A fresh approach to stellar benchmarking

An avalanche of data is about to revolutionize astronomy, but the options for validating those data have been limited. High-precision measurements from the Hubble Space Telescope enable a much-needed alternative option.
Rachael Beaton is in the Department of Astrophysical Sciences, Princeton University, Princeton, New Jersey 08544, USA.

Search for this author in:

Try this experiment: extend your thumb at arm’s length and close one eye at a time. Your thumb will seem to ‘jump’ between two positions as you switch the eye that is closed. That jump is known as parallax. If you measure the jump as well as the distance between your eyes, you can use trigonometry to calculate the distance to your thumb. Astronomers use parallax, on a much greater scale, to measure distances to astronomical objects. Writing in The Astrophysical Journal Letters, Brown et al.1 report that they have achieved this for the nearby star cluster NGC 6397, using the Hubble Space Telescope. Their method will provide a crucial means of validating the wealth of parallax data released this year from the European Space Agency’s Gaia mission2.

It is a challenge to find a topic in astronomy that does not rely on the astronomical distance scale — a collection of methods applied in series to determine distances that are too large to be measured directly. Distances are used as conversion factors for deriving the physical quantities of celestial objects from observations and, therefore, they are essential for constructing models of the Universe. The foundations of the astronomical distance scale are trigonometric parallaxes for individual stars. These parallaxes enable us to calibrate the physical properties of those stars, which can then be used to infer properties of ever more distant stars, star clusters and galaxies. On the largest distance scales, they can even be used to calculate the size of the Universe.

A stellar parallax was first measured in 1838 by the astronomer Friedrich Bessel3. Many more have been recorded since. Until earlier this year, about 2 million reliable measurements4 had been made. This sounds like an impressive number, but effectively spanned only the astronomical cul-de-sac in which the Sun resides. That number increased to roughly 1 billion following the release in April of data from Gaia2, which surveyed a region well beyond the Solar System, almost halfway across the Galaxy. Until the publication of Brown and colleagues’ data, only one technique — very-long-baseline interferometry5 — was capable of measuring parallax directly on such distance scales. This was a concern because astronomers worldwide are poised to use the Gaia data in their research, and so it would be desirable to have more than one direct method for measuring stellar parallaxes to help validate the Gaia data.

The basic experimental set-up for measuring stellar parallax is identical to that described for observing your thumb. First, two images of the same astronomical object are taken with an interval of six months (Fig. 1a). This ensures that they are captured at positions separated by the diameter of Earth’s orbit around the Sun, in the same way that observing your thumb from each eye provides two viewpoints that are separated by a known distance. Second, the apparent displacement of the target star is determined. This involves measuring the position of the star in each image with high precision, and then measuring that position in relation to a set of reference objects (stars or galaxies) from the same image. Both tasks are conceptually simple yet tricky to achieve in practice. Brown et al. address them in interesting ways.

Diagram showing how to use parallax to measure astronomical distances

Figure 1 | Using parallax to measure astronomical distances. Objects viewed along two lines of sight have different apparent positions relative to their background, and the distance between those positions is known as the parallax. a, Distances to a nearby star are determined by measuring the parallax between two images of the star that were taken six months apart. Because the distance between the two observation positions is known to be the diameter of Earth’s orbit around the Sun, the distance to the star can be calculated using trigonometry. b, Brown et al.1 have measured the distance to the nearby star cluster NGC 6397. A camera on the Hubble Space Telescope took two long exposures six months apart, so that the cluster was visible as a ‘streak’ that results from the telescope’s orbital motion around Earth (the apparent drift of the cluster has been exaggerated, for clarity). Each position along the streak provides a different measurement of the position of each star in the cluster, thereby allowing the apparent positions of the stars to be measured more precisely than from ‘snapshot’ images. These measurements enabled the distance to the cluster to be determined.

To measure the position of their target star cluster, Brown and colleagues took the two images with a camera on the Hubble Space Telescope using a long exposure, so that the cluster’s stars ‘drift’ across the images as a result of the telescope’s orbital motion around Earth (Fig. 1b). This technique, known as spatial scanning6,7, produces images of the target as a ‘streak’. Every position along the streak provides a different measurement of the position of each star in the cluster.

The images of NGC 6397 taken by Brown et al. comprise more than 1,000 individual measurements, which increases the overall precision by more than 30-fold, compared with a conventional ‘snapshot’. Moreover, each measurement was made for numerous stars in that cluster. Spatial scanning has previously been used by researchers from the same group to study single stars, several thousand light years away, that are exceptionally bright6,7, but Brown et al. are the first to apply this technique to faint stars in a cluster at these sort of distances. (NGC 6397 is about 7,500 light years, or 2,390 parsecs, from Earth.)

The authors then used the same spatial scanning technique to measure the position of non-cluster stars in the background star field with incredible precision, enabling them to determine the displacement of cluster stars relative to each non-cluster star. But these relative parallaxes must be put into an absolute frame of reference, and setting such a frame is a complicated task. To do this, Brown and colleagues required coarse estimates (accurate to ±15% of the true value) of the parallaxes for the non-cluster stars. The authors obtained these by determining the type and size of each star, and then assigning each the mean physical properties of its class, from which its distance (and therefore its parallax) can be determined8.

The Gaia mission also uses a scanning technique to obtain the positions of target objects, but sets the absolute frame using a sample of quasars (point-like galaxies that are unfathomably far away) from across the entire night sky9,10. Brown and colleagues’ frame of reference has systematic uncertainties that are distinct from those of Gaia, and it could therefore provide a direct, independent means of testing the Gaia reference frame if it were to be expanded to include more star clusters.

The highly precise, long-distance parallax measurements provided by Gaia are a leap forward for astronomy. But, as in all fields of science, precision is not the only source of uncertainty. It is also crucial to understand the systematic uncertainty that is associated with a reference frame, partly so that this parameter can be included in data analyses, but also to devise a better means of establishing the frame. Systematic uncertainties can be reduced only by the addition of fresh, independent information, such as that provided by Brown and co-workers. The work involved in establishing these safeguards can be tedious and is often overlooked, but it is the bedrock of scientific progress.

Nature 558, 33-35 (2018)

doi: 10.1038/d41586-018-05306-7
Nature Briefing

Sign up for the daily Nature Briefing email newsletter

Stay up to date with what matters in science and why, handpicked from Nature and other publications worldwide.

Sign Up


  1. 1.

    Brown, T. M. et al. Astrophys. J. Lett. 856, L6 (2018).

  2. 2.

    Gaia Collaboration. Preprint at (2018).

  3. 3.

    Bessel, F. W. Mon. Not. R. Astron. Soc. 4, 152–161 (1938).

  4. 4.

    Lindegren, L. et al. Astron. Astrophys. 595, A4 (2016).

  5. 5.

    Reid, M. J. & Honma, M. Annu. Rev. Astron. Astrophys. 52, 339–372 (2014).

  6. 6.

    Riess, A. G., Casertano, S., Anderson, J., MacKenty, J. & Filippenko, A. V. Astrophys. J. 785, 161 (2014).

  7. 7.

    Casertano, S. et al. Astrophys. J. 825, 11 (2016).

  8. 8.

    Benedict, G. F et al. Publ. Astron. Soc. Pacif. 106, 327–336 (1994).

  9. 9.

    Gaia Collaboration. Preprint at (2018).

  10. 10.

    Gaia Collaboration. Astron. Astrophys. 595, A1 (2016).

Download references