© 2005 Nature Publishing Group

NATURE|Vol 438|3 November 2005 NEWS & VIEWS

39

removing these foreground signals might leadto a spurious result.

Although several groups have tentativelyclaimed to have seen fluctuations from earlystarlight on the basis of previous infrared spacemissions3,4, the increased depth of the IRACimages means that Kashlinsky et al.1 couldreduce the confusing signals from foregroundgalaxies to fainter upper limits (Fig. 1). Theypresent several tests to demonstrate the relia-bility of their detected signal and the originthey claim for it. The pattern of structure theysee, for example, is statistically similar at fourdifferent infrared wavelengths, and is consis-tent across fields whose sky coordinates arevery different (so an origin in the Milky Way isunlikely). The signal does not change betweentwo measurements taken six months apart, aswould be the case if its origin were ‘zodiacal’light reflected by local interplanetary dust inthe Solar System.

The authors use various methods to removethe signals coming from faint galaxies, noneof which significantly alters their result.Remarkably, however, the total contributionof foreground galaxies, estimated by extrapo-lating to include those too faint to be resolvedin the deepest IRAC images5, is small com-pared with the residual signal, which theauthors ascribe to primordial stars. Kashlin-sky et al. argue that this is further proof oftheir signal’s origin, because only a very largeerror in the amount of foreground galaxyclustering would make the residual signal dis-appear. Because the clustering observed in thebackground will be diluted if the first starsswitch on gradually over a range in cosmictimes, the prominent signal might conceiv-ably arise because it was produced in a veryshort time interval. But the imbalance stillseems a surprising outcome, and a number ofuntested assumptions involved in allowing forunobserved galaxies could represent a weak-ness in the analysis.

Kashlinsky and colleagues, perhaps wisely,do not interpret their signal in much detail.Theorists have predicted the level of fluctua-tions that might be seen in such experiments6;but their calculations, too, depend on manyimponderables. The genuine excitement inthis work lies in the practicality of detectingthe stellar radiation from hitherto uncharteddistances corresponding to a time when theUniverse was barely 100 million years old. Notbad for an 85-centimetre telescope! ■

Richard S. Ellis is at the California Institute ofTechnology, MS 105-24, Pasadena, California91125, USA.e-mail: rse@astro.caltech.edu

1. Kashlinsky, A., Arendt, R. G., Mather, J. & Mosely, S. H.Nature 438, 45–50 (2005).

2. Abel, T. et al. Science 295, 93–98 (2002).3. Wright, E. L. & Reese, E. D. Astrophys. J. 545, 43–55

(2000).4. Matsumoto, T. et al. Astrophys. J. 626, 31–45 (2005).5. Fazio, G. et al. Astrophys. J. Suppl. Ser. 154, 39–43 (2004).6. Santos, M., Bromm, V. & Kamionkowski, M. Mon. Not. R.

Astron. Soc. 336, 1082–1092 (2002).

be a specific task for the 6.5-metre-apertureJames Webb Space Telescope — successor tothe 2.5-metre Hubble Space Telescope — thatis projected to enter orbit in 2013. Kashlinskyand colleagues’ observations1 come from amore modest-sized instrument, the InfraredArray Camera (IRAC) on board NASA’sSpitzer Space Telescope. The technique theyuse — that of analysing ripples in the back-ground sky — is well established in cosmol-ogy: observations of ripples at microwavefrequencies produced soon after the Big Bang(the so-called cosmic microwave background)gave us precise measurements of the con-stituents and geometry of the Universe. IRACdetects light at infrared wavelengths between3.6 and 8 micrometres, and is sensitive to thelight of young stars that has been redshifted(moved to longer wavelengths) by a factor ofbetween 20 and 50 by the expansion of the cos-mos since its emission. This light would haveoriginated at a time when the Universe wasonly 100 million years old — an infant interms of its stately 13.7 billion years today.

Spitzer, with its 85-centimetre aperture,does not have the angular resolution to resolveindividual sources at these enormous dis-tances. But it might well be capable of detect-ing distant stars’ combined output as structurein the light produced behind that of fore-ground sources. And here lies the observa-tional challenge: Spitzer’s IRAC detectors arealso sensitive to diffuse light that is producedwithin the Solar System, in clouds of cool gasin our Galaxy and, crucially, from galaxies —those close by, as well as those a long way back in cosmic time. Even a minor blunder in

COSMOLOGY

The infrared dawn of starlightRichard S. Ellis

The modest-sized but successful Spitzer Space Telescope has detectedfluctuations in cosmic light at infrared frequencies. Is this the signature ofthe first population of stars that formed in the Universe?

On page 45 of this issue, Kashlinsky et al.1 pre-sent observations that reveal clustering in thedistribution of cosmic infrared light over andabove that expected from the combined effectof known galaxies. This excess signal couldconceivably be light from stars that switchedon when the Universe was just a tiny fractionof its present age. The authors do not detectany individual sources, nor can they pinpointprecisely when in cosmic history these signalswere produced. Nevertheless, their result islikely to provoke much discussion among cosmologists.

Searching for cosmic sources that ignitedwhen the Universe was very young is at thefrontier of our current observational capa-bilities. Deep surveys with space-based telescopes and the larger ground-based tele-scopes allow us to observe very faint sourcesthat are thought to have originated in earlycosmic times. Much progress has been madein tracing how early stellar systems changedand grew to become ‘respectable’ galaxiessuch as our own Milky Way. But the results ofnumerical simulations2 suggest that the veryfirst stars may have been quite different fromthose that came later. They probably con-tained only atoms of light nuclei produced inthe Big Bang — so no heavy elements such ascarbon and oxygen. They are also likely tohave been very massive, over a hundred timesmore massive than the Sun. When theseheavyweights switched on, they would haveburnt up their hydrogen in only a few millionyears, shining intensely and briefly in a bril-liant dawn of starlight.

Detecting and studying these first stars will

Figure 1 | Sign of the times? A deep infrared exposure taken with the Spitzer Space Telescope. Whencontributions from the known stars and galaxies are subtracted (black areas), the fluctuations in thebackground emission show a residual signal over and above that expected from the past history ofgalaxies. This may arise from the first massive stars that ignited when the Universe was only 100million years old. (Courtesy of A. Kashlinsky et al.1.)

3.11 News & Views 031 CDS 28/10/05 4:51 PM Page 39

Nature Publishing Group© 2005