Less than a week remained before Andrew Lyne’s scheduled talk at the American Astronomical Society’s January meeting in Atlanta. There the British astronomer planned to discuss his discovery of the first solid evidence for a planet orbiting another star, some 30,000 light-years from Earth.
Lyne’s discovery was, in every sense, a very big deal. When first announced last July, it made front-page news the world over. His findings were especially surprising to astronomers because the planet he’d found was circling a type of star called a pulsar--a small, dense, rapidly spinning core sometimes left behind after a supernova explodes. Since supernova explosions are among the most violent events in the universe, any planet unfortunate enough to be in the vicinity of one should have been instantly annihilated or blown far away from the star. Nevertheless, the evidence for this planet’s existence seemed impeccable.
What Lyne and his colleagues at the University of Manchester in England actually observed when they looked at the pulsar was not a visible body orbiting the star remnant--that would have been impossible to see at such a great distance. Rather, they detected a cyclical variation in the signal coming from the pulsar itself. As a pulsar spins, it emits a tightly focused beam of radio waves, like the beacon from a lighthouse. Most pulsars emit their signals with a regularity that rivals the best atomic clocks; the one Lyne observed, however, appeared to speed up and slow down in a regular six-month cycle. Although the team tried hard to come up with an alternative explanation, in the end they had to conclude that the fluctuation in the signal was being caused by the gravitational pull of a large planet, one with about ten times the mass of Earth. They did so with some reluctance, certainly: as they noted in their paper, this was the first detection of a planetary-sized body outside the solar system.
Their fellow astronomers reacted cautiously at first, but most soon agreed that a planet was the most reasonable explanation for the pulsar’s signal. Yet now, getting close to the Astronomical Society meeting, something seemed wrong. reviewing data from the most recent observations, Lyne noticed a small error in the assumed position of the pulsar. He did some calculations to correct for the mistake.
Five minutes later I froze in horror, he says. I saw the planet evaporate. Then it was only a matter of time before I figured out exactly what had happened and why.
What had happened was a simple human error. Although the British team detected the planet only last summer, Lyne discovered the pulsar itself seven years ago. At that time he got what seemed to be a reliable fix on the distant body’s position, and he used these measurements to make his calculations concerning the existence of the planet. But the measurements he took in 1985 turned out to be imperfect.
Ordinarily, astronomers will determine the location of a pulsar by using radio telescopes to analyze the strength and direction of the incoming signal. They will then check their work against mathematical models designed to predict when future signals should arrive at Earth. If subsequent observations match these predictions, they know they’ve figured correctly. But because of an oversight (this pulsar was one of some 300 they had observed over seven years), Lyne’s team eliminated this step.
The problem was exacerbated when the astronomers later tried to refine their computations by correcting for Earth’s orbit around the sun; this is a routine computational precaution, since Earth’s orbit is slightly elliptical and could therefore cause errors to appear in a pulsar’s signal. But since the first calculation was imprecisely done, so was this one. When corrections were made in both steps, the signal stabilized.
It was a bitter pill to swallow, says Lyne’s colleague Matthew Bailes. Within days Lyne and Bailes submitted a retraction to the British journal Nature, in which they had published the paper announcing the discovery. An editorial commenting on the admission says that Lyne and Bailes should take pride in the directness of their acknowledgment this week of their mistake, calling their conduct a model of how these things should be done.
When Lyne ultimately spoke at the meeting in Atlanta, it was to discuss the upcoming publication of the retraction. With the journal due out the next day, however, most of the astronomers already knew the bad news. After Lyne finished his address, he received a long ovation.
We applauded him for his honesty, says Stan Woosley, an astronomer from the University of California at Santa Cruz. He is a very careful scientist. He recognized his error and reported it promptly. You could contrast this to the cold-fusion fiasco.
Ironically, at the same meeting Alexander Wolszczan, a radio astronomer at Arecibo Observatory in Puerto Rico, gave a talk about two-- and possibly three--planets orbiting a different pulsar that he and Dale Frail, an astronomer at the National Radio Astronomy Observatory in Socorro, New Mexico, had detected. Wolszczan and Frail had published their find only one week before, also in Nature, but the astronomical community had known about it for months. Would Lyne’s retraction now damage Wolszczan’s work?
