Contact: Janice Fink
UA News Bureau, 205/348-3782

Sources: William Keel (205/348-1641) and Raymond White (205/348-1640). Images accompanying this release are available on the World-Wide Web at


For release: 9:20 a.m. EST, Friday, 9 January, 1998

(Washington, D.C.) - New images presented today from a pair of satellite telescopes offer a unique view of the obscuring dust in galaxies, taking advantage of rare cosmic silhouettes to help resolve a debate about how much of this dust exists in galaxies, where it is, and how much it matters.

Astronomers William Keel and Raymond White III, of The University of Alabama in Tuscaloosa, presented the findings at the Washington meeting of the American Astronomical Society, using data from NASA's Hubble Space Telescope (HST) and the European Space Agency's Infrared Space Observatory (ISO) to provide the clearest mapping yet of the dust in distant galaxies.

The amount of dust in typical galaxies, and its effect on what we see, has been a topic of controversy since a 1990 paper by Edwin Valentijn challenged the long-held view that galaxies are basically transparent. Valentijn used statistical evidence from galaxy catalogs to suggest that most galaxies are virtually opaque. Subsequent statistical analyses have shown that such conclusions can be biased by the ways in which galaxies were selected for the study, or selected for entry in the existing catalogs in the first place. To avoid such ambiguities, Keel and White have taken a more direct approach by examining the rare cases where a foreground galaxy is partially covering a more distant background galaxy. Using ground-based telescopes, they culled through thousands of galaxy pairs to find those few that are symmetric enough and viewed in just the right way to show off their dust against the spotlight of a background companion. New data from two orbiting observatories provide views of a chosen few of these objects in unprecedented detail.

The investigators presented images, taken with Hubble's Wide Field Planetary Camera 2 (WFPC2), of two striking pairs of galaxies, each with a spiral galaxy in front of a smooth elliptical companion. Both are too faint to appear in the familiar NGC catalog, and were first listed in the Arp-Madore (AM) catalog of the southern sky. AM1316-241 is located about 400 million light years away (redshift z=0.033) in Hydra, while AM0500-620 is in the deep southern constellation Dorado, about 350 million light years away (redshift z=0.028). The images resolve structures in these galaxies as small as 175-200 light-years in size, improving more than ten times the discrimination of detail on the best earlier images.

"A glance at these spectacular images makes the same points we had spent years reaching by detailed analysis and modelling from ground-based data", says Keel. "They are as interesting for what we didn't expect to see as for what we did."

In fact, the astronomers did see what they expected to find, that the dust is patchy and clumped, largely aligned along the spiral arms. "Having most of the absorbing dust in the spiral arms, where most of the light originates, is what caused the statistical studies to wrongly conclude that spirals are opaque," says White. This fact was implicit in comparison of their earlier results through various filters in visible and near-infrared light, but from the ground only the very largest of these dust patches, thousands of light-years in extent, could be seen clearly. While not exactly surprising - this kind of distribution is obvious from most detailed pictures of spiral galaxies - this technique using overlapping galaxies avoids many ambiguities that crop up in trying to measure the dust content of an individual galaxy.

The new data also reveal surprises. The dustiest patches that appear in the HST images aren't very dark, since at least 20% of the blue light comes through, and even more of the near-infrared light. This seems to violate expectations based on looking sideways through the arms of our own galaxy. In addition, the dusty spiral arms don't show as much fine structure as the researchers might have expected. The dust patches smooth out at sizes of 500 light-years or so, rather than the full range of sizes, going down as small as can be measured, suggested by fractal models of interstellar material. The two galaxies observed by HST also differ in the width and texture of the dust within spiral arms, despite being spiral galaxies of similar overall type.

The ISO infrared observations complement the exquisite Hubble imaging, in their ability to peer into the dusty regions for traces of star formation and in measuring the total amount of dust, based on its own emission of far-infrared radiation. They show that there is not significant star formation in the dusty regions mapped with HST, giving further limits on how much of the action in these galaxies takes place where it's difficult to see.

Knowing the amount of dust found in galaxies is important in a wide range of questions in astrophysics. These results suggest, for example, that absorption by dust in intervening galaxies is unlikely to account for the fact that we don't see many quasars with very high redshift; more likely, we are looking beyond the time in which they formed. Closer to home, attempts to balance the books on energy flow in galaxies, both what is emitted by stars and that which is absorbed by dust and subsequently radiated deep in the infrared, have been stymied by our lack of knowledge of just where the dust lies and how it is situated with respect to the brightest and hottest stars.

