Exoplanets: Beginnings

Many years ago, as I was writing up my thesis, my advisor burst into my office with the hot news.
Someone had announced a possible discovery of an extrasolar planet!

It seems strange now, but back then we really did not know if there were any other planets around other stars.
A lot of astronomers thought that it was quite likely that there were planets around other stars, maybe even most stars, but we had no data.

We were getting there, there were astrometric and radial velocity searches underway, with people like Walker, Griffin and Latham developing techniques which would clearly, eventually, maybe, in a decade or two, tell us about the presence of Jovians.
New high precision radial velocity surveys were being set up, in particular by Mayor in Switzerland, and Marcy in California.

Nature had published the possible detection of a low mass companion around HD114762, which is either a low mass brown dwarf or a high mass Jovian, depending on how you look at it.

Several other claims of exoplanet detection had been published, and then retracted, generally because of unsuspected systematic errors, or understandable but slight overinterpretation of marginal data.

Exoplanetes Exotiques

PSR B1829-10 was exciting, it was different, the data was good. 10 Earth masses, few month orbital period!
For some hours I contemplated suspending work on my thesis, which was on companions to pulsars and how they might come to be there, and doing something, anything, on this discovery.
Then, I played with some equations, and sent a cryptic e-mail to my advisor, noting that the signal looked worryingly like a buried second harmonic of the Earth’s orbit.
There were some e-mails between people above my paygrade, several people had noticed this, and then the retraction came.
The Solar System barycenter had not been correctly subtracted from the radio wave data, and the detection was invalid.
The retraction and handling of the related publicity, by the way, was exemplary – and precisely how discovery retraction ought to be well handled.

But, the stage had been sent, in particular the understanding that radio pulsar measurements were precise enough to detect planets – not just giant planets, but Earth mass planets.

A few months later, Alex Wolszczan announced the discovery of PSR B1257+12 with the first confirmed extrasolar planets orbiting it.
This was a tour de force of astronomical discovery.

Two planets, with masses of about 4 Earth masses, and third planet found later in the data, with a mass of 1/50th that of the Earth. That is still the lowest mass exoplanet found, and the sensitivity of pulsar searches remains about three orders of magnitude better than the current best optical radial velocity searches. Pulsar data has sensitivity equivalent to velocity variations of mm/sec!

PSR B1257+12

This was real, and in quick order the predicted planet-planet orbital perturbations were seen in the data.
There was a proliferation of theory papers, as always, coming up with increasingly outlandish theories for how a pulsar, and a recycled millisecond pulsars at that, could possibly have planets.

pulsar planets

There is now a good model for the formation, by Hansen and collaborators, with the planets formed within a narrow annulus of metal rich debris evolving viscously, probably from fallback from the supernova. This has implications, not just for formation channels for millisecond pulsars, but also for terrestrial planet formation in general (see also here).

A workshop was hastily put together to discuss the pulsar planets.
It was held at the end of april 1992, at Caltech. I was invited, and told to put something together for the meeting.
So I did: Planets in Globular Clusters.

Globular Cluster M4

Globulars are full of pulsars, good for timing, and there’s lots of stars. Maybe we could find planets there – globular stars are also low metallicity, but a quick calculation showed planets might still form in the more metal rich clusters, although maybe with lower efficiency than for metal rich stars, and the pulsar planets might form through a different mechanism from regular planets.
I showed 1257+12 planets might form and persist around cluster pulsars, though they might become dynamically perturbed by neighbouring stars; and, that a different channel existed where pulsars could steal planets from normal solar like stars, and that these would have a qualitatively different orbital distribution from planets formed in situ.

The meeting was very exciting, the Rodney King riots broke out the day before the meeting started. We drove down from Santa Cruz on the 101, into LA listening to the World Famous KROQ and the sudden traffic advisories about not exiting the freeways near downtown… The deli around the corner from the condo we stayed in was shot up that night, and the ambient atmosphere was a bit tenser than normally in SoCal.

At the meeting, Don Backer grabbed me, and showed me the data on PSR B1620-26 (Steve Thorsett had previously tagged the system in his PhD thesis as having anomalous timing behaviour) – it was immediately obvious that there could be a long period jupiter mass planet in the system, and I started arguing the case for such.

For several years, starting at the 1994 Aspen Center for Physics meeting on Pulsars, a minor debate raged over PSR B1620-26 – the initial solution was consistent with anything from a sub-jupiter mass planet to a black hole!
But as more data came in, the constraints tightened until it was clear the object orbiting in the system had to be sub-stellar.
Then a fortuitous encounter over a good south coast pinot noir at a Kavli Institute meeting lead to the realization that serendipitous Hubble data existed for the field, and in quick order this provided independent confirmation that the second pulsar planet was there.

