperbolic flyby
$L \to 0$ at fixed $E<0$	$e \to 1$ with $a$ finite	degenerate radial ellipse — plunge through the center line
$U(\infty) \neq 0$	compare $E$ to $V_{\mathrm{eff}}(\infty)$ , not to 0	sign-of- $E$ shortcut fails off-Kepler

Common trapNegative total energy always means the orbit is bound

Only if the potential vanishes at infinity — the rule is really $E < V_{\mathrm{eff}}(\infty)$ , and for Kepler these coincide because $U(\infty) = 0$ . Shift the potential by a constant (perfectly legal — physics fixes only potential differences) and the dividing line shifts with it; a potential like $U = -k/r + c$ or a confining $U \propto r^2$ (where every orbit is bound, whatever the sign of $E$ ) breaks the sign test outright. Two more traps ride along. First, $E$ in this classification is the energy of the relative motion in the center-of-mass frame, built with the reduced mass $\mu$ — for a binary star, blindly using one star's lab-frame kinetic energy misclassifies the orbit. Second, the parabola at exactly $E = 0$ is a mathematical knife-edge: real measured orbits are always $e$ slightly below or above 1, and calling a long-period comet "parabolic" is shorthand for "bound, but with $r_{\max}$ too far to distinguish from escape at current precision".

Worked examples

1 · Three launches from Earth's surface (confidence builder)

Ignoring drag and Earth's rotation, an object is launched horizontally from Earth's surface ( $r = 6.371\times10^6\ \mathrm{m}$ , $GM_\oplus = 3.986\times10^{14}\ \mathrm{m^3/s^2}$ ) at 8.0, 11.19, and 12.0 km/s. Classify each trajectory and find what each becomes.

Step 1. The threshold is the local escape speed, where the specific energy $\varepsilon = \tfrac12 v^2 - GM_\oplus/r$ crosses zero: $v_{\mathrm{esc}} = \sqrt{2GM_\oplus/r} = \sqrt{2 \times 3.986\times10^{14}/6.371\times10^6} = \sqrt{1.251\times10^8} = \boxed{11.19\ \mathrm{km/s}}$ .

Step 2. At $v = 8.0\ \mathrm{km/s}$ : $\varepsilon = \tfrac12(8.0\times10^3)^2 - \dfrac{3.986\times10^{14}}{6.371\times10^6} = 3.20\times10^7 - 6.26\times10^7 = -3.06\times10^7\ \mathrm{J/kg} < 0$ : a bound ellipse. Its size: $a = -GM_\oplus/2\varepsilon = 3.986\times10^{14}/6.11\times10^7 = 6.52\times10^6\ \mathrm{m}$ , so the far apsis is $2a - r = 6.67\times10^6\ \mathrm{m}$ — apogee altitude $\approx\boxed{300\ \mathrm{km}}$ . The launch point is perigee; the "orbit" would skim the launch site each revolution (which is why real rockets also raise perigee).

Step 3. At $v = 11.19\ \mathrm{km/s}$ exactly: $\varepsilon = 0$ , $e = 1$ : a parabola. The object coasts out forever, arriving at infinity with zero speed — the knife-edge case.

Step 4. At $v = 12.0\ \mathrm{km/s}$ : $\varepsilon = 7.20\times10^7 - 6.26\times10^7 = +9.44\times10^6\ \mathrm{J/kg} > 0$ : a hyperbola. The leftover speed at infinity is $v_\infty = \sqrt{2\varepsilon} = \sqrt{1.888\times10^7} = \boxed{4.34\ \mathrm{km/s}}$ . Note the nonlinear bookkeeping: launching only 0.81 km/s above escape buys 4.34 km/s at infinity — the well refunds energy, not speed, since $v_\infty = \sqrt{v^2 - v_{\mathrm{esc}}^2}$ .

