The number that survives the encounter

In 1889 a comet was found that nobody could place. Its orbit matched no comet on record, and there was no obvious reason to think it had been seen before — except that a search of the archives turned up a comet, observed years earlier, that had since passed close to Jupiter. If the two were the same object, everything about its orbit had changed: its period, its distance from the Sun, its shape, the direction of its long axis.

Félix Tisserand’s answer was that one combination of those quantities cannot change, however violent the encounter, and that computing it settles the identification. It is still how a comet is recognised after an encounter has destroyed its orbit, and the same expression, read forwards rather than backwards, is the map of everywhere a spacecraft can reach without firing an engine.

Where it comes from #

The Tisserand parameter is the Jacobi constant in disguise, and seeing that is most of the understanding.

In the circular restricted three-body problem — Sun, planet on a circular orbit, and a massless third body — nothing is conserved in the inertial frame, because the planet’s motion makes the potential time-dependent. In the rotating frame the potential is static, and there is one integral: the Jacobi constant

$$C_{\rm J} = 2\Omega(x,y) - v_{\rm rot}^2,$$

which is the quantity whose level sets say where a body cannot go. Rewrite it in terms of the small body’s heliocentric orbital elements, drop the terms of order the planet’s mass, and it becomes

$$T = \frac{a_{\rm p}}{a} + 2\sqrt{\frac{a}{a_{\rm p}}(1-e^2)}\,\cos i.$$

Every element in that expression changes during an encounter. The combination does not, because the Jacobi constant does not, and the Jacobi constant does not because the rotating frame does not know what time it is.

The test, and it is a real one #

An invariant that is asserted rather than checked is worth nothing, so the claim deserves an integration that could refute it.

The closest approach in that run is 0.03 of Jupiter’s orbital radius — 0.16 AU, a distant pass by planetary standards — and it changes the semi-major axis by a factor of two and a half. That is the ordinary violence of a gravity assist, and it is why an invariant that survives it is worth having — the same violence a spacecraft is deliberately aimed into.

What it sorts #

The comet population divides at two values of the parameter, and the division is not a convention adopted for tidiness — it is a statement about dynamics that turns out to be a statement about origin.

$T > 3$: no encounter with Jupiter is possible, because $v_\infty^2 = (3-T)v_{\rm J}^2$ would be negative. Most asteroids are here. So is 2P/Encke, at 3.03, which is why Encke is the awkward case in every classification — dynamically it is an asteroid and physically it is a comet.

$2 < T < 3$: the Jupiter-family comets. Low inclination, short period, and dynamically dominated by Jupiter. They come from the scattered disc beyond Neptune, and they are passed inward by a sequence of encounters, each of which moves them along their own contour.

$T < 2$: the nearly isotropic comets — Halley-type and long-period. Their approach speed exceeds Jupiter’s orbital speed, they arrive from every direction including retrograde, and they come from the Oort cloud. 1P/Halley itself has $T = -0.61$, and the sign is the retrograde $\cos i$ doing its work.

That last division deserves a moment, because it is the one that carries the most information. A comet’s Tisserand parameter is not a label attached at discovery — it is the surviving trace of where the object spent the time before anybody saw it. A body handed inward from the scattered disc arrives with an inclination of a few degrees and a $T$ just under three, because that is what a sequence of Jupiter encounters starting from a low-inclination reservoir produces. A body dropped in from the Oort cloud arrives from a direction chosen at random, so its $\cos i$ is uniformly distributed and its $T$ scatters across the whole range including the negative one. Two reservoirs, two distributions of one number, and the number is measurable from an orbit fitted to a few weeks of astrometry.

Read forwards, it is a map #

The same expression, used the other way, is the central tool of interplanetary mission design, and the change of viewpoint is a change of which quantity is unknown.

Write the small body’s heliocentric velocity at the planet’s own orbital radius as $\mathbf v = \mathbf v_{\rm p} + \mathbf v_\infty$, with $V = v_\infty/v_{\rm p}$ and the pump angle $\alpha$ between the two. Then

$$\frac{a}{a_{\rm p}} = \frac{1}{1 - V^2 - 2V\cos\alpha},\qquad 1-e^2 = \frac{a_{\rm p}}{a}\,(1+V\cos\alpha)^2,$$

and substituting both into the definition gives $T = 3 - V^2$ identically. So sweeping the pump angle traces out a contour, and the contour is the set of every orbit reachable at that arrival speed.

