Figure H.I. The work on the acceleratir^g universe was Science Magazine's "Science Breakthrough of the Year" for 1998. Reprinted with permission from SciencE, vol. 282. December 18, 1996 IIlustration. John Kascht, Albert Einstein™ represented by the Roger Richman Agency, Inc , Beverly Hills, CA Copyright 1998 American Association for the Advancement of Science.

from our claim that A is real. Aside from the embarrassment, there would be regret. We were beginning to like A.

And we weren't the only ones. Science, the leading science journal in the United States, sclectcd the accclcrating universe as its top "Scicncc Breakthrough of the Year" at the end of 1998. Brian Schmidt won Australia's first "Malcolm Mcintosh Prize for achievement in the physical scienccs." Brian was in demand, at least in Australia, where the Australian Broadcasting Company callcd him in for an interview on the "The Age of the Universe" show, hosted by John Doyle, known to a wider audience as the TV announcer in the movie Babe.

For another voice, they asked me to come to a TV studio at 11 p.m., which was convenient for the Australians. Jayne and I had tickets to the Red Sox game at Fenway. Stupidly, I drove so I could get over to WGBH after the game, and even more stupidly I parked in one of those lots marked "E-Z Out" where they block your exit with as many cars as possible after you have left the scene. The Cleveland Indians and the Red Sox were engaged in a titanic struggle as the clock ticked past 10:30. Relief pitchers paraded in for both teams. Contrary to the oath I had sworn to my grandfather in the Fenway bleachers in 1959, I needed to leave the game before the last out to be on Australian TV on a program hosted by a comedian. My car wras wedged in the parking lot. I backed and filled while Jayne gestured with an imaginary steering wheel, slowly ooching around to the angle where I could escape. Then I heard a sickening sound of metal bending, as I warped the curve of my Saab's dcxir on the bumper of a Chevrolet. There was no going forward and there wras no retreat. At that instant, i realized it is better to do the work than to talk about it on 'IV.

Plenty of people were working on the accelerating universe. There was a flood of theoretical papers concerning the dark energy, which was quickly seen to be a frontier of physics.1 The idea that the universe was composed principally of vacuum energy with negative pressure, required by these astronomical observations but nowhere seen in terrestrial laboratories, meant that an important problem in basic physics was not yet solved. Presumably, this has something to do with the fact that there is not yet a complete "theory of everything" that treats gravitation on the same quantum footing as clcctromagnetism, the weak nuclear force, and the strong nuclear force. Effects of virtual particles and their antiparticles that spring out of the vacuum and annihilate one another are staples of theoretical physics for electromagnetic effects. Amazingly enough, the Casimir effect and the Lamb shift arc both laboratory experiments that show these wild ideas have real consequences that agree with the facts. The vacuum is a lively place.

For the gravitational equivalent of vacuum effects, there are as yet no laboratory experiments and no well-established theory, just the evidence from supcrnovac for an accclcrating universe. But there are adventurous ideas. My union brothers at the Institute for Theoretical Physics had been among the pioneers of string theory, which people tell me works best in a bulky 11-dimensional space. There is some hope that an explanation for the small value of the cosmological constant in our membrane of three space dimensions and one time dimension might drop out of this cogitation. So I have been skimming the abstracts of papers with titles like "A Scalar-Tcnsor Brane World Cosmology." Despite holding a union card in the International Brotherhood of Theorists, my true orientation is much like that of Hale: "I confess the subtleties of the theory are altogether too much for my comprehension." Astronomers learn general relativity now, so Hale's modesty seems quaint. Someday wc may have to understand 11 -dimensional M-theory or its dcscendcnts to understand cosmology, though for now the ferment is among "the very few competent to discuss the matter with authority."

The idea of looking for the fingerprints of dark energy in super-novae beyond a redshift of one was not ours alone. The Supernova Cosmology Project was still months ahead of us in many ways, and they had already found one, SN 1998eq. During the interim before supcrnovac arc reported to the International Astronomical Union to rcccivc their designations, each team uses its own nicknames for the candidates, a little like the names for hurricanes. Saul Perlmutter is a cultured fellow, a Harvard graduate and a violinist. Their team decided to call their really high redshift candidates according to an alphabetical list of composers, starting with Albi-noni.; So far, the data on Albinoni, which is said to be at a redshift of 1.2, have not been published, though in talks they show it lying below the line you'd expect if misleading effects like age or dust arc most important, and in the general direction for a genuine cosmological effect.