2 · 'Oumuamua: convicting an interstellar visitor (exam level)

The object 1I/'Oumuamua passed perihelion at $q = 0.256\ \mathrm{au} = 3.83\times10^{10}\ \mathrm{m}$ with hyperbolic excess speed $v_\infty = 26.3\ \mathrm{km/s}$ ( $GM_\odot = 1.327\times10^{20}\ \mathrm{m^3/s^2}$ , 1 au $= 1.496\times10^{11}\ \mathrm{m}$ ). Find its specific orbital energy, semimajor axis, eccentricity, and perihelion speed.

Step 1. Energy from the excess speed via (6): $\varepsilon = \tfrac12 v_\infty^2 = \tfrac12(2.63\times10^4)^2 = \boxed{3.46\times10^8\ \mathrm{J/kg}}$ . Positive — unbound, and enormously so: no planetary encounter inside the Solar System can absorb that much specific energy, so the object arrived unbound. Verdict: interstellar.

Step 2. Semimajor axis from (5): $a = -GM_\odot/2\varepsilon = -1.327\times10^{20}/6.92\times10^8 = -1.92\times10^{11}\ \mathrm{m} = \boxed{-1.28\ \mathrm{au}}$ . The negative sign is the hyperbola's bookkeeping.

Step 3. Eccentricity from the perihelion relation $q = a(1-e)$ , valid for all conics: $e = 1 - q/a = 1 - \dfrac{0.256}{-1.28} = 1 + 0.200 = \boxed{1.20}$ . This matches the fitted value that made headlines — the first solidly hyperbolic eccentricity ever measured for a Solar System visitor.

Step 4. Perihelion speed from vis-viva (7): $v_q^2 = GM_\odot\left(\dfrac{2}{q} - \dfrac{1}{a}\right) = 1.327\times10^{20}\left(5.22\times10^{-11} + 5.21\times10^{-12}\right) = 1.327\times10^{20} \times 5.74\times10^{-11} = 7.62\times10^9$ , so $v_q = \boxed{87.3\ \mathrm{km/s}}$ — briefly one of the fastest objects in the Solar System, all of it borrowed from the Sun's well and repaid in full on the way out, except the $26.3\ \mathrm{km/s}$ it brought from interstellar space.

Origins · 1 minWilliam Rowan Hamilton and how this was discovered

William Rowan Hamilton

1805–1865 · Trinity College Dublin

Hamilton found the orbit classification hiding in velocity space. In 1846 he asked what curve the tip of a Kepler orbiter's velocity vector traces as the orbit unfolds — the hodograph — and proved a jewel of a theorem: it is always a perfect circle, for every conic orbit. The classification then becomes a matter of where the origin of velocity space sits: inside the circle for an ellipse, exactly on it for a parabola, outside it for a hyperbola (the excess speed $v_\infty$ is the distance from origin to the circle's nearest point, and the unreachable arc corresponds to the asymptotes). The child prodigy who read Persian at ten had already reshaped optics and dynamics around his characteristic function — the seed of Hamiltonian mechanics — and his hodograph circle survives as the slickest known route to deriving the Kepler orbits, periodically rediscovered, most famously in Feynman's "Lost Lecture".

History

1609

— Kepler's Astronomia Nova: Mars moves on an ellipse with the Sun at a focus — the bound class discovered empirically, before anyone knew what energy was.

1687

— Newton's Principia derives all conic sections from the inverse-square law and notes comets may travel "in orbits very nearly parabolic" — the full classification in one stroke.

1705

— Halley applies Newton's method to 24 comets, recognizes the 1531, 1607 and 1682 apparitions as one bound object, and predicts its 1758 return — energy classification's first forecast.

1846

— Hamilton's hodograph theorem: the velocity tip of any Kepler orbit traces a circle, and the orbit's class is read from whether the origin lies inside, on, or outside it.

2017

— Pan-STARRS discovers 1I/'Oumuamua with e = 1.2: the first measured solidly hyperbolic visitor, and the energy argument certifies its interstellar origin.

Going further

Texts. Goldstein, Classical Mechanics, 3rd ed., ch. 3 (§3.3–3.7) · Morin, Introduction to Classical Mechanics, ch. 7 · Landau & Lifshitz, Mechanics, 3rd ed., §14–15 · Taylor, Classical Mechanics, ch. 8.

