Direct observations of the earliest luminous objects are only now beginning to illuminate the study of the origin and formation of galaxies. Each step beyond the previous limit of discovery holds the promise that new techniques will reveal the precursors of the galaxies we see today. On page 586 of this issue, Chen, Lanzetta and Pascarelle1 report the discovery of an object that probably has the largest redshift ever measured, z = 6.68. In the model of the expanding Universe, high redshifts — detected as shifts in spectral lines to longer wavelengths — of galaxies and quasars directly indicate their great distances from Earth. A redshift of 6.68 corresponds to a time when the Universe was only 5% of its present age. The previous record holder, reported last year2, was found at a distance or ‘look-back’ time corresponding to when the Universe was 1% older.
Three factors have made the discovery of such high-redshift galaxies particularly difficult. First, such objects are not merely dimmed with distance by the cosmological equivalent to the inverse square law; galaxies are also extended, and surface brightness is dimmed by (1 + z)4, so, only the brightest clumps in a system will be detectable above the sky and detector background. Second, there is no redshifted signal from the ultraviolet to the red part of the spectrum because the flux is completely absorbed by hydrogen in intergalactic clouds. Finally, the light produced and emitted in spectral lines by hydroxyl ions in the Earth's own atmosphere can swamp faint signals emitted in the far red by the most distant objects.
Chen et al.1 exploited the unique capability of the Space Telescope Imaging Spectrograph for discovering faint objects. The Hubble Space Telescope orbits above the atmosphere and focuses images very sharply, allowing slitless spectroscopy in the far-red part of the spectrum. Potential high-redshift galaxies are distinguished from galaxies of lower redshift through differences in their spectral energy distributions.
The unique spectral signature of these proto-galaxies at high redshift is generated by a region of newly formed hot stars, which excite the remnant of the gas cloud out of which they were formed into emission, producing features such as the Lyman-α emission line in the ultraviolet spectrum of atomic hydrogen (Fig. 1). The Lyman-α line is a particularly strong line in the emission spectra of active galactic nuclei and quasars, which, if they are at high redshifts, appears shifted into the visible part of the spectrum. The combined ultraviolet radiation — the continuous spectrum from the hot stars and the emission lines from the gas — then propagates through an absorbing medium of more diffuse gas surrounding the region of star formation. The strong hydrogen absorption from this galactic gas, and from numerous other clouds along the line of sight, creates the final spectral signature of an absorption step and trough.
Has such a spectral signature been unambiguously identified in this case, to claim the current redshift record? Chen et al. have detected one emission line at near-infrared wavelengths of somewhat marginal significance, and the absorption step and trough is what you would expect from absorption in the galaxy and the intergalactic medium. The spectrum is consistent with the emission line being Lyman-α, but the redshift measurement is not yet of a quality to be widely and uncritically accepted as decisive. Traditionally, evidence of two spectral features of known wavelength ratio or a spectral template match of relatively high fidelity is preferred. Some additional confirmation is required here for highest confidence, although the result seems very promising.
This candidate galaxy may be unusual, even among the few known very high-redshift objects, for two reasons. One is that the apparent ultraviolet luminosity is quite strong. On the basis of the known link between star formation rate and ultraviolet luminosity, the authors speculate that this result may indicate a trend to higher star-formation rates for individual galaxies, as we look back further in time. The other reason is that absorption from intergalactic hydrogen in the ultraviolet appears to be nearly 100% at this redshift of 6.68. At slightly lower redshifts, say 4.5, there is always some residual flux in the blue part of the emission spectrum. If the soup of intervening intergalactic clouds at z ∼ 6 is largely composed of neutral gas, then the total hydrogen absorption would be opaque, as observed. To have some residual flux at a slightly lower redshift suggests that the clouds are partially or nearly totally ionized over a very short interval of cosmic time. If total hydrogen absorption at z ∼ 6 is confirmed with a higher signal-to-noise ratio observation, it would signal a sudden change in the properties of the intergalactic medium, perhaps by the ‘switching on’ of quasars and galactic star formation, both of which produce strong ionizing ultraviolet radiation.
The record for the highest redshift object in the known Universe is transient these days, and the differences in cosmic ‘look-back’ time for large jumps in redshift are very small. The significance of this discovery lies more in the promise of a systematic study of these most distant objects, of which only a handful are known today. There are critical questions waiting for answers. Most importantly, what are these objects and what is their evolutionary fate? Their spectra can be interpreted as revealing active star formation at the rate of 4-25 solar masses per year3. Such values are lower limits4 because of the obscuring effect of intervening dust. Brighter objects have low heavy-element content, implying that the starlight comes from one of the earliest generations of star formation5. The typical mass is still unknown, although it is likely to be sub-galactic. The suspicion is that these objects appear luminous because of strong star formation.
We observe these objects as they were in the past, but whether they will ultimately become typical spiral galaxies like our Milky Way, or the cores of massive elliptical galaxies is a mystery. The answer will help us choose between competing theories of galaxy formation. High-redshift objects ( z > 3) may be regarded as the precursors to large, ‘normal’ galaxies because their approximate space density at ∼10% of the current age of the Universe is consistent with the density of luminous present-day galaxies5. Observations favouring such objects as the ancestors of elliptical cores are that many of the high-redshift galaxies are nearly spherical, and are unlikely to evolve into flattened rotating disks.
An argument for these ancient galaxies as precursors of spiral galaxies is that there are few close neighbours, which would be expected in the environment of a cluster where ellipticals are found6. Observed high-redshift objects are much more compact than present-day galaxies, so they might be sub-clumps that will ultimately merge. The objection to this picture is that you might expect to see small clusters of such objects, bound together and poised for coalescence. Typically, however, single objects have been detected. Perhaps such sub-clumps light up with newly formed stars at slightly different times, so that only one star-forming region per proto-galaxy is detectable at any very early epoch5,7.
The advent of near-infrared spectroscopy with adaptive-optics — which compensate for the effects of the Earth's atmosphere — at ground-based telescopes, and stronger near-infrared capability in space, should put the study of the earliest luminous objects, now at the limit of our grasp, onto a sound footing.
Chen, H.-W., Lanzetta, K. M. & Pascarelle, S. Nature 398, 586– 588 1999).
Hu, E. M., Cowie, L. L. & McMahon, R. G. Astrophys. J. Lett. 502, L99 –L103 1998).
Steidel, C. C., Giavalisco, M., Pettini, M., Dickinson, M. & Adelberger, K. L. Astrophys. J. Lett. 462, L17–L21 (1996).
Armus, L., Matthews, K., Neugebauer, G. & Soifer, B. T. Astrophys. J. Lett. 506, L89–L92 (1998).
Lowenthal, J. D. et al. Astrophys. J. 481, 673– 688 (1997).
Giavalisco, M., Steidel, C. C. & Macchetto, F. D. Astrophys. J. 470, 189– 194 (1996).
Trager, S. C., Faber, S. M., Dressler, A. & Oemler, A. J Astrophys. J. 485, 92–99 (1997).