Not according to Woosley: The kind of error that influenced Lyne’s work is not present in Wolszczan’s, he says. There is almost no reservation in the astronomical community to accepting this for what it is announced to be, that is, at least two planets around a pulsar. Woosley’s confidence comes mostly from the knowledge that Wolszczan took additional readings of the pulsar’s position--which is located 1,300 light-years from Earth--and used equations that automatically correct for the ellipticity of Earth’s orbit.
Wolszczan found his planets using Arecibo’s 1,000-foot-diameter radio telescope while searching the sky for pulsars in February 1990. he too made his discovery by analyzing fluctuations in the arrival times of pulses from one pulsar. But in this case the fluctuations occurred in two separate cycles: 66 days and 98 days. These cycles are the orbital periods of two planets, Wolszczan says, one with a mass 2.8 times Earth’s and a second 3.4 times greater. Other, subtler flutters in the pulsar could even indicate a third planet.
Last September I finally decided there was no way to find any other explanation but planets for these signals, he says. I sat on it for another month and kept observing the pulsar and using the planetary model, based on one year’s worth of observations, to predict observed arrival times. The model kept producing correct predictions. There are, of course, still risks involved, and I guess the British guys have proved that very dramatically. But at the same time one cannot be overly conservative. That is not how progress is made.
Perhaps the most intriguing feature of Wolszczan’s discovery is that in a few years he may be able to prove the planets’ existence beyond any reasonable doubt. If there are two planets circling the pulsar, he says, they should interact gravitationally. Wolszczan predicts that this should show up as yet another systematic change in the pulsar’s signal every five and a half years.
That would constitute a really one hundred percent solid proof that we do have planets around a pulsar, he says. It would be difficult to come up with an effect that would ideally mimic planetary motions.
Another crucial distinction between this discovery and Lyne’s is that Wolszczan’s planets orbit a millisecond pulsar. It is easier to explain the presence of planets around a millisecond pulsar than around an ordinary one, says Wolszczan.
Millisecond pulsars are especially fast pulsars that complete a single revolution once every few thousandths of a second; known millisecond pulsars have rotation rates ranging from 1.6 to 6.2 milliseconds and are about a billion years old. Average pulsars take between half a second and a second to complete a spin and may slow down even further after a few million years. Of the 500-odd pulsars discovered, fewer than 50 are millisecond pulsars.
Millisecond pulsars spin so fast, astronomers believe, because they were once part of a binary star system. The star that will become a pulsar explodes, leaving behind a compact, extremely dense core, called a neutron star, that measures as little as ten miles across. All stars rotate slowly, but as the core contracts into the even-smaller pulsar, the rate of spin increases, much as a twirling skater moves faster and faster as he pulls in his arms.
If a companion star is hovering nearby, the enormous gravitational tug of the pulsar may start drawing off some of its mass. As matter spirals down onto the dense, spinning body, it transfers some of its momentum to the pulsar, thus increasing the spin rate even further. After several hundred million years the pulsar may consume so much mass from its partner--perhaps all of it--that it achieves the millisecond spin rate. In the past three years astronomers have detected two pulsars that are cannibalizing their companions in this way. The first of these unusual stars has been dubbed the black widow pulsar.
Such pulsars would be more likely to include planets than ordinary pulsars simply because they have the raw material on hand. Typically, not all of the matter drawn from the companion star is sucked into the pulsar; some of it begins to orbit the pulsar instead, and eventually this may accrete into a distinct body. On the whole, Lyne says, it’s somewhat less surprising that a millisecond pulsar, rather than an ordinary pulsar, should have planets, because we do believe that at some stage it had circumstellar material.
Even though astronomers will now first look at other millisecond pulsars for evidence of planets, Lyne remains convinced that the search for planets around pulsars similar to the one he observed should continue. We should not let this ‘undiscovery’ discourage us from looking for planets around normal pulsars, he says. Other astronomers agree. Just because one planet has been ruled out doesn’t mean there aren’t others, says Woosley.
There is no shortage of models explaining how an ordinary pulsar could be accompanied by a satellite. While most of them differ on the finer points of the planet-forming process, they all agree on at least one thing: the orbiting body could not form before a supernova explosion and still be around afterward.
What seems to be very unlikely is that these objects could survive a supernova explosion, says Doug Lin, a colleague of Woosley’s at the University of California at Santa Cruz. Then the question is, how do you come up with a scenario that would explain the formation of planets in a post-supernova event?