Provenance. Equations and classification verified against Goldstein 3rd ed. §3.3–3.7, Landau & Lifshitz Mechanics §14–15, Morin Ch. 7, Taylor Ch. 8. A verifier should recompute: (1) $r_0=L^2/\mu k$ and $E_{\min}=-\mu k^2/2L^2$ by minimizing $V_{\mathrm{eff}}$ ; (2) $e=\sqrt{1+2EL^2/\mu k^2}$ from the turning-point quadratic; (3) $a=-k/2E$ from $a=p/(1-e^2)$ ; (4) the stability bound $n>-3$ for circular orbits in $F\propto r^n$ . All numbers are standard symbolic textbook results; no empirical constants asserted. Worked-example numbers on this page ( $v_{\mathrm{esc}} = 11.19\ \mathrm{km/s}$ , apogee altitude $\approx 300\ \mathrm{km}$ at 8.0 km/s, $v_\infty = 4.34\ \mathrm{km/s}$ at 12.0 km/s; 'Oumuamua: $\varepsilon = 3.46\times10^8\ \mathrm{J/kg}$ , $a = -1.28\ \mathrm{au}$ , $e = 1.20$ , $v_q = 87.3\ \mathrm{km/s}$ ) are illustrative arithmetic recomputed from the boxed formulas.

Builds on: The effective potential · Reduction to the radial equation · Angular momentum conservation in central fields · The two-body problem and reduced mass. Leads to: Conic-section orbits · The Kepler problem · Kepler's first law · Orbital elements and parameterization.

How can an orbiting satellite have negative energy? Negative energy sounds unphysical

Nothing is unphysical about it — only the zero point is a convention. We choose $U(\infty) = 0$ , i.e. "at rest infinitely far away" scores zero. Anything closer in sits in gravity's debt: a bound satellite has less energy than the free reference state, hence a negative number. The magnitude $|E|$ is exactly the binding energy — what you must pay to lift it to rest at infinity. Kinetic energy alone is always positive, of course; negativity happens because the potential debt $-GMm/r$ outweighs it. And the convention has a real convenience built in: with this zero point, the sign of $E$ is the entire bound/unbound classification. Choose a different zero and the physics is identical but the dividing line moves off zero — which is precisely the pitfall the warning box flags.

Why do I need both E and L to know the orbit? Isn't energy enough?

They control different levers. The semimajor axis — the orbit's size, and via Kepler's third law its period — is fixed by energy alone: $a = -k/2E$ , no $L$ anywhere. The shape at that size is $L$ 's job through $e = \sqrt{1 + 2EL^2/\mu k^2}$ : at fixed $E < 0$ , maximal $L$ gives the circle ( $e = 0$ ), and shrinking $L$ stretches the ellipse toward the degenerate radial plunge ( $e \to 1$ ) without changing $a$ or the period at all. Two satellites can share an energy and a period while one rides a circle and the other dives nearly into the atmosphere. Conversely two orbits can share $L$ (same semi-latus rectum) at different energies — that is the conic family in the second figure. One number sizes the orbit; the other shapes it.

Do exactly parabolic orbits ever occur in nature?

Strictly, never — $E = 0$ is a single point on a continuum, a measure-zero target no real object hits exactly. What nature serves constantly is near-parabolic: long-period comets from the Oort cloud fall in from tens of thousands of au, where their orbital energy is a hair below zero, so they arrive with $e = 0.9999\ldots$ and fits that hover around 1 within measurement error. The label "parabolic comet" in catalogs means "bound, but with aphelion too distant to pin down". The distinction has teeth: a genuinely hyperbolic fit that cannot be explained by measurement error or planetary nudges — 'Oumuamua's $e = 1.2$ , Borisov's $e = 3.36$ — is the signature of an interstellar object, which is why orbit fitters fought over those error bars so carefully in October 2017.

The formula a = −k/2E gives a negative semimajor axis for hyperbolas. What does a negative length mean?

