Info

figure 6.2 The Hubble diagram for type la supernova«. Note that the velocity is proportional to distance, as noted in 1929. Hubble's original Hubble diagram (figure 5 4) extended only out to 2000 kilometers per second, where the individual motions of galaxies added w the scatter This Hubble diagram extends out to 30.000 kilometers per second, one-tenth the speed of light, where the cosrrological Hubble flow is large compared to any galaxy's individual motion among its neighbors Courtesy of Adam RJess, Harvard-Smithsonian Center for Astrophysics several well-measured SN la to construct a template. The assumption underlying Bruno's work was that all the SN la were identical. This was plausible in the 1980s: there was a theoretical reason to think exploding white dwarfs at the Chandrasekhar limit were all the same and the observations, mostly photographic, were nor good enough to see subtle differences clearly. After finishing his thesis, Bruno came to work with me in Cambridge as a postdoc at the Harvard-Smithsonian Center for Astrophysics. Data assembled by our ream and techniques we developed in the 1990s helped make the SN la the best standard candles for measuring distances to galaxies.

Bruno had some of the traditional Swiss characteristics; he was very careful, thorough, and self-critical. These are good properties in someone dealing with light curves, where there are many ways to err. But Bruno also had some not-so traditional properties. The joke is that Switzerland doesn't have an army, it is an army. Every male is supposed to serve in the army and then keep a rifle (not just a Swiss army knife) at home in good condition for national defense. Einstein was a pacifist during World War I, when he was in Berlin, and a spokesman for world peace even after he helped instigate the Manhattan Project to build a nuclear weapon. But when he was a young man, Einstein was called for the Swiss army— and he was rejected for flat feet and varicose veins. Bruno was also called for the Swiss army, and his feet and veins were fine, but he chose "service without a weapon," as a matter of principle. Bruno is a person who thinks for himself and stands up for his own opinions. These are good properties if you are a Swiss citizen, Gustav Tam-mann's student, or my postdoc.

For many years, the impeccably dressed, careful, and energetic Tammann had collaborated fruitfully with Allan Sandage. By the late 1980s, they were embarked on a program with IIST to discover cepheids in galaxies with well-observed supernovae where Bruno's template was useful. The idea was simple. You make a list of galaxies that have had well-observed supernovae. You use Bruno's template to figure out the peak apparent brightness. Next, select the galaxies that are close enough so that HST can find their cepheids.

Then comes the heavy lifting: you must convince the Space Telescope Time Allocation Committee to let you take many HST images of your target galaxy to find the cepheid variables and to determine their periods and apparent brightnesses. Armed with the apparent brightness of cepheids of known period in a galaxy, you can figure out its distance, just as Hubble did, by comparing those stars to the cepheids in the Large Magellanic Cloud. Now you turn the problem around: if you know the distance to the galaxy and, from Bruno's work, the apparent brightness of the supernova, you can do the arithmetic to see what the intrinsic power output of a supernova is. Do this for enough galaxies to average out the errors (if they are purely Gaussian errors) and you learn the true brightness of SN la. Present values are somewhere around 4 X 105 solar luminosities.

The last step is to use a distant set of supernovae found in galaxies out in the Hubble flow where redshift reflects cosmic expansion, not random motions. Use the known luminosity and measured apparent brightness to compute the distance to each one. Divide the velocity (in kilometers per second) by the distance fin mega-parsecs), average to decrease the errors, and, tot la t you have the i lubble constant.

This program sounds simple, but it isn't. First, supernovae are rare. Since a supernova of this type goes off in a galaxy once every century or so, and we've only been aware of supernovae since Zwicky and Baade's pioneering work in the 1930s, there is only a small number of good cases where a well-observed supernova erupted in a galaxy within HST's limited range to detect the cepheids. Allan Sandage, Gustav Tammann, Abi Sahaf and their collaborators, who have been carrying out this program with HST, list 9 supernovae discovered in nearby galaxies. They find a Hubble constant of 60 ± 6 kilometers per second per megaparsec, up from earlier measurements that placed the Hubble constant in the 50s.

Wendy Freedman, three decades younger than Sandage, but also working at the Carnegie Observatories, has been the leader of a "Key Project" team measuring the Hubble constant with HST. Observational cosmology is not exactly a contact sport, but it helps til be tough and competitive if you are working in the same field as Allan Sandage, and at the same institution, and especially if you get a different answer. Wendy and her sister used to play on the University of Toronto's women's hockey team—as an experienced right wing, the rough and tumble of astronomy doesn't bother her too much. Wendy's group has used a number of other methods besides supernovae to measure distances to galaxies. She's looking to see whether they agree, to get around the particular systematic problems of each technique. She calls this "cross-checking" her data, which is bad in hockey, but good in observational astronomy. When the Key Project team first reported their results in 1994, they got relatively high values for Htl—near 80 kilometers per second per megaparsec, which corresponds to a meager allocation of 12 billion years since the Big Bang. Her group's present value of H0— 72 kilometers per second per megaparsec is widely asserted to be good to 10%. History suggests that we are always confident, but rarely correct. Or, perhaps we really are nearing the end of smoking out errors in the Hubble constant. This is not quite in agreement with Sandage's team, but the differences are getting smaller.8

