What was the first exoplanet discovered?
Well, y'know, that depends on what you mean by first. The first one announced that turned out to be real? Or the first data ever taken that showed the existence of one?
Or the first one announced, then retracted, then shown to have been real in the first place? Because arguably, that's the one.
OK, let's take a step back. An exoplanet is an alien world orbiting another star. Astronomers had been searching for them for decades, but they're really, really hard to spot directly. Planets are faint and stars are bright — that brightness ratio can be billions to one — and so far away that any planet appears to be extremely close to its host star in images, lost in the glare.
So astronomers turned to indirect techniques. One is called the reflex velocity method (or sometimes the radial velocity method, which means essentially the same thing and conveniently has the same initials). I've explained this before, but in a nutshell a planet orbits a star due to the star's gravity. But a planet has gravity too! So in reality they both orbit their mutual center of mass, called the barycenter. The planet makes a big circle because the star's mass dominates, and the star in reflex makes a little circle because the planet's mass is so much smaller.
As the star moves around in a circle, sometimes it's headed toward us, and sometimes away (this is called radial velocity). We can detect that motion as a Doppler shift in its light, a red shift when it heads away from us and a blue shift when it heads toward us.
This method is pretty good, but it's very tricky to measure. The star's motion is extremely small — sometimes just a few meters per second, which I can easily achieve on my bicycle, for example — and that's hard to detect. Plus there are a lot of other things that can happen that look like the star is moving around due to a planet, when really it's just sitting there all planetless.
That brings us to 1988, when a team of astronomers announced they had found what looked like a "substellar companion" — a planet — orbiting the star Gamma Cephei. The star is a little over 44 light years from Earth, and is actually a binary, two stars orbiting one another. One is an orange giant (a star that was once like the Sun but has used up all the hydrogen fuel near the end of its life and has swollen up hugely in size) and the other a far fainter red dwarf. They orbit each other at a distance of several billion kilometers, and take over 70 years to make one circuit.
The astronomers were observing the behavior of Gamma Cephei A, the orange giant. They very carefully measured its light over the course of several years, and could easily see the Doppler shift due to the red dwarf binary companion. But they also saw another signal as well: a 2.7 year variation in the star's velocity (by about 25 meters per second toward and away from us) that looked like it was due to a Jupiter-mass planet! Given the modern nomenclature given to planets, it would have the name Gamma Cephei Ab.
Hurray! Except not so much. Not long after, it was found that sometimes one certain type of orange giant star (called a K1 III star) can actually have internal physical processes that make it look just like a planetary companion tugging on the star, even when no planet is there. And guess what: Gamma Cephei A was classified as a K1 III. Ouch. Worse, this variability can take two or more years, with a velocity of around 20 meters per second, just what they found for their "planet."
D'oh! It looked like their planet was actually just Gamma Cephei A doing its own solo thing. They issued a retraction, saying the planet wasn't real.
But wait again! A few years later, in 2003, a group of astronomers which included some of the same ones from the original team got more accurate data using high-precision detectors. Combining all the data they had and analyzing it revealed something very interesting indeed: The variation was on a 906-day period (about 2.5 years), and it looked very stable. That was odd; the internal variability in orange giants actually changes with time, so the period changes too. Having it be constant at 906 days over observations made from 1982 to 2002 was suspicious.
As it happens, that internal variability also manifests itself in other ways, like in the star's magnetic field. Now, if you take a spectrum (breaking the light up into thousands of individual colors) you can identify different elements in the star's atmosphere. Calcium is strongly affected by a star's magnetic field, so a spectrum looking at calcium can tell you what's going on with the star's magnetism.
In the retraction, they found the light from calcium was variable on a 2.5 year period, which was close enough to their finding for the "planet" that it supported the idea that the variation was actually internal to the star, and not from a planet. However, when they found later that the period of the oscillations of the star were actually rock-solid at 906 days, they went back and rechecked the calcium data. They got a surprise: What they found there was that the period was actually 2.1 years, well short of the 2.5 year variation they saw in the velocity data.
And then another very cool thing happened. It turns out the star was misclassified. It wasn't an orange giant, a K1 III; it was a subgiant, a K1 IV star (subgiants are smaller than giants). These don't undergo those variations!
Boom. The star was stable, which meant the variations in the data they saw really were from a planet.
And just like that, the exoplanet Gamma Cephei Ab was real.
The data showed it has a mass of about 1.7 times that of Jupiter (a least; it could be much more) and orbited Gamma Cephei A at a distance of about 320 million kilometers (roughly equivalent to the asteroid belt around the Sun between the orbits of Jupiter and Mars).
Hurray! Even better, in 2018 follow-up observations of the system by yet another team of astronomers using Hubble were able to make even better measurements of the stars and planet and found the exoplanet to have a mass of at least 9.4 times that of Jupiter, but probably not much more.
So it turns out the planet Gamma Cephei Ab was seen in that original data! And that's the fun bit, where this plays into the historical record…
The first confirmed exoplanet was announced in 1992, orbiting a pulsar, the dead remnant of a massive star that had exploded (and in fact there are three exoplanets orbiting that pulsar). For a long time it was considered the first exoplanet discovered.
But Gamma Cephei Ab was announced in 1988! Since the retraction was itself retracted, it's actually the first confirmed exoplanet to be discovered.
And this gets weirder. Thrown into this mix of firsts is the fact that a spectrum of a white dwarf (the dead remnant of a star that was once like the Sun) was reexamined in 2014 and found to have clear evidence of a rocky planet that once existed around the star but was torn apart by the dwarf's immense gravity. That spectrum was taken in — get this — 1917!
So the first evidence of an exoplanet was found in 1917, but wasn't understood until 2014. In the meantime, Gamma Cephei Ab was found, lost, and found again, and in the meantime of that, other planets were discovered, too. That includes a massive planet orbiting the star HD 114762, found in 1989 (just after Gamma Cephei Ab) but not confirmed until 2012.
Good heavens. Literally.
You'd think the discovery of something as concrete as a planet would be clear-cut, but it's anything but. When you're on the bleeding edge of the technology available, looking at data right on the edge of detection, there are bound to be difficulties. Our lack of complete understanding of how stars behave is also a monkey in the wrench there. All in all, the confirmation of the first true exoplanet was a mess.
But I wonder, as I always do in situations like this… what else is out there? What other planets hide in ratty data, old data, observations that were discarded by astronomers at the time? How many planets have we seen but missed?