PSR 1620-26b

The oldest planet, and still my favourite

It also confirmed, what we were beginning to suspect, that planets really are ubiquitous.
Not a problem.
I am also sure that more pulsar planets will be discovered.

There are a lot of interconnects in astronomy.
In 1999 a group of us, lead by Ron Gilliland and Tim Brown, got some (ok, a lot of) Hubble time to look for transits of planets in a globular cluster – 47 Tuc specifically.
(Don’t know what fool suggested that:
“Observation is most probable in the rich high-density precore collapse clusters such as 47 Tuc.”).

We found no planets in 47 Tuc…
Which was a puzzle. Which I mentioned to my then graduate student.
We did get a lot of practise doing high precision relative photometry, showed that space-based relative photometry really had the necessary precision, and that false positives due to grazing binaries, variables and blends could be eliminated with high confidence.

A decade later, the student even figured out why – Solar Tides. The planets close enough to the star for the Hubble search to have detected them (two transits less than 4 days apart), would have been destroyed, not by the tide of the star on the planet – we checked that! – but by the subtler tide of the planet on the star, the slowly rotating old star.

Dissolving planets…

Nice piece of work.

The original exoplanet discovery has been somewhat overshadowed by the later discovery of exoplanets around solar type stars – that was pretty exciting also, but that is another story.


  1. #1 Markk
    October 4, 2010

    Reading this I feel another age of miracle and wonder moment similar to what I felt about reading about modern genetic understanding – from about the same time frame – my adult lifetime. Just wholesale areas of new knowledge. In my adult lifetime … wow.

    Did anybody in, say, 1982 (when I got my Bachelors) to 1989 have any models or proposals anything like we see now in exoplanets? Or was there simply no model at all? It is amazing how cruddy we are extrapolating past data. It makes me very queasy about all these bubble universe or even holographic theories when there so much stuff that we likely have no idea about between where we are now and any kind of evidence for them.

  2. #2 Steve T.
    October 4, 2010

    It was a good thing that I edited the proceedings for the 1992 Planets Around Pulsars meeting, or I’d have missed it all, because the excitement of the riots pretty much crowded the science out of the minds of those of us who had to worry about how to feed scores of scientists when our conference banquet site abruptly closed, and where we were going to house people who couldn’t get from campus to their hotels. Since then, I only organize meetings a long ways away from home: Aspen, Crete, Santorini, Moriond, ….

    Thanks for the historical summary! I always love to see that image of Methuselah, even if it is only an artist’s conception.

  3. #3 Steinn Sigurdsson
    October 4, 2010

    @Steve – ah, yes, good times… I always felt sorry for the couple of people who to stuck at LAX coming in and didn’t get there till the last day of the meeting.

    @Markk – there was surprisingly little speculation on exotic planets, outside science fiction at least – even with a previous claimed detection around PSR B0329+54.
    General theory of Solar System formation was relatively mature, but there were hints of problems, as note, eg by Ward and Goldreich & Tremaine, but in retrospect the migrating planets and richer exoplanet morphology could have been predicted robustly.

    Theorists really are frightfully conservative, predictions are rare, and observers like to tease us.

  4. #4 Dave Bacon
    October 4, 2010

    Great post. Reminds me of why I almost became an astrophysicist (somehow the interstellar tedium isn’t useful in quantum computing :))

  5. #5 Steinn Sigurdsson
    October 4, 2010

    It is all one – my Ph236 essay for Roger was on information and black holes, and non-conservation of weakly conserved quantities.
    And Seth Lloyd made some very distracting arguments at Murray’s seminars, quite fascinating stuff.

  6. #6 Lab Lemming
    October 5, 2010

    So where are all the white dwarf planets?

  7. #7 Steinn Sigurdsson
    October 5, 2010

    @Lab – We’ve looked. There is one good candidate out there, pending followups last I heard. Not one of ours, unfortunately.
    Several searches underway that will be complete down to few jupiter masses, with some having better sensitivity.
    Direct imaging of white dwarf planets ought to be a medium term industry, especially if JWST flies.

  8. #8 Matt Kenworthy
    October 5, 2010

    I’ve been giving talks on exoplanets, pointing out the original pulsar discoveries, when someone asked the (obvious) question: “Why are there no more pulsar planet discoveries since then?”

    I could think of plausible answers, but I don’t know what the real answer is. Is it the lack of millisecond pulsars on the sky? There are no ‘stable’ pulsars on the sky?

    I assume everybody chased after more pulsar planets, but I’ve not heard anything since….

