Exoplanet detection
A Socratic walk-through of exoplanet detection — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #How can we detect a planet we cannot see, orbiting a star we can barely resolve?
A planet emits no light of its own worth speaking of, and it sits beside something a billion times brighter. Even the star is, in most telescopes, a single unresolved point. So the honest starting position is that we cannot see the planet at all — and yet the catalogue now holds more than five thousand of them. If direct sight is off the table, what is left? Only this: the planet must do something to the star that we can measure. What could a small dark body possibly do to its star?
Reasoning it through
REASONING #Two things, it turns out, and they are the two workhorses.
The first is blocking. If the orbit happens to be edged toward us, the planet passes in front of the star once per orbit and takes a bite out of the light. How big a bite? The star is a disc, the planet a smaller disc, and the fraction removed is simply the ratio of their areas — the square of the ratio of their radii. Run the numbers and the method's whole character falls out. Jupiter is about a tenth of the Sun's radius, so a Jupiter crossing a Sun-like star dims it by roughly one percent. The Earth is about one hundred and nine times smaller than the Sun, so an Earth dims it by about 0.008 percent — eighty-odd parts per million. Both are detectable, but they are not remotely the same problem, and a telescope that comfortably finds the first may never see the second.
The second is wobble. We speak of a planet orbiting a star, but neither orbits the other; both orbit their shared centre of mass, so the star traces a small circle of its own. That motion alternately carries the star toward us and away, and the light we receive is shifted in wavelength accordingly — redder as it recedes, bluer as it approaches. Measure the shift in the star's spectral lines over time and a periodic wobble reveals a companion. Jupiter swings the Sun at about thirteen metres per second; the Earth manages roughly a tenth of a metre per second, slower than a walking pace, across a hundred and fifty trillion kilometres of vacuum. The first is well within modern instruments. The second is at or beyond their edge.
Now hold both methods up and ask what each misses, because that is the more interesting question. A transit requires the orbit to be nearly edge-on from our vantage — otherwise the planet passes above or below the star's disc and blocks nothing. The chance of that alignment is roughly the star's radius divided by the orbital radius, so a planet hugging its star at a twentieth of an Earth's orbit has perhaps a one-in-ten chance of transiting, while an Earth at an Earth's distance has about one in two hundred. And you must catch several transits to be confident, which means watching for several orbital periods: three years for a one-year planet, three weeks for a four-day one. The wobble method carries the same bias for different reasons — the velocity swing grows with planet mass and shrinks as the orbit widens, and a long period again demands years before one full cycle is visible.
So both methods, by entirely independent mechanisms, favour the same thing: big planets, close in, orbiting fast. Which is why the first exoplanet found around an ordinary star — 51 Pegasi b in 1995 — was a Jupiter-mass object completing an orbit in about four days, a configuration nobody had thought possible, and why the early catalogue looked like a galaxy of "hot Jupiters". Were hot Jupiters common? Not especially. They were visible. The strange catalogue was a portrait of the instruments, not of the sky.
The analogy
THE ANALOGY #Imagine trying to census the wildlife of a forest at night using only two instruments: a tripwire across one narrow path, and a very sensitive scale under the ground. The tripwire catches only animals that happen to use that path; the scale registers only animals heavy enough to press it. Compile the results and you will conclude the forest is full of large animals that walk in straight lines — and you will be wrong, not because the instruments lied, but because everything else left no signal.
The tripwire tells you almost nothing about what it caught, whereas a transit and a wobble together tell you a great deal — the transit gives the planet's radius, the wobble gives its mass, and dividing one by the other gives a density that distinguishes a rock from a gas ball.
Clarifying the model
THE MODEL #Three refinements connect the pieces.
First, the two methods measure genuinely different quantities, and neither is complete alone. A transit gives size but no mass. A radial-velocity wobble gives only a minimum mass, because a tilted orbit produces a smaller line-of-sight swing than the same planet seen edge-on, and the tilt is unknown. This is why the pairing is so valuable: a transiting planet is known to be nearly edge-on, so its wobble yields a real mass rather than a lower bound.
Second, neither signal is unambiguous alone. A dimming can be a background eclipsing binary blended into the same pixel; a periodic velocity shift can be produced by starspots rotating across a star's face. Confirmation is elimination, not a single detection.
Third, and most importantly: a bias is not a flaw to be apologised for but a known quantity to be corrected. Because transit probability and sensitivity limits can be calculated precisely, we can ask what true population must exist to produce the sample we got. That correction is how the field reached its most striking result — that planets between the size of the Earth and Neptune are the commonest kind, though they were the last to appear in the catalogue. The honest caveat: such corrections depend on how completely a survey's sensitivity is characterised, and the inferred abundance of genuinely Earth-like planets in Earth-like orbits remains the least certain number in the field, precisely because it sits where both methods are weakest.
A picture of it
THE PICTURE #How to readMove right for planets closer to their star and up for larger, heavier ones — the two properties both methods reward. The top-right quadrant is where signals are loud for either technique, and it is no coincidence the hot Jupiter sits there and was found first. Bottom-left is an Earth in an Earth-like orbit: small enough that the transit is faint, far enough that the wobble is slow and alignment unlikely. The distance between those two points, not any fact about planets, is what the early catalogue was really measuring.
What became clearer
WHAT CLEARED #We do not see exoplanets; we see what they do to their stars, and each method sees only one of those effects. Blocking gives a radius, wobbling gives a mass, and both effects grow with size and shrink with distance — so the catalogue was always going to fill from the top-right corner inward. Recognising that the early sky of hot Jupiters was an artefact of sensitivity rather than a fact about planets is what turned a biased sample into a usable census.
Where to go next
ONWARD #- How transmission spectroscopy reads a planet's atmosphere from the extra light absorbed during a transit.
- Why microlensing and direct imaging fill in exactly the wide-orbit region where the two workhorse methods go blind.
Key terms
TERMS #| Term | What it means |
|---|---|
| Transit | the passage of a planet across its star's disc as seen from Earth, dimming it by the ratio of their areas. |
| Radial velocity | the component of a star's motion along our line of sight, measured from the Doppler shift of its spectral lines. |
| Minimum mass | the lower bound on a planet's mass returned by radial velocity alone, since the orbital tilt is unknown. |
| Selection effect | a systematic difference between what a survey can detect and what actually exists, correctable when the sensitivity is well characterised. |
Every term the collection defines is gathered in the glossary.