Fundamentally, the accuracy of the Hubble constant, H„, and the expansion age, t(J, inferred from l/H0 for supernovae depends on distances to nearby galaxies that have hosted SN la, which can also be measured with cepheids. The distances to those galaxies depend on comparing their cepheids with the same type of star in the Magellanic Clouds. This step-by-step measurement of the universe leads to an odd situation—our knowledge of the size and age of the entire universe depends on measuring the distance to the nearest galaxy, the LMC. It is frustrating but true that we haven't got this local problem completely solved. The most recent revisions to the Hubble constant depend on improved measurements of cepheids in the Magellanic Clouds, not on supernovae 10,000 times farther away.

How would we know if today's distance scale based on cepheids were wrong? One way is to compare independent methods of measuring distances to the same galaxies. Cepheids underpin most methods for finding the distances to galaxies, yet there is a handful of ways to measure extragalactic distances that doesn't depend on these stars. If independent methods give the same answer, perhaps both methods are measuring the distance accurately. If they disagree, then somebody is wrong.

Supernovae provide two distance-measuring methods that have nothing to do with the distance scale based on cepheids. First is the amazing ring in SN 1987A. When the supernova was first sighted in February 1987 George Sonneborn, at Goddard Space Flight Center, and I compiled a detailed record of the supernova's changes with the International Ultraviolet Hxplorer satellite. The first change was a little disappointing: the ultraviolet light from the supernova faded very rapidly, plunging by a factor of 1000 in the first three days after discovery. But then, after about 90 days, the ultraviolet spectra began to show something curious: narrow emission lines from highly ionized nitrogen atoms. The fact that SN19H7A showed emission lines suggested that this light was coming from gas that had been excited by the supernova explosion. The fact that the lines were narrow meant that the velocity range of the emitting gas was very small. Since the supernova itself was ripped apart by a violent explosion that sent the outer layers flying out at 10% of the speed of light, small speeds for the emitting gas ruled out supernova debris. What was this stuff?

My Swedish colleague, Claes Fransson, had a good idea for something simple that would cover the facts. What if there were a shell of gas around the supernova, perhaps exhaled by the pre-supernova star? If it was located at the right distance, then the powerful flash of light from the explosion would take many months to reach that shell, excite it, and make it glow. A flash of ultraviolet light would rip the electrons off nitrogen atoms to ionize them, but it wouldn't give the gas much of a kick up lo high speeds.

If this story was right, then the emission lines we were seeing would gel stronger over the next several months. If the shell were big enough, light travel time would matter for us in 1988 just as it did for Ole Homer in 1676: the near side of the shell would be many light-months closer than the far side. In fact, it took about 400 days for the lines to reach their maximum strength, which we took to mean that the pre-supernova star was surrounded by a spherical shell with a diameter of about 400 light-days.

All of this was very interesting information about the last gasps of a massive star, but it did not yet provide an independent distance to the Large Magellanic Cloud. That came after the 1990 launch of the Hubble Space Telescope. There were hints from early ground-based data that there was something near the supernova that lit up in the months after the explosion. The earliest images from HST, even with its flawed initial vision, showed that supernova 1987A was surrounded by glowing gas, just as Claes had predicted. (See figure 3 3.)

Except, as usual, nature was wilder than our imaginations. The gas was not in a simple shell around the explosion, but in a ring, presumably the inner boundary of a flattened donut of gas. Fven with the blurry version of an uncorrected 11ST picture, you can measure the angular size of the ring. It turns out to be about 1.6 arc-seconds. Since the blurring effect of the Earth's atmosphere is typi cally around 1 arcsecond at a good astronomical site, 11ST, sited above the atmosphere, was essential for this measurement. Now, if you know how large the ring is from timing the rise to maximum of the narrow emission lines in the IJV, and you figure out the tilt of the ring from its shape or the light curve, and you also know the angle the ring covers as viewed from our galaxy, it is not hard to compute the distance to SN 1987, and hence to the Large Magellanic Cloud.