The bending limit is worth its own line, because it says which planet is worth visiting. The maximum turn satisfies $\sin(\delta/2) = 1/(1 + r_p v_\infty^2/\mu)$, so what matters is $\mu/r_p$ at the closest approach the body permits — its surface gravitational potential, in effect. Jupiter’s is enormous and its cloud tops are far from its centre; the Earth’s is small. That ratio, not the planet’s mass alone, is why every mission to the outer system is designed around Jupiter.

What was actually measured #

The parameter’s first use was an identification and it is worth stating as a measurement.

Comet Lexell, discovered in 1770, passed 0.015 AU from the Earth — the closest cometary approach on record — and was never seen again. It had passed close to Jupiter in 1767, which put it on the orbit that brought it near the Earth, and passed Jupiter again in 1779, which removed it. Le Verrier and later Tisserand showed that the pre-1767 and post-1779 orbits, entirely different in every element, share a Tisserand parameter, and the identification of the three arcs as one object rests on that agreement and on nothing else.

The modern version is routine and continuous. Every comet and asteroid in the Minor Planet Center’s files carries a computed $T_{\rm J}$, and the number is used to decide which dynamical class an object belongs to — a decision that would otherwise require integrating its orbit backwards through encounters that are chaotic and therefore not reliably integrable. A conserved quantity is the only way to say anything about a trajectory that cannot be computed, and this is the clearest example the solar system provides.

There is a third use, and it is the one that turns the parameter into a survey instrument. Because $T$ is preserved through encounters but not through the non-gravitational forces a comet feels — outgassing pushes on a nucleus and changes its energy — a population whose Tisserand parameters have drifted measurably from their expected values is a population whose members have been active. That comparison has been made for the Jupiter family and the drift is small, of order a thousandth, which puts an upper bound on how much rocket effect the average nucleus has experienced over its dynamical lifetime. An invariant is useful for what breaks it as well as for what preserves it, and here the breakage is a measurement of cometary activity made without observing any activity.

The Kirkwood gaps are the counter-example worth having #

It is tempting, having found an invariant that sorts a population, to read every structure in the small-body distribution as a Tisserand effect. Most of them are not, and the distinction is instructive. The two do interact, and the interaction is what actually delivers a Jupiter-family comet. A body pushed out of a resonance acquires an eccentricity large enough to cross Jupiter’s orbit; from that point onwards it is on a Tisserand contour and is passed inward by encounters. The resonance sets the trajectory going and the invariant governs everything after.

The encounter as a rotation on a sphere #

The derivation above treats the invariant as an algebraic accident of the Jacobi constant. There is a second way to see it that makes it obvious rather than surprising, and it is the form in which the parameter is actually used for statistical work on comet populations.

Öpik’s encounter theory reduces the whole event to a single geometrical statement. Far from the planet, the small body’s heliocentric velocity is the planet’s velocity plus an excess $\mathbf v_\infty$. Close in, the planet’s gravity turns $\mathbf v_\infty$ and does nothing else to it. So the encounter is a rotation of one vector, and the small body’s state before and after differ only in where that vector points on a sphere of fixed radius.

Two angles then describe everything. The pump angle $\alpha$ between $\mathbf v_\infty$ and the planet’s motion decides the new semi-major axis, since it decides how much of the excess adds to the heliocentric speed. A second angle around the planet’s velocity direction — the crank angle — decides how much goes out of the plane, and therefore the new inclination. An encounter picks a point on that sphere; the geometry of the approach decides which point; and every orbit corresponding to every point has the same $T$ by construction.

Nothing here needs the restricted problem or its rotating frame. The invariance follows from one sentence — a flyby changes the direction of the excess velocity and not its magnitude — and the Jacobi constant is the reason that sentence is true rather than an independent fact. Written this way the theory also becomes a machine for generating encounter statistics: distribute the impact parameters uniformly, rotate accordingly, and integrate over the sphere, which is how the transfer rates between the scattered disc and the Jupiter family are computed without integrating any individual orbit.

Where the classification frays #

A boundary drawn in one number will misplace objects, and the interesting ones are exactly those it misplaces — because a dynamical class and a physical class are different things, and the parameter measures only the first.

Encke. At $T_{\rm J} = 3.03$ it cannot encounter Jupiter at all, and yet it is unmistakably a comet: it has a tail, it outgasses, it has a nucleus of ice and dust. Whatever brought it to its present orbit is not available to it now, and the usual account involves non-gravitational forces acting over thousands of years to move it across the boundary from the Jupiter family — one of the few places where the rocket effect of outgassing is invoked as a dynamical mechanism rather than a nuisance.

The damocloids. Objects with $T_{\rm J} < 2$, on Halley-type orbits including retrograde ones, showing no activity whatever. They have the dynamics of long-period comets and the appearance of asteroids, and the accepted reading is that they are extinct cometary nuclei — the same bodies after their volatiles are exhausted or buried.

