News & Views | Published:


A problem of distance

How far is the Pleiades star cluster from Earth? The latest measurement suggests that there is a problem with data from the Hipparcos satellite, which will have repercussions for estimating other astronomical distances.

The European Space Agency's satellite Hipparcos1 has enjoyed a successful astrometric mission, measuring the distances and motions of more than 100,000 nearby stars. But its measurement of the distance to the well-known star cluster the Pleiades (or Seven Sisters; Fig. 1) has caused a major controversy2. The cluster is more than 100 parsecs from Earth, equivalent to a distance of 300 light years. But exactly how much more than 100 parsecs is a matter of dispute. Xiaopei Pan and colleagues3 (page 326 of this issue) offer a plausible resolution.

Figure 1: How far?


The Pleiades star cluster, also known as the Seven Sisters, is more than 300 light years from Earth. Exactly how far away the cluster is, however, is controversial. A measurement by the Hipparcos satellite conflicts with an estimate based on stellar brightness. A new calculation by Pan et al.3 suggests that the Hipparcos data may be flawed.

The oldest direct method of measuring the distance to a star is through 'parallax', the small shift in a star's position in the sky caused by the Earth's orbital motion. From this comes the name of a unit of distance commonly used by astronomers — the parsec (pc), which corresponds to roughly 3×1013 kilometres. At a distance of 100 pc, the parallax is only 0.01 arcseconds (another common unit: an arcsecond is 1/3,600 of a degree). But the Earth's atmosphere blurs stellar images, such that, from the planetary surface, it is difficult to measure parallax with an accuracy better than 0.01 arcseconds. So although the parallax technique has been used4, not everybody accepts the measurements as an accurate distance for the Pleiades.

Traditionally, the distance to the Pleiades cluster has instead been estimated using a method called main-sequence fitting. This takes advantage of the empirical relation between the colour of a star and its brightness, which is theoretically well understood for stars in the main sequence (stars powered by the fusion of hydrogen into helium are 'main sequence', and remain so for roughly 90% of their observable lives). First, the distance to the Hyades cluster, the nearest star cluster to Earth, was measured directly through simple geometrical techniques. Then the main-sequence stars in Pleiades were compared with those in Hyades. Because the apparent brightness of a star is inversely proportional to the square of its distance, the distance to Pleiades was worked out5 at 132±4 pc.

Enter Hipparcos. In orbit above the Earth's atmosphere, this satellite has provided measurements of parallax that are accurate to 0.001 arcseconds, a major breakthrough. But for stars at a distance of 132 pc, even Hipparcos could not measure parallaxes to an accuracy better than 13%. However, by combining the measurements made of many stars in the Pleiades cluster, that accuracy could be improved, and its distance was announced6 to be 118±4 pc — a serious conflict with the distance obtained from main-sequence fitting.

The consequences of this discrepancy are profound. If Hipparcos is right and the main-sequence-fitting result is wrong, then not only are the distances that have been calculated to stellar globular clusters wrong, but we are also wrong about the estimates of cluster ages and a seemingly robust way to measure the size of the Universe. On the other hand, the Hipparcos measurements may have suffered some unknown systematic error, which should be uncovered lest a similar error affect the future, more accurate astrometric mission called Gaia7.

Many attempts have been made to resolve the dispute, but none has succeeded. This latest attempt, by Pan et al.3 using the Palomar Testbed Interferometer in California, is a major step towards a firm distance determination, although I doubt that it will convince hard-core sceptics. The principle of the method is simple and elegant. One of the brightest stars in the Pleiades cluster, named Atlas, is a binary system. The two stars move around each other under gravity, with an orbital period of 290.8 days. Pan et al. measured the semi-major axis of this orbit to be 0.0129±0.0001 arcseconds, a measurement that is more robust than parallax.

Ideally, the next step (which is missing so far) would be to measure the radial velocities of the two stars. If the changes in radial velocity are combined with the orbital period and the angular size of the orbit, a direct measure of the distance to the system is obtained. Unfortunately, the two stars rotate so rapidly that it is difficult to measure their radial velocities (their absorption lines overlap). But it is not impossible. In fact, another group is working on such measurements, and once that work is complete I expect the controversy to be resolved — to the satisfaction of some but the discomfort of others.

Pan et al.3 do not have those radial velocities, so they have used a less direct, but still simple, method. In a nutshell, it is based on two relations — Kepler's third law and the mass–luminosity relation — that connect the unknown distance of the Pleiades cluster with the mass of the Atlas binary. The authors conclude that the distance is 135±2 pc. So if Pan et al. are right, Hipparcos is wrong.

I suspect that Hipparcos enthusiasts will not accept this measurement: using the mass–luminosity relation is equivalent to using main-sequence fitting, which they claim has problems. But I think Pan et al. have got it right. There could well be a systematic error in the Hipparcos data, owing to its highly eccentric orbit. This orbit was not planned but was caused by a failure of some of the satellite's boosters shortly after launch. Initially it seemed that the mission was lost, but thanks to the ingenuity of the Hipparcos team, the satellite was still able to complete its task. I would not be surprised, though, if there turns out to be some residual problem with the defective orbit that has caused an unfortunate shift in the Hipparcos data8. If we are to have confidence in our measurements of astronomical distances, not just the distance to the Pleiades cluster, this problem must be solved.


  1. 1

    Perryman, M. A. C. et al. Astron. Astrophys. 323, L49–L52 (1997).

  2. 2

    Mermilliod, J.-C., Turon, C., Robichon, N., Arenou, F. & Lebreton, Y. in ESA SP-402: Hipparcos Venice '97 (eds Perryman, M. A. C. & Bernacca, P. L.) 643–650 (European Space Agency, Noordwijk, 1997).

  3. 3

    Pan, X., Shao, M. & Kulkarni, S. R. Nature 427, 326–328 (2004).

  4. 4

    Gatewood, G., Kiewiet de Jonge, J. & Han, I. Astrophys. J. 533, 938–943 (2000).

  5. 5

    Meynet, G., Mermilliod, J.-C. & Maeder, A. Astron. Astrophys. Suppl. Ser. 98, 477–504 (1993).

  6. 6

    van Leeuwen, F. Astron. Astrophys. 341, L71–L74 (1999).

  7. 7

    Perryman, M. A. C. in ESA SP-485: Proc. 1st Eddington Workshop on Stellar Structure and Habitable Planet Finding (eds Favata, F., Roxburgh, I. W. & Galadi, D.) 229–234 (European Space Agency, Noordwijk, 2002).

  8. 8

    Narayanan, V. K. & Gould, A. Astrophys. J. 523, 328–339 (1999).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Further reading

  • Astrophysics in 2004

    • Virginia Trimble
    •  & Markus Aschwanden

    Publications of the Astronomical Society of the Pacific (2005)


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.