We found that the distance to the LMC is about 165,000 light ycars, which is the same distance that Wendy Freedman and her asscx:iatcs have been using for the beginning of the cepheid-based distance scale. So we agree, using completely independent paths. It could be chance, but it could be we are both right,v

Wc invented another method based on supernovae to check the cephcid distance scale. While SN la are thermonuclear explosions in white dwarfs, type II supernovae result from the collapse of a massive star. When the outside of a SN II is ejected, it is still mostly hydrogen. The properties of the expanding, cooling atmosphere can be computed in cletail by a very smart graduate student. Ron Eastman, who worked with me at the University of Michigan and at Harvard before joining the scientific staff at Liv-ermore, did this computation in 1989 and worked out a refined methfxl for comparing the models to the data for SN 1987A. Repeated measurements of the temperature, speed, and brightness of the supernova atmosphere supply enough information to figure out how large the atmosphere is, and to compute the distance to the explosion. For SN 1987A, the distance comes out again to be near 165,000 light-years, in good agreement with the conventional distance.1"

In 1994, for his Ph.D. thesis, my Harvard graduate student Brian Schmidt applied our "expanding photosphere method" to alt the available data for explosions since 1969. Interestingly, some of the galaxies with SN II data and expanding photosphere distances arc-also galaxies in Wendy Freedman's Key Project sample. The results agree very well, which suggests either we are both doing something wrong or both doing something right. Since the two approaches are completely independent, we suspect this is a clue that we are doing something right.

More than 20 years after I gave a talk to our Visiting Committee and had a dull lunch with the aged Harlow Shapley, I was the department chair at I Iarvard, trying to set up the program for another Visiting Committee. Wallace Sargent from Caltech was the chair of the Visiting Committee. Since I had been a graduate student at Caltech, Wal knew me pretty well, and I was eager to show that there were good things happening at Harvard. Brian Schmidt seemed the natural choice to talk to this outside group. He had an independent measurement of the Hubble constant, which was a hoi topic. Plus, Brian is a charming guy, a lively speaker, and had proved to be a very good teaching assistant. If he could deal with Harvard undergraduates, I reasoned he could deal with Caltech professors. Brian gave an excellent description of his work, impressing the committee with the science and wowing them with his presentation.

At the end, Wal, wrapping up, said, "Well, today we have seen the debut of Kirshner, Junior."

Both Brian and I turned equal shades of crimson. I wonder who sat next to Brian at lunch. Perhaps some legendary figure from the past. I don't know because he must have been seated out at the edge of the rcx>m.

So, what time is it? On the face of it, a cosmic age of 14 billion years from a Hubble constant of 70 is in the same ballpark with the ages of the globular cluster stars at 12 billion years or the cooling rime for the oldest white dwarfs stars in the Milky Way at 10 billion. If the universe has been expanding at a constant rate, then the cosmic ages seem concordant. Stars that formed a few billion years after the hot beginning are younger than the universe as a whole. This is good, because you should not be older than your mother.

But the concordance of ages is spoiled if the universe has been slowing down. In that case, the present expansion is a treacherous guide to the rate since the beginning of time. In fact, this was a very serious problem through the 1980s and 1990s. If the Hubble constant wen? really 80 kilometers per second per megaparsec, as

Cosmic Expansion Astronomy

figure 6.3. Brian SehmTdt explain« the expanding photosphere method to his Ph.D.

advisor in 1994 The computer screen shows Schmidt's Hubble diagram for type II supernova«, derived using the expanding photosphere method to measure distances Courtesy of Harvard News Office.

figure 6.3. Brian SehmTdt explain« the expanding photosphere method to his Ph.D.

advisor in 1994 The computer screen shows Schmidt's Hubble diagram for type II supernova«, derived using the expanding photosphere method to measure distances Courtesy of Harvard News Office.

suggested by Wendy Frcedman's team in their first report in 1994, and if ft were equal to one, as many theorists believed, then deceleration would make the age of the universe embarrassingly small compared to the ages of stars. If the universe had been expanding more rapidly in the past, just like observing a tiring marathon runner, you would overestimate the time on the course if you ignored the slowing down. In this case, £2 = 1 would imply a true cosmic age close to 8 billion years, which was not in good accord with evidence from stars. Something was wrong with this picture. Was it Ho or w as it ft?

And finally, when do we get there? If ft is low, we never do— the universe expands without limit. If ft is one, universal expansion slows, but never turns around—wc get closc, but wc never arrive. And if ft is bigger than one, at some distant time in the future the universe will reach a maximum extent. We will have arrived, but we will also see the awful prospect of what lies ahead: universal contraction, the undoing of all the effects of hundreds of billions of years of cosmic change in a fiery Big Crunch. All of this sounds like the stuff of mythology, but we have slowly expanded the boundaries of measurement and rational discussion. The problem is nor conceptual, it is quantitative: can wc make the measurement well enough to trust the answer? And finally, those conclusions ignore the cosmological constant. If the total energy density of the world is made up of some matter that gravitates, and some dark energy that makes the universe spring apart, all bets arc off.

Was this article helpful?

0 0

Post a comment