Woosley, Lin, and University of California colleague Peter Bodenheimer have published a paper of their own, arguing that planets could form around a normal pulsar from the debris the supernova left behind. Astronomers believe that when a supernova explodes, its outer layers jet away, perhaps at speeds faster than a tenth the speed of light. While the outer layers are exploding, the innermost regions of the star implode; in effect, they are moving inward at negative velocity.
But in between those two limits you have material with a whole range of velocities, from zero to twenty thousand miles per second, says Woosley. Some of this matter would have too much angular momentum to fall back down to the young pulsar right away and could become gravitationally trapped around it, forming a large orbiting disk. Friction from the pulsar would then cause the inner portion of the disk to slow down and eventually fall into the star remnant; the outer portion would move fast enough to increase the circumference of its orbit. In about a million years it would cool and coalesce into planets.
On the surface, this scenario sounds plausible enough, but Johns Hopkins astronomer Julian Krolik believes it’s unlikely. It’s one of those pictures that might happen, Krolik says. But it’s hard to say why you would expect it to happen. There has to be enough mass in the explosion, with energy just in the borderline of escape. And then it has to form a disk in the right place in order to put a planet in that position. So while there is nothing that violates physical law in anything they suggest, there isn’t any particularly compelling reason to think that it should work out. Krolik adds that if planets form from supernova debris, we should see many more of them circling pulsars.
In his theory’s defense, Lin says that planets may indeed be orbiting other pulsars, but if the planets are small enough, we would have no way of detecting them. Low-mass planets would not be picked up by the current generation of surveys, he says. Even with a ten-Earth-mass planet, the detection is marginal. If the planet is much lower in mass, you would never pick it up.
Krolik has his own ideas about how a planet could form around a normal pulsar. With some modifications, he says, the same model could apply to a discovery like Wolszczan’s. Krolik’s model begins with a binary star system. One of the stars goes supernova, but the other star withstands the explosion. Stars are pretty robust when it comes to surviving a supernova, he says. close binaries in which one is a normal star and one has gone supernova exist in considerable numbers.
After exploding, the first star would form a neutron star (later to become a pulsar) and, like all young neutron stars, begin emitting tremendous amounts of radiation. This electromagnetic output would slowly heat up the companion star, causing it to swell. Ultimately the outer layer of the star would waft off into space. With less matter, and thus less gravity, the star would swell further and further, losing layer after layer. when the star was pared away to less than a hundredth of a solar mass, it would begin to cool. When it did, it would shrink down again and harden into a planetlike body.
Alternatively, says Krolik, rather than cooling to form a planet, the companion could be pulverized into gas by the gravity of the neutron star. The smaller, denser body could absorb some of the gas from the companion, helping to increase the speed of its spin to the millisecond pulsar rate. The rest could form a ring that would eventually congeal into one or more planets. The point at which this might happen is a matter of debate.
There are a number of quantitative knobs in this that no one knows how to set, Krolik says. At what stage does the companion become unstable? If a disk does form, how many planets will form, and at what distances? No one knows how to set those knobs.
Another model offers an even more exotic scenario for the formation of pulsar planets. Philipp Podsiadlowski, Martin Rees, and Jim Pringle at the University of Cambridge in England suggest that a pulsar could pick up companions by colliding with a star that was surrounded by planets of its own. The cosmic crack-up would destroy the original star, causing the pulsar to take its place at the center of the system. The original star would be reduced to a huge sphere of hot gas enveloping the inner planets. The high temperature would melt most of the planets, but some larger or more distant ones could survive. After about 100,000 years, the sphere of gas would be blown off by the stellar wind or fall back onto the pulsar, leaving the surviving planets behind.
Podsiadlowski admits that such collisions would be rare, but he says that a few hundred have probably occurred in our galaxy. If you do have a collision with a solar system, what we describe should happen, he says.
At this stage it’s too early to say which of these theories will survive the test of time. But whatever happens, Andrew Lyne may very well have a hand in future discoveries. We’ve got a very large data base on pulsar timing that we’re going to study very closely, he says. We’ve probably got fifteen hundred pulsar years of data--several times more than any other observatory has. We’re going to be studying that with great care over the next few months.
Lyne is also planning an all-sky survey of pulsars--and their possible planets--throughout the heavens. For this study, he will be using radio telescopes in England and Australia so he can cover both the northern and southern hemispheres. We don’t know where these programs are going to lead, Lyne says. If we did it wouldn’t really be research.