The Hubble Space Telescope " core sample" of Andromeda's halo stars completely avoided the galaxy's disk. All of the other images in this article come from the target area (box), which lies 51 arcminutes — about 11 kiloparsecs or nearly 36,000 light-years — from M31's nucleus. Bill Schoening, Vanessa Harvey REU program, NOAO, AURA, NSF Clicking on the images will take you to larger versions on Hubblesite.org.

Visible even to the unaided eye as a foggy spindle of light, the Great Galaxy in Andromeda, or M31, is the single most studied object outside our own galaxy. Just 2.5 million light-years away, it is the nearest giant spiral, one roughly similar in size, mass, and type to our own Milky Way galaxy. A puzzling difference, however, has been the surprising abundance of elements heavier than helium among the stars of Andromeda's halo — a spherical region that surrounds the galaxy's disk — relative to comparable Milky Way stars. A team led by Tom Brown of the Space Telescope Science Institute in Baltimore has solved the puzzle by taking the deepest-ever look at M31.

Astronomers knew that the Milky Way halo was deficient in those elements, a feature characteristic of stars that formed early in the galaxy's history. " We knew that the Milky Way halo was metal-poor and old," Brown said, " and we knew that the Andromeda halo was metal rich, but we didn't know how old it was." Telescopes could only resolve bright, giant stars in M31, stars that tell astronomers very little about the history of star formation there.

That changed last year, when astronauts installed the Advanced Camera for Surveys (ACS) on the Hubble Space Telescope. The new instrument gives Hubble the largest field of view it has ever had — 202 by 202 arcseconds, or about twice that of the Wide Field Planetary Camera 2 (WFPC2) — and increases its sensitivity, allowing astronomers to reach objects at least a full magnitude fainter than those in the famous Hubble Deep Field. " Now we can see the faint dwarfs, stars as faint as our own sun and fainter" in Andromeda, Brown said.

Tom Brown's Andromeda deep field is an 8,000-by-8,000-pixel mosaic of four images obtained with the Hubble Space Telescope's Advanced Camera for Surveys. It reveals some 300,000 halo stars and possibly as many as 10,000 background galaxies never before seen. NASA, ESA, and T. M. Brown (STScI)

By analyzing the distribution of color and brightness in a stellar population, astronomers can tell just how old that population is. Sorting the stars by color and brightness creates the familiar color-magnitude diagram: blue stars on the left-hand side, red stars on the right, stacked bottom to top in order of increasing brightness. At some point in the history of a related star group, most stars will lie on a diagonal band that runs across the middle of the plot. That band, called the main sequence, represents stable stars that burn hydrogen in their cores as our sun does today.

The brightest, bluest stars soon exhaust their core hydrogen and move toward the right-hand side of the plot, where red giants reside. Over time, progressively fainter stars of more moderate brightness, temperature and mass follow suit. The main sequence unzips as stars peel off to become evolved giants. The color-magnitude plot shows a kink on the main sequence just where stable stars have begun to leave it. The color and brightness of those turnoff stars reflect the amount of time elapsed since the whole group formed.

Brown and colleagues wanted an image deep enough to examine halo stars below the main sequence turnoff. But deciding where to point the Hubble Space Telescope required them to find a " sweet spot" between two competing factors. " We wanted it crowded enough to minimize contamination by both foreground stars in our own galaxy and background galaxies behind Andromeda, and we wanted to have good population statistics so we could really disentangle a complicated star formation history," Brown said. Competing with that, the field had to be sparse enough so that they could get accurate measures of the brightness of individual stars. The group also wanted the " cosmic core sample" to include one of Andromeda's hundreds of globular clusters, compact spheres made from hundreds of thousands of aged stars. They also decided to calibrate their data by observing several well studied Milky Way globular clusters — including 47 Tucana and M92 — with the same filters used in the deep image.