During 1999, John Tonry of the University of Hawaii led the charge toward higher redshift for our high-z team. John is a creative astronomer who pcrfcctcd a new way to find the distances to galaxies based on how grainy they look. I le is an independent guy and a software wizard of the first magnitude John has deftly constructed his own version of exquisite sky-subtraction software and recxam-

ined the whole Rube Goldberg scaffolding of the high-z team's way of doing things. This incrcdiblc effort not only protects us against a missing K or minus sign in the computer code, but has improved our ability to find supcrnovac at redshifts of one and beyond. As you look for supcrnovac at higher redshift, you not only must find fainter objects, you must look further to the red, bccausc that's where the light from very distant supcrnovac is shifted. Then you have to deal with more light from the sky, which glows in the near infrared, and with CCD detectors that generally do not work as well at those wavelengths as they do for observations at visible wavelengths

So everything is working against you: the objects are fainter and redder. Because they are redder, you have to use detectors where they arc less effective, and you have to contcnd with a brightly shining sky even when the moon is not a factor. But the reward for doing these difficult observations would be to find supernovae that reveal the autograph of A. It seemed worth the effort.

To tune the search for higher redshift, we changed tactics. Since the tai^et supcrnovac were going to tx? fainter than the ones wc had previously sought, wc made our exposures longer. Sincc they were going to be at higher redshift, we shifted the filter of our exposures farther to the red. And since the sky was more of a problem, we realty needed the improved software that Brian Schmidt and John Tonry developed to subtract one frame from another. In keeping with our policy of incorporating every good idea we can find, wc improved our image subtraction by incorporating a scheme developed by Christoph Alard into our data pipeline.

On 2 and 3 November 1999 our team discovered 20 supcrnovac using the giant 12,000 x 12,000-pixel CCD array camera on the Can-ada-France-Hawaii Telescope by subtracting images taken the month before. Confirming images were taken at Cerro Tololo, using the 4-mctcr Blanco Telescope. Spcctra obtained in the next 10 days at the Kcck by John Tonry and by Alex Filippenko showed that 12 of our candidates were SN la. Two of these proved to be especially interesting targets with redshifts greater than one. To provide a little contrast with the other team, we took a lowbrow path and named our candidates after cartoon characters instead of cultural figures.

So wc had Rocky and Bullwinkle, Boris and Natasha, and Fearless Leader as candidates. The ones that were most interesting were Velma and Dudley Do-Right. Although measuring each spectrum was tough going, these appeared to be at redshift 1.05 and 1.2.

In December we continued to follow the decline of these super-novae to measure the shape of the light curve. At high redshift, time is stretched out by expansion just the way the wavelengths of light are strctched. This means that in a month of our time, the supernova only ages by two weeks. So our December observations were well placed to see wrhat the supernova was doing in the first two weeks after maximum light. Since I was in Hawaii anyway for my honeymoon, having just gotten married during the full moon in Decem-bcr, I went up to the summit of Mauna Kea to observe with my graduate student Saurabh Jha. My wife, Jayne Loader, compensated for the psychic injury by having her toenails painted at the Mauna Kea Beach Hotel.

Saurabh had already mastered all the details of observing at the University of Hawaii's 2.2-mctcr telescope on Mauna Kea. At sunset, the mountain is a fabulous place: an extinct (wc hope) volcanic landscape without vegetation. The clouds of the trade winds were below lis, and above us only 6o percent of the air that Jayne was breathing at sea level. By midnight, the absence of air was beginning to bother mc. Legend has it that you can't think straight at the 13,796-foot summit. But if you can't think straight, why should we believe what you say about how you are thinking? In any ease, I was getting a headachc, my gums were sore, and I felt a little short of breath. But Saurabh seemed alert and we were getting cxccllent data. It was fun. If you're running the show, you get to choose the music, so Saurabh was playing an austere minimalist composition by Steve Rcich on his CD player. As the rhythmic figures subtly wove into patterns, I tried to ignore my aching gums and to think about A. We were working at a high enough redshift to see beyond the era of acceleration, back to the time when matter ruled the universe. I could see the galaxies forming in dark matter lumps as they rushed outward from the Big Bang, the whole expansion slowing due to the tug of dark matter, and then the steady push of the cos-mological constant shifting the balance and driving space outward faster and faster, fading to red and leaving us alone in the darkness, gasping for breath. Or maybe I was just oxygen-starved. When I awoke, the integration was done, and it was time to shift to the next target.

