HOW CAN the Universe be younger than the stars it contains? Astronomers have
been struggling with this bizarre question since at least the early 1990s, when
many researchers concluded that the Universe began somewhere between 8 and 11
billion years ago. But the oldest stars were thought to be 15 billion years old.
Something had to be wrong.
Since then, astronomers have wrestled with both parts of the
equation—the age of the Universe and the age of the stars. Then, last
year, the world media triumphantly broadcast the answer. A European satellite
had solved the “age problem” and everyone could rest easy. In a way, they were
right: the age problem does seem to have been solved. But the real answer has
come from a very different source.
You can understand why the media were excited. It looked like the sweet
victory of an underdog. The European Space Agency’s unglamorous Hipparcos
satellite, relegated to the drudgery of measuring distances to stars, had
usurped NASA’s sexy Hubble Space Telescope and solved cosmology’s greatest
paradox. Hipparcos made the Universe older, the stars younger and everybody
lived happily ever after.
Advertisement
A year later, however, those claims are under fire. “I was really looking
forward to hearing the results,” says Wendy Freedman of the Carnegie
Observatories in California. “But after looking at the data, my reaction was one
of disappointment.”
Freedman is not the only astronomer to criticise the Hipparcos claims. But
the good news is that astronomers pursuing other avenues of research have made
great progress towards resolving the age problem. For one thing, Hipparcos
apart, estimates of the Universe’s age have increased. For another, new models
of how stars develop suggest that the oldest stars are considerably younger than
previously thought. Put the two together and the Universe no longer appears
younger than its oldest stars.
Although the Hipparcos claims swirl in controversy, no one questions the
importance of the satellite’s work. It was launched in 1989 to measure the
distances to over 100 000 stars, and its work will be invaluable in many
branches of astronomy. The problem with measuring stellar distances is that you
can’t rely only on a star’s brightness. Sure, the farther away a given star is,
the fainter it will seem. But because some stars shine more brightly than
others, a feeble nearby star can look as bright as a more powerful distant star,
just as a candle on your dinner table can look as bright as a streetlight.
So to calculate distances, more information is needed. One approach, taken by
the designers of the Hipparcos satellite, is to use parallax—the apparent
shift in a star’s position when it is viewed from different vantage points. The
smaller the shift, the more distant the star. But parallax reaches only so far,
because the maximum distance between different vantage points is limited to
opposite points on the Earth’s orbit around the Sun. Stars farther than a few
hundred light years away have parallaxes too tiny for current instruments to
detect.
However, the distances to stars in other galaxies can be measured with the
help of a clever trick. Knowing the distance and measuring the apparent
brightness of a nearby star tells astronomers its actual or intrinsic
brightness. Armed with this information, they can observe the apparent
brightnesses of similar kinds of stars in other galaxies and thus—assuming
the intrinsic brightness is the same—they can establish the distances.
This is where it gets interesting for the age of the Universe question.
Galaxy distances yield cosmology’s most contentious number—the Hubble
constant. As a galaxy’s light travels through space toward Earth, space expands
and stretches the light wave, so the farther the galaxy is from Earth, the
farther its light has travelled through space and been stretched, or
red-shifted. The Hubble constant is the ratio of the amount the light has been
red-shifted to the distance of the galaxy. It expresses how fast the Universe is
expanding, and hence its approximate age. If the Hubble constant is high, as
Freedman’s team has found, the Universe is expanding fast and so must be young,
because it has taken relatively little time since the big bang to attain its
present size. But if the Hubble constant is low, the Universe is expanding
slowly and is old.
In theory, to determine the Hubble constant, all you need to know is a
galaxy’s red shift and its distance. The red shift is easy, but the distance is
not. Because of controversies over how best to measure the distance to far-off
galaxies, the modern Hubble war has raged for over 20 years. Passions run high.
“Calm down, Allan, calm down,” Allan Sandage, staunch advocate of a low Hubble
constant, says to himself. “You’re not an evangelist, Uncle Allan.” He pauses,
then adds: “You bet I am.”
Sandage, also of the Carnegie Observatories, has been working on the Hubble
constant for decades. Back in 1975 he and Gustav Tammann, a Swiss astronomer,
published a Hubble constant of about 55. This implies that the Universe’s age is
12 to 16 billion years—old enough to accommodate 15-billion-year-old
stars. But a year later, the late Gérard de Vaucouleurs claimed the
Hubble constant was twice as high, implying that the Universe is half as old. By
the early 1990s, many investigators had calculated that the Hubble constant was
around 80, less than de Vaucouleurs’ figure, but still high enough to be
troublesome: it puts the Universe’s age at 8 to 11 billion years.
The Hubble war centres on distances, and no distance indicator is more sacred
than a class of stars called Cepheids, yellow supergiants that periodically
expand and contract. Cepheids have many virtues. They are easy to find because
they brighten and fade dramatically. What’s more, they are young and shine so
brightly they can be seen in galaxies other than our own. But most importantly,
the period at which they pulsate is closely related to their intrinsic
brightness: the longer the period, the greater the brightness. So measuring the
period, which is easy, reveals the Cepheid’s intrinsic brightness. Comparing
this with the apparent brightness yields the distance to the Cepheid—and
to its galaxy.