  9. #9 Lassi Hippeläinen
    October 5, 2010

    “…noting that the signal looked worryingly like a buried second harmonic of the Earth’s orbit.”

    Sometimes you have to do some navel-gazing…

    During my short involvement in space engineering in early nineties, there was some discussion about quantized red shift. (Haven’t heard anything about that for a long time.) My first reaction was to suspect the equipment. When you dig deep enough in your observations, you can see the pixels of your detector.

    “…a different channel existed where pulsars could steal planets from normal solar like stars…”

    Some planets will feel offended if you call them satellites. True wanderers (“planetes”) are not bound to a single star, they wander all around the globular cluster…

  10. #10 Steinn Sigurdsson
    October 5, 2010

    @Matt – there are about 3,000 pulsars now, including several hundred millisecond pulsars; limitations on planet discovery include a) equipment downtime for a while, b) lack of high cadence followup, c) faintness, and d) lots of the MSPs are in globulars where timing is more complicated

    I am reasonably confident there will be more, interesting, pulsar planet discoveries, both in clusters and the field.

    @Lassi – quantized redshift has pretty much vanished, it was pondered upon but never really went anywhere.
    There are certainly free-floating planets in clusters and stellar associations – we don’t know how efficiently planets are formed in globulars, but there could be more planets in them than stars; I’d not be surprised if the number of free floating planets in the galaxy is also comparable to or larger than the number of stars.
    I think it is pretty clear that the number of bound planets is, on average, several per star.

  11. #11 andy
    October 5, 2010

    It is kind of weird that pulsar planets have ended up so forgotten, especially in all the excitement over super-Earths. I suspect we’ll be having “smallest known exoplanet” claims for objects over an order of magnitude too massive for quite a while to come.

    Regarding free-floating planets/sub-brown dwarfs in globulars, wouldn’t they mostly have been ejected by now?

  12. #12 Steinn Sigurdsson
    October 5, 2010

    @andy – significantly not ejected, they’d have mass segregated to the outskirts with some significant differential tidal stripping, but timescale for everything become long in the outskirts of globulars so you ought to retain a lot of the planets in the outer half of the cluster – actually true for both the free-floaters and the bound planets.

  13. #13 Lab Lemming
    October 6, 2010

    Haven’t globular transit surveys all come out negative?

  14. #14 Steinn Sigurdsson
    October 6, 2010

    There was the Hubble transit survey of an inner portion of 47 Tuc and a survey of the outskirts of 47 Tuc, from Australian, ground based. A couple of surveys of nearby open clusters.
    All negative – the “dissolving planets” paper reconciles the issue, planets with short orbital periods are tidally destroyed in globulars, simply because of the longer timescale, but still there to be seen in the field.

  15. #15 Lab Lemming
    October 7, 2010

    Are any of the Kepler released candidates in the open clusters in its FOV?

  16. #16 JohnD
    October 7, 2010

    Don’t know if any of the candidates are in the clusters, but 4 clusters should be in the Kepler FOV. The trick will be if enough planets are discovered. Also, recent work might suggest that null results from open cluster surveys are still consistent with field transit surveys, see arxiv:1009.3013

  17. #17 Nick
    October 7, 2010

    I may have missed it, but how far away is this planet? And how long would it take for a shuttle to get to this, if it was to be possible?

  18. #18 Steinn Sigurdsson
    October 7, 2010

    Which of the planets?
    1620-26 is about 7,000 light years away.
    The others are tens to hundreds of lights years away, mostly.

    The shuttle is a Low Earth Orbit vehicle, it does not go further than few hundred kilometers from the Earth.

    There are about 10 trillion kilometers in a light year.

  19. #19 Lab Lemming
    October 8, 2010

    I assume that Debes et al. assume that hot jupiter hosts rotate at less than once per 4 days (like the sun)- is that sensible?

    Given that, wouldn’t their model predict hot brown dwarfs and hot super jupiters to raise larger buldges and decay faster?

  20. #20 JohnD
    October 9, 2010

    @Lab Lemming

    You do need a star that is rotating slower than the mean motion of the planet, and more massive things will decay faster.

  21. #21 Erica
    October 13, 2010

    What are your thoughts about the Godilocks planet discovered recently? Do you think that there is a possibility for life outside of Earth?

  22. #22 nick
    November 15, 2010

    still no life supporting – mass, radiation, chemistry etc

    Drake still gives minimal numbers

  23. #23 coolstar
    December 21, 2010

    “Theorists really are frightfully conservative, predictions are rare, and observers like to tease us.”
    That was meant in jest right? because it is hilarious!