The main-belt comets. Objects at $T_{\rm J} > 3$, in the asteroid belt, on ordinary asteroidal orbits, that develop tails. Some are outgassing genuine ice; others are shedding dust after an impact or from rotational break-up, which produces a tail with no volatiles involved at all.

Each of these was a surprise in the same way. The parameter sorts by history and the telescope sorts by appearance, and the two disagree at the edges — which is not a failure of either but the reason both are quoted. An object’s $T_{\rm J}$ says which reservoir its orbit is consistent with; its spectrum says what it is made of; and the cases where they conflict are where the population’s evolutionary sequence becomes visible.

The map’s two readings — the contours themselves and the reachable set a single flyby delivers — are each worth drawing at a second arrival speed, since the arrival speed is what the launch fixes and everything else follows from it.

Moving between contours on purpose #

The mission-design half of the essay says a spacecraft cannot leave its contour without an engine. The interesting consequence is what happens when it uses one, because the cheapest place to spend fuel is not where intuition puts it.

A resonant flyby is the first ingredient. Leave a planet on an orbit whose period is a simple ratio of the planet’s — three to two, say — and the spacecraft returns to the same place at the same time as the planet, with no cost at all, ready for another encounter. A sequence of these walks along the contour in steps, each turning $v_\infty$ a little further, and Galileo’s two Earth passes and Cassini’s pair at Venus were exactly this.

The second ingredient is $v_\infty$ leverage. A small burn at aphelion, far from the planet, changes the spacecraft’s heliocentric energy by very little — but it changes where the spacecraft crosses the planet’s orbit, and therefore the excess velocity at the next encounter, by a great deal. A few tens of metres a second applied at the far end of a resonant orbit can buy hundreds of metres a second of $v_\infty$ at the return, which is a jump between contours at a fraction of the price of making the jump directly.

That is the manoeuvre behind the deep-space burn in a $\Delta V$-Earth-gravity-assist trajectory, and it is why the burn is placed where nothing appears to be happening rather than at the flyby, where the geometry is most dramatic. The Tisserand map prices the jump and the leverage decides where to pay for it, and between them they explain the otherwise baffling shape of every modern outer-planet trajectory: a spacecraft launched towards Jupiter that first goes inward to Venus, twice, and then twice past the Earth.

There is a further reason the leverage manoeuvre is placed where it is, and it is worth stating because it is the same argument as the burn’s efficiency. A change in speed applied where the spacecraft is moving slowly changes the orbit’s shape by more, per unit of propellant, than the same change applied where it is moving fast changes the orbit’s size — so the two effects a designer wants, a large change in arrival geometry and a small change in energy, are both maximised at the same place.

That is also why a designer’s first question about a candidate route is where its deep-space manoeuvres fall rather than how many flybys it has. A trajectory with four encounters and no burns is cheaper than one with two encounters and a burn at the wrong end of an orbit, and the map is what makes the comparison before either is integrated.

Where the model stops #

Three approximations are buried in the derivation, and each is visible in real data.

The parameter assumes the planet’s orbit is circular. Jupiter’s eccentricity is 0.048, and that alone makes $T_{\rm J}$ wander by a few thousandths over a comet’s orbit — enough that objects near a family boundary move across it.

It drops terms of order the planet’s mass, $10^{-3}$ for Jupiter. That is the size of the residual drift seen in careful integrations, and it is why $T_{\rm J}$ is quoted to two or three decimals and not more.

And it assumes one perturbing planet. A comet that encounters Saturn as well has a Tisserand parameter with respect to each, and neither is conserved through an encounter with the other. Objects on Centaur orbits between Jupiter and Neptune are handed between planets in exactly this way, and their dynamical lifetimes — a few million years — are set by how quickly the handovers scramble whichever invariant was holding. And the resonant structure that sits on one contour, at an arrival speed between the two.

One more reading shows what the invariant is worth once a spacecraft is on it.

Where this ladder goes next #

This rung establishes the invariant and what it sorts. The rungs above it go in two directions.

One follows the mission-design use into its own machinery: how a sequence of assists is planned as a walk along contours, how the intersections between contours at different planets define the transfer opportunities, and how the whole thing is priced against the alternative of simply burning fuel — which is a comparison the low-thrust case settles differently again.

The other follows the invariant into the regime where it stops being one. The Jacobi constant is exactly conserved and the Tisserand parameter is not, and the gap between them is where the interesting dynamics of the Centaurs lives. There is also a family of trajectories for which the Jacobi constant alone decides everything — the ones threading the necks at the Lagrange points, which transit or fail to transit according to a boundary with no width at all.