In the past four years, astronomers have discovered at least 30 planets in orbit around nearby stars1,2,3,4,5,6,7,8,9,10,11. Their detection relies on the fact that the force of gravity that keeps a planet in its orbit has an opposite reaction on the star, causing it to wobble in its own tiny orbit about the common centre of mass. From the size of the stellar wobble, astronomers can calculate a lower limit for the planet's mass, but they can't tell whether they are seeing the effect of a lighter planet in an orbit seen edge-on, or a heavier one in the same orbit viewed differently — until now. On page 751 of this issue, Cameron et al.12 argue that they have detected light reflected from a planet orbiting a nearby star, providing a direct measure of its mass and size.
All the stellar wobbles indicating extrasolar planets have so far been found using the Doppler technique, which measures changes in the star's velocity along the line of sight. What is actually measured is a small, systematic change in wavelength (Doppler shift) of the many absorption lines that make up the star's spectrum. Jupiter, for example, causes the Sun's velocity to vary by ∼12.5 m s−1 during its 12-year orbit, whereas a planet much closer to its parent star produces a larger velocity perturbation (in the case of the planet orbiting 51 Pegasi, ± 60 m s−1). The Doppler method favours systems quite unlike our own, with heavy giant planets close to their stars producing the strongest and most easily detected gravitational tugs. Some have orbits as short as a few days, rather than decades, and many extrasolar planets are heavier than Jupiter. Although these discoveries have serious implications for theories of planet formation, the indirect detections cannot tell us anything about the radius or composition of these planets.
The planet detected by Cameron et al.12 is orbiting every 3.3 days around the star τ Boötis, which is 50 light years away near the constellation Boötes. The planet's tight orbit causes a relatively large stellar Doppler shift of ± 470 m s−1, implying a mass 3.9 times or more that of Jupiter. The planet is so close to τ Boötis (20 times closer than the Earth is to the Sun) that even the Hubble Space Telescope can't distinguish them. The combination of close orbit, high mass and the fact that τ Boötis is hotter and bigger than our Sun, means this planet is one of the hottest known.
The planet must be brilliantly illuminated, and is a prime candidate for direct optical detection. Although it will reflect to us only about 0.01% of the total starlight from τ Boötis, this light will be strongly Doppler shifted because of the planet's high speed. Cameron et al. find evidence for exactly such an effect, a faint planetary spectral component that is strongly blueshifted by 74,000 m s−1 when the star is redshifted by 470 m s−1 (redshift indicates an object moving away from us, whereas blueshift indicates an object approaching Earth). This result directly reveals that the planet weighs 470/74,000 × (1.25 solar masses) 4= 8 Jupiter masses (MJ). This mass is consistent with the lower mass limit from the amplitude of the stellar wobble, provided that the plane of orbital motion is being viewed from above, with the axis of rotation tilted 29° in our direction.
The strength of the planet's signal tells us about its size and brightness. Assuming an albedo (the amount of light reflected) of 55%, similar to that of Jupiter, Cameron et al. deduce that the planet must be 1.6–1.8 times bigger than Jupiter — that is, 30–50% larger than expected from theory13. A smaller albedo would imply an even larger radius. In addition, the authors find the planet to be blue–green in colour, something mildly at odds with emerging theories of the reflectivity of extrasolar planets14,15,16,17. So, if true, the Cameron et al. findings will demand new explanations.
Unfortunately, the detection of reflected light is extremely difficult and so reliability is not high. Most worryingly, when the data are analysed at orbital periods other than the known period of the planet, the authors find spurious signals similar to (or larger than) that taken to be the planetary signal. Indeed, there is a 1 in 20 chance that the apparent planetary signal is just random noise. It is also true that previous independent observations18 did not find a planetary signal from the same star and in the same region of the spectrum where Cameron et al. find their strongest result. We expect that more extensive observations made next spring, when τ Boötis is once again high in the night sky, will resolve the issue.