It is bookkeeping that keeps every conic formula working with one sign convention. Geometrically, a hyperbola has a real length scale $|a|$ — the distance from the vertex to where the asymptotes cross — and all the Kepler machinery survives with $a < 0$ inserted consistently: the perihelion is still $q = a(1-e)$ (negative times negative gives positive, since $e > 1$ ), vis-viva still reads $v^2 = GM(2/r - 1/a)$ where the $-1/a$ now adds speed, and $v_\infty^2 = -GM/a = GM/|a|$ . You can avoid negative lengths by writing separate hyperbola formulas with $+|a|$ , as some texts do, but then every equation forks into bound and unbound versions. The single convention with signed $a$ is why one line of code in an orbit propagator handles planets and 'Oumuamua alike.

Escape velocity doesn't depend on which direction I launch. Why not — surely launching straight up is different from sideways?

Escape is an energy criterion, and kinetic energy has no direction: $\varepsilon = \tfrac12v^2 - GM/r \ge 0$ cares only about speed. Launch at 11.2 km/s straight up, sideways, or at 45° (drag ignored) and every trajectory reaches infinity — they are different shapes of the same energy class, parabolas of different angular momenta. What direction does change is $L$ , hence the eccentricity and the path taken: a sideways launch curves away on a wide parabola, a vertical launch is the degenerate $L = 0$ straight line. One caveat hides in the idealization: direction matters in practice through what was excluded — drag path length, Earth's rotation bonus, and, for the vertical shot, whether you clip the planet on the way around. The pure two-body classification is genuinely isotropic.

A gravity assist changed a spacecraft's energy, but energy is conserved. Where does the energy come from?

From the planet — conservation holds, but the ledger spans both bodies. In the planet's rest frame the flyby is a textbook hyperbola: the spacecraft arrives and leaves with the same $v_\infty$ , energy unchanged, only the direction of the velocity rotated. Now step into the Sun's frame: the planet itself moves at ~13 km/s (Jupiter), and the rotated velocity now adds to the planet's velocity differently than the incoming one did. If the spacecraft departs more aligned with the planet's motion, its heliocentric speed — and energy — has increased; the planet loses exactly that energy and slows by an immeasurably small amount (its mass is ~10²² times larger). Voyager 2 gained about 10 km/s at Jupiter this way. The classification bookkeeping is frame-dependent: hyperbolic (unbound) relative to the planet, while the heliocentric class jumps from ellipse toward hyperbola.

Why is the energy in the classification built with the reduced mass instead of the satellite's actual mass?

Because the thing whose orbit is being classified is the relative coordinate $\mathbf{r} = \mathbf{r}_1 - \mathbf{r}_2$ , and after the two-body reduction that coordinate obeys Newton's law with the reduced mass $\mu = m_1m_2/(m_1+m_2)$ . The conserved quantities that classify it are the relative motion's energy $E = \tfrac12\mu\dot{\mathbf{r}}^2 + U(r)$ and angular momentum $L = \mu r^2\dot\phi$ — both computed in the center-of-mass frame. For a satellite around Earth, $\mu \approx m$ and the distinction is invisible. For a binary star with equal masses, $\mu = m/2$ , and using one star's lab-frame kinetic energy (which includes center-of-mass motion) misclassifies bound pairs as unbound. Rule of thumb: subtract the center-of-mass motion first; classify what is left.

Does the ellipse–parabola–hyperbola classification survive general relativity?

Its logic survives; its menu grows. Around a Schwarzschild black hole the effective potential gains a $-GML^2/c^2r^3$ term that bends the centrifugal barrier over into a crest, so alongside "bound" and "unbound" a third class appears: capture — orbits, even ones with $E > 0$ , that clear the crest and spiral in, something Newtonian point-mass gravity forbids for any $L \neq 0$ . Bound orbits also stop closing: $\Delta\phi > \pi$ generically, giving perihelion precession (Mercury's 43″/century was the first confirmation), and near the horizon exotic zoom-whirl orbits appear. The classification tool — draw $V_{\mathrm{eff}}$ , place the energy line, count the walls — transfers verbatim; only the curve changed. That is the real lesson of this article: the method outlives the potential it was learned on.