Keel and White are quick to point out areas for further work in this direction. These galaxy pairs only tell us about their outer regions, where the backlighting is strongest; the inner parts of galaxies, richest in heavy chemical elements and perhaps in dust, are much more difficult to explore. HST observations scheduled for the next year, both of a direct galaxy superposition extending their own program and by several other research groups using the new infrared instrument NICMOS on board HST, are likely to improve this situation. As analysis proceeds of the ISO data in the far-infrared, where all we see is the dust component of galaxies, overall understanding of the most typical dust grains is improving as well. It is this kind of measurement that can tell us the total amount of dust, which is important in assessing how it is massed together on scales beyond the ability of even the Hubble telescope to discern directly.

This research was funded by NASA, in an HST research grant through the Space Telescope Science Institute and through support of U.S. participation in the ISO mission.


Astronomers have learned, usually the hard way, that "what you see isn't always what you get" on a cosmic scale. In recent years, several astronomers came to this realization in the context of something that many had long thought a settled issue - can we see through galaxies? It was obvious from the earliest photographs of spiral galaxies that there is absorbing material in front of some of the starlight. In fact, James Keeler at Lick Observatory produced a detailed comparison of the appearance of dust structures in spiral galaxies years before Edwin Hubble established their nature as independent galaxies like our own Milky Way. Even so, most astronomers were convinced by Erik Holmberg's results more than 40 years ago that the overall loss of light from dust is small, and therefore dust is not an important factor in our measurements of ordinary galaxies. Holmberg compared galaxies' average surface brightness with their inclinations in reference to our line of sight, reasoning that if dust is not important, edge-on galaxies will have a higher surface brightness with the same amount of light packed into a slimmer area.

This comforting conclusion came into question around 1990, with a combination of theoretical studies by astronomers at the University of Wales in Cardiff and a statistical analysis by Edwin Valentijn of the European Southern Observatory, which together showed that available data were equally consistent with the notion that spiral galaxies were very dusty, and that we might be seeing less than half of their starlight as it is absorbed by dust grains. These reports touched off a flurry of further research, since there is a heavy astronomical investment in knowing how many stars there are in galaxies versus how much starlight we actually see. Whether spiral galaxies are largely transparent or opaque has ramifications for the nature of dark matter, the star formation rates in galaxies, and the observability of quasars, among the most distant objects in the Universe.

Astronomers estimate the visible masses of spiral galaxies by adding up the light from their stars and by knowing the typical masses of such stars from measurements in our own Milky Way galaxy. But spiral galaxies contain much more mass than is observed in visible stars and gas. This additional mass, detected only through its gravitational effects on galaxies, is attributed to so-called "dark matter," the nature of which is one of the most important unsolved problems in astronomy. If spiral galaxies are opaque to their own visible light, then astronomers may be underestimating the amount of normal stellar matter they contain and attributing too much mass to the "dark matter."

Infrared telescopes, such as the Infrared Astronomical Satellite, have found many galaxies which are strong emitters of far-infrared radiation, thought to be a sign of intense bursts of star formation. Since most stars are formed deep in dusty clouds of interstellar gas, their starlight is not directly visible in the optical, but light from the young stars heats the surrounding dust, which in turn radiates at infrared wavelengths that can escape the dusty clouds. However, if dust is very widespread and thick throughout galaxies, the stars which heat the surrounding dust may not be just the young stars, so the observed infrared radiation would not be the good measure of star formation rates that it is usually assumed to be. Clarifying the dust content of typical galaxies is important to understanding how they evolve over time, as interstellar gas is transformed into stars.

The transparency or opacity of spiral galaxies may also limit how deeply we can see into space. The most distant known objects are quasars, very energetic galactic nuclei, which are seen in greater and greater numbers the further out astronomers probe. But beyond a certain distance, about 85 per cent of the way to the edge of the observable universe, quasars are no longer detected. Some astronomers suggest that this cutoff is due to looking through intervening galaxies on the way to distant quasars, with these galaxies shadowing their light. However, if spiral galaxies are largely transparent, they cannot cause the quasar cutoff.

The grains in interstellar space, produced in the outer atmospheres of stars near the latest stages of their lifetimes and in stellar explosions, are tiny, thousandths of a millimeter or less in size. Like the dust in Earth's atmosphere, they allow more red than blue light to pass through, reddening background objects. In sufficient quantity, these grains can block nearly all the visible light from distant objects.

- 30 -

University of Alabama press releases are available on the World Wide Web at