In the meantime, unimpeachable direct evidence of an extrasolar planet has been found in the light from star HD209458 because of the transit of a planet across its stellar disk19,20. At the moment when the star is furthest from us, the starlight is seen to dim by about 2% for three hours, as the planet passes in front. To see a transit we must be viewing this system nearly edge-on, so the planet's mass must be the minimum given by the star's wobble, 0.63 MJ. From the amount of dimming, and the known size of the star, the planet is found to be 1.27 times the radius of Jupiter (RJ), independent of any assumptions about albedo. This is in harmony with theoretical predictions13 of between 1.2 and 1.6 RJ for a fast-orbiting Jupiter-like planet. Any last doubts about the existence of Jupiter-mass planets are removed by this direct measure of mass and size. Not only is the planet real, we now know that it is a gas giant made mainly of hydrogen, not rock — a solid planet with a mass of 0.63 MJ would have to be three times smaller than Jupiter.
Our understanding of planetary science will be enormously increased by direct measurements of the physical properties of extrasolar planets. The detection of reflected light from a planet, if confirmed, is the first step towards spectroscopic studies of extrasolar planets that would enable us to study their chemical composition and atmospheres. A journey that begins with giant planets close to their primary stars should in time lead to a better understanding of Earth's place among the many planetary systems that seem to surround us.
Butler, R. P. & Marcy, G. W. Astrophys. J. 464, L153–L156 (1996).
Butler, R. P., Marcy, G. W., Williams, E., Hauser, H. & Shirts, P. Astrophys. J. 474, L115–L118 (1997).
Butler, R. P., Marcy, G. W., Vogt, S. S. & Apps, K. Publ. Astron. Soc. Pacif. 110, 1389–1393 (1998).
Cochran, W. D., Hatzes, A. P., Butler, R. P. & Marcy, G. W. Astrophys. J. 483, 457–463 (1997).
Fischer, D., Marcy, G. W., Butler, P., Vogt, S. S. & Apps, K. Publ. Astron. Soc. Pacif. 111, 50–56 (1999).
Latham, D. W., Stefanik, R. P., Mazeh, T., Mayor, M. & Burki, G. Nature 339, 38–40 (1989).
Marcy, G. W. & Butler, R. P. Astrophys. J. 464, L147–L151 (1996).
Marcy, G. W., Butler, R. P., Vogt, S. S., Fischer, D. A. & Liu, M. C. Astrophys. J. 520, 239–247 (1999).
Marcy, G. W., Butler, R. P. & Fischer, D. A. Bull. Am. Astron. Soc. 194, 14.02 (1999).
Noyes, R. W. et al. Astrophys. J. 487, L111–L114 (1997).
Mayor, M. & Queloz, D. Nature 378, 355–359 (1995).
Cameron, A. C., Horne, K., Penny, A. & James, D. Nature 402, 751–755 (1999).
Guillot, T., Burrows, A., Hubbard, W. B., Lunine, J. I. & Saumon, D. Astrophys. J. 459, L35–L38 (1996).
Goukenleuque, C., Bezard, B., Joguet, B., Lellouch, E. & Freedman, R. Astron. Astrophys. (submitted).
Marley, M. S., Gelino, C., Stephens, D., Lunine, J. I. & Freedman, R. Astrophys. J. 513, 879–893 (1999).
Seager, S. & Sasselov, D. D. Astrophys. J. 502, L157– (1998).
Sudarsky, D., Burrows, A. & Pinto, P. http://xxx.lanl.gov/abs/astro-ph/9910504
Charbonneau, D. et al. Astrophys. J. 522, L145–L148 (1999).
Charbonneau, D., Brown, T. M., Latham, D. W. & Mayor, M. Astrophys. J. Lett. (in the press).
Henry, G. W., Marcy, G. W., Butler, R. P. & Vogt, S. S. Astrophys. J. Lett. (in the press).