Boxes show the location of detail images within the ACS mosaic. NASA, ESA, and T. M. Brown (STScI)

The field they selected lies 51 arcminutes, or 36,000 light years, from M31's nucleus and contains the globular cluster GC 312. Between December 2, 2002, and January 11, 2003, Hubble peered into the target zone for 121 orbits and made 250 separate exposures. The resulting image has a total exposure time of 3.5 days and penetrates to magnitude +31. An estimated 300,000 never-before-seen halo stars and thousands upon thousands of new background galaxies pepper the frame.

With the star data in hand, said Brown, " We did the classical age diagnostics that have been applied to stars in our own galaxy for decades." While most of the stars fall into the age range found among stars in the Milky Way's halo — 11 to 13 billion years — about 30 percent of Andromeda's halo stars turn out to be much younger, between 6 and 8 billion years old. Since younger stars develop in an environment more chemically diverse than that in which the halo stars formed, such a significant population of younger stars explains the unusually high fraction of heavy elements in M31's halo. Brown presented his findings at the Space Telescope Science Institute's annual symposium in May 2003.

Brown sees three possible explanations for the presence of the young stars. Perhaps repeated mergers of small galaxies have been pummeling M31's disk and blowing newly formed stars out into the halo. Or perhaps a single larger galaxy collided with Andromeda and some of the interloper's young stars became mixed with the halo population. Alternatively, a galactic collision somehow may have fueled a new wave of halo star formation, producing the young stars right where we see them. Which of those scenarios works best, he said, is a problem for the theorists. The findings lend support to the view that giant galaxies like Andromeda and our Milky Way have grown by cannibalizing many smaller galaxies throughout their histories. And it appears that M31 has taken part in a major interaction much more recently than has the Milky Way.

A tight, orderly spiral (1) invites comparison to the far more distant galactic shreds that make up much of the background in this part of the Andromeda deep field. The reddish arc (2) at first glance suggests a galaxy that has been gravitationally lensed by an unseen object. On closer inspection, it appears to be a chance superposition of two edge-on disk galaxies. A ruddy disk galaxy (3) and a few bright stars in the Andromeda Galaxy's halo dominate this part of the field. More distant galaxies appear as irregular patches and nebulous clumps. The face-on spiral (4) appears to have a warped arm, most likely evidence that a galactic intruder long ago disrupted the galaxy's graceful spiral form. The " blue blob" (5) may be a galaxy distorted by recent collisions, or perhaps two galaxies in the process of merging. GC 312 (6), a compact globular cluster containing some 100,000 stars, is one of several hundred similar " star balls" that orbit M31. Even in this crowded star field several background galaxies emerge. ESA, and T. M. Brown (STScI)

Between talks at the symposium, I pointed out to Tom Brown that he seemed almost embarrassed by all the attention given to those background galaxies. His rendition of the halo image naturally emphasizes the stars, but the version released to the public — and which you see here — was enhanced differently in order to show off the galaxies. " I was after the stars," he smiled, " and in any case it's unclear how useful the galaxy images will be simply because all those stars are in the way."

He mentioned that he was recently asked how an eight-meter space telescope would perform for this type of work. " It could do in an hour what took Hubble three-and-a-half days," he told me. As we talked, he stretched his arms over his head, framed an imaginary field of view with his fingers, and mimed taking a snapshot. " With an instrument like that, you could pin down star formation history throughout the Local Group," he said, moving his arms as if to frame up another section of sky. " If such a telescope ever materializes, that's exactly what I plan to propose."

See also:

• " Star formation histories in the Local Group," T. M. Brown, New Astronomy Reviews, November 2005
• " Halos of Spiral Galaxies. III. Metallicity Distributions" , M. Mouhcine, R.M. Rich, H.C. Ferguson, et al., The Astrophysical Journal, November 2005
• " Age Constraints for an M31 Globular Cluster from Main Sequence Photometry" , Thomas M. Brown, Henry C. Ferguson, Ed Smith, et al., The Astrophysical Journal, October 2004