There is a catch, however. To calibrate exactly what luminosity goes with
what period, astronomers must know the distances to at least a few Cepheids.
Unfortunately, Cepheids are so rare that even the nearest, Polaris, also called
the North Star, is 430 light years away, and its parallax is too small for
ground-based astronomers to measure accurately. Instead, astronomers had to
estimate the Cepheid period-luminosity relation by observing Cepheids in star
clusters of known distance and in the Large Magellanic Cloud, a nearby galaxy
whose distance has been determined by other techniques.
Hipparcos, however, was able to detect the parallaxes of over twenty
Cepheids. “The local Cepheid calibration is now fixed much more firmly,” says
Michael Feast of the University of Cape Town in South Africa. In 1997, based on
Hipparcos’s parallaxes, Feast and Robin Catchpole of the Royal Greenwich
Observatory announced that the Cepheids were 10 per cent farther away than had
been thought. This would mean the Hubble constant is 10 per cent lower, and the
Universe 10 per cent older.
But most astronomers disagree. “I don’t take the Cepheid result that
seriously,” says Michael Bolte of Lick Observatory in California. “They had lots
of measurements that were right on the hairy edge of what Hipparcos could do,
and it’s hard to average lots of lousy measurements to get the right answer.”
And a team led by René Oudmaijer of Imperial College in London claims
that parallax errors led Feast and Catchpole to overestimate the Cepheids’
brightnesses.
Feast defends his work. “Yes, the Cepheids are far away,” he says, “but I
think it is possible to analyse the data in the way we did to get away from
biases.” He also disputes another recent attack. Freedman and Barry Madore of
the California Institute of Technology point out that dust dims the Cepheids’
light. When Madore and Freedman analysed the Cepheids for which they have both
Hipparcos parallaxes and data about intervening dust, they found that the
Cepheids were no brighter than previously thought. Feast counters that Madore
and Freedman’s extraction of a few Cepheids from the full sample creates a bias
that invalidates their work—a contention Madore and Freedman reject.
In view of all this controversy, most researchers feel that Hipparcos has
done little to resolve questions about the Hubble constant and hence the age of
the Universe. But there is another way to solve the age conflict: if you can’t
make the Universe older, simply make the troublesome stars younger. And that’s
just what astronomers have done. Some of the oldest stars reside in star-packed
clusters called globulars. Until recently, models of how stars evolve predicted
that the globular clusters were 15 billion years old—older than the
Universe. But those ages have now come down.
Soon after Feast’s Cepheid announcement, Neill Reid of the California
Institute of Technology reported new Hipparcos distances for the globular
clusters. Knowing how far away the clusters are tells astronomers their
intrinsic brightness. This in turn reveals the clusters’ ages, because the
brightest stars die first, followed by dimmer ones, so the clusters grow
steadily dimmer with age. Unfortunately, even the nearest globulars are
thousands of light years away from Earth, so Hipparcos could not detect their
parallaxes. Instead, Reid had to use the trick of measuring the intrinsic
brightnesses of nearby stars and looking for the same kind of stars in the
far-off clusters.
Bright young things
Cepheids are short-lived stars and there are none left in the ancient
globular clusters, so Reid used much fainter stars called subdwarfs, which
abound in globular clusters and can also be found relatively close to the Earth.
When Reid analysed these nearby subdwarfs, he found that their intrinsic
brightnesses implied that the oldest globulars were 10 to 15 per cent farther
away than previously thought. This means their stars are 20 to 30 per cent
brighter, and the clusters billions of years younger. The oldest clusters, says
Reid, are about 12 billion years old.
Freedman, who criticised the Cepheid data, thinks the subdwarf work is more
sound. “Many of the parallaxes there are of high quality,” she says. “It looks
to be a fairly convincing result.” But Reid’s work has also been attacked.
Frédéric Pont of Geneva Observatory, Switzerland, and his
colleagues also analysed Hipparcos parallaxes of nearby subdwarfs but calculated
that the distances were smaller—similar to previous estimates. However,
because of improvements in the theoretical models of stellar ages, Pont’s group
also says the globulars are younger than had been thought, about 13 billion
years old.
Whatever the subdwarf verdict, the estimates of the ages of globular clusters
have decreased, independent of the Hipparcos results. Just two years ago, Don
VandenBerg, of the University of Victoria in Canada, working with Michael Bolte,
and Peter Stetson of the Dominion Astrophysical Observatory in Canada, concluded
that the oldest globulars were 15 billion years old. Now he says that
improvements in stellar models have reduced the estimated ages of globular
clusters by billions of years.
There are several reasons. For example, old stars are known to have higher
ratios of oxygen to iron than young stars, and incorporating the correct ratios
into the models makes the globular stars younger. Another factor is helium, the
second most abundant element after hydrogen. Most stars convert hydrogen into
helium and die soon after the hydrogen in their cores is used up. But because
helium is heavier than hydrogen, some of the helium outside the core sinks down
into the core, speeding up a star’s death—a phenomenon that previous
stellar models ignored. These and other effects, says VandenBerg, mean that the
globular clusters are several billion years younger than had been thought. Even
using the results of Pont’s group, which VandenBerg favours, the latest models
give an age of only 13 billion years for the oldest globular clusters.
So it seems that the new estimates of stellar ages have helped save the
day—even the highest estimates of the Hubble constant produce a Universe
almost old enough to contain the globular cluster stars. But the questions
remain—just how big is the Hubble constant and just how old is the
Universe? Even if Hipparcos’ measurements of the Cepheid scale were universally
accepted, they would not resolve the controversy over the Hubble constant,
because both sides use the same scale. Nevertheless, the two sides do seem to be
converging. Freedman’s team, which claimed in the early 1990s that the Hubble
constant was in the low 80s, is now reporting lower values, implying an older
Universe. She and other astronomers have used the Hubble Space Telescope to
measure the distance to many more galaxies containing Cepheid stars. These
results are used to calibrate other distance indicators that reach farther than
Cepheids.
At the moment, Freedman’s team thinks the Hubble constant is about 72, which
implies a Universe between 9 and 12 billion years old. This is closer to the
current Sandage value, 56, which implies a Universe 12 to 16 billion years old.
“They’re going down at the rate of about 2 units a year,” Sandage notes wryly.
Sandage’s favourite technique for measuring the Hubble constant employs type Ia
supernovae, which occur when white dwarf stars exceed a critical mass and
explode. This mass should be the same for all white dwarfs, which means that the
explosions should be equally luminous, making them excellent distance
indicators. Furthermore, they pour out so much light that they can be seen
billions of light years away. By looking for Cepheids, Sandage can calculate the
distances to nearby galaxies that have spawned type Ia supernovae. These
distances reveal exactly how luminous a type Ia supernova is, allowing type Ia
supernovae in more distant galaxies to be used as a measure of their galaxies’
distances—and thus of the Hubble constant.
Best of three
“Type Ia supernovae are one of the most promising means of measuring the
Hubble constant,” says Freedman. “But would I put all of my eggs in that basket?
Absolutely not.” She prefers to hedge her bets by choosing several of the best
distance indicators and averaging the results. Sandage counters that he uses
three other distance indicators as well, all of which yield the same result as
type Ia supernovae.
David Branch of the University of Oklahoma has no problem adopting just one
indicator. “I’ll put my faith in the best one,” he says, “and I think that type
Ia supernovae are at present the safest determination of the Hubble constant.”
In a review article due to appear this autumn, Branch says that type Ia
supernovae give a Hubble constant of 60, which puts the Universe’s age at 11 to
15 billion years.
Robert Kirshner of Harvard University, whose type Ia supernovae results
indicate a Hubble constant of about 65 notes that the two sides have come closer
together. “I don’t think it’s just social pressure,” he says. “I think the
measurements really have got better.” Key among these has been the detection of
Cepheids in many more galaxies.
But Canadian astronomer Sidney van den Bergh, a long-time observer of the
Hubble war, is less optimistic. “I just feel somewhat more unsure now than I did
a few years ago,” he says, “simply because the best distance indicators, the
things that we really trust—like the Cepheids—are now up in the air
at the 10 per cent level. That’s sort of a sad situation, when your most basic
measuring sticks are in question.”
Freedman and her colleagues will soon use a new camera on the Hubble Space
Telescope to observe Cepheids at longer wavelengths, which penetrate dust better
and so more accurately indicate Cepheids’ true brightnesses. Eventually,
astronomers hope for a better satellite than Hipparcos, capable of more precise
measurements of the parallaxes of nearby Cepheids—and even globular
clusters and the nearest galaxies—and for a better space telescope than
Hubble, to detect Cepheids in very distant galaxies.
Meanwhile, many
astronomers celebrate the recent progress on determining the numerical value of
the Hubble constant. “It’s really quite amazing to see the convergence,” says
William Harris of McMaster University in Canada, noting that most values for the
Hubble constant now fall between 60 and 80. “Seventy is a good round number that
almost anyone can live with.”
Can Sandage live with a Hubble constant of 70? “Of course not!” How about 65?
“No! Why do you want me to do that? The precepts that give that are absolutely
ɰDzԲ.”
The problem of the Universe being younger than its stars may be over, but the
impassioned fight over the Hubble constant goes on.
- Further reading:
The Hubble Wars Chapter 15 of Lonely Hearts of the Cosmos
by Dennis Overbye (Macmillan, 1991) - Older than the Universe? Chapter 17 of The Alchemy of the Heavens
by Ken Croswell (Oxford University Press, 1996)