The University of Hawaii tests your mettle by making you ol>-scrvc at the summit, but the Keck Observatory compensates for the effects of altitude by spending money to let the observers work at sea level. Since you never need to touch the telescope anyway, controlling the instrument through a computer console, you might as well get some oxygen. Technicians at the summit babysit the telescope, but the observers arc down in Waimca, using a fast computer link to control the data-taking instruments. If they were cm the summit, they would also be in a control room, using identical computers, so there's not much to lose and a lot to gain in mental alertness and physical comfort. This gives you more chance to choose your next move intelligently, taking into account the weather, how sharp the images are, and your list of targets.

There's a slow-scan 'IV link that lets you observe the telescope operator sitting patiently in a chair, and for the operator to sec you, too, frantically trying to calibrate the latest observation fast enough to decide what to do next. That communication is gcx)d enough. You don't become close friends with Wayne the operator, encountering him only in this distant way, but that's a reasonable price to pay for a brain that works.

Sometimes strange things happen—one night it was pouring down Hawaiian rain in Waimca while we were taking spectra of supernovae. Up on the summit, in Wayne's world, they were above the moisture of the trade winds and observing conditions were excellent. As we were debating which object to do next, Barbara Schaefcr, the head of all the Keck telescope operators, looked into the data room on her way home from a late night at the office. We described what we were doing—going after redshift one supernovae. We were trying to decide whether to do another of these nearly impossible objects or to do something easier where we would be sure to get a useful result. Barbara composed her face in serenity, placed her palms together, and, in the cheddar-sharp nasalities of upland Wisconsin intoned the koan of Keck: "When conditions are good, do the hard thing." Then she went home to her cats.

The logical extension of this mode of observing will be to use fast network connections so you can observe in Hawaii without leaving Berkeley or Pasadena or Cambridge. Instead of amusing the scantily clad tourists and locals by arriving in Kona with a down parka and heavy boots for the summit, or sleeping through a perfect day in Waimea, you will be at your officc with the phone ringing, and students waiting outside the door, where the only respite will be the chance to sleep through faculty meetings. This will be followed by a long night of observing. We will call this progress.

There are many large new telescopes coming into operation, including the twin Gemini 8-mctcr telescopes on Mauna Kea and in Chile. To put the Gemini Observatory on the scientific map, they organized a conference on "Astrophysical Timescales" and invited me to speak. It seemed like a great place to discuss how A affects the estimate of time elapsed sincc the Big Bang, so I said yes. But when it came time to plan the travel, I realized there were other timescales 1 could not alter. There was no way to get to Hilo, Hawaii and back without missing a lccturc in my undergraduate class at I larvard. They don't ask us to teach all that much, so I try to be there every time. Instead, John Tonry hopped over from Honolulu to report on the results from our high-z observations at the Gemini conference. This was better, and not just because of logistics: John had been doing most of the work and it seemed right that he should give the talk.

Although the final analysis was not quite done and John's conference report has not been refereed, so it doesn't have the weight of a real journal article, it does show which way the finger of fate is pointing. If the distant supernovac were yet fainter, it would be bad news for A. If the distant supernovac came out a little brighter than that, it would be the signature of cosmic deceleration in the early universe, and a clear sign that the effects we were observing were cosmological, not the result of "a changing population of supernovae" feared in an ancient News and Views. John showed that the data from Dudley Do-Right, at z = 1.2, and his high-redshift friends, came out a little bit brighter than you'd otherwise expect. In the context of a cosmology with A, our new data favor a universe that is accelerating now, but was decelerating in the distant past, 7 billion years ago. This stop-and-go universe is good news for dark energy.

Since the Gemini meeting in Hilo was a conference on cosmic timescales, John also spelled out what A means for the conncction between cosmic expansion and cosmic age. If you have a Hubble constant of 72 kilometers per second per megaparsec, then l/Nv is 14 billion years. If there were no acceleration and no deceleration, that would be the real elapsed time since the Big Bang. In an = 1 universe completely dominated by dark matter, the deceleration of the universe means that the present rate of expansion is lower than the average, so the universe is younger than it currently appears, with an elapsed time of about 9 billion years. This conflict with stellar ages of around 12 billion years was one of the rhetorical gambits advanced for A before the observational evidence from super novae.

In an accelerating universe, the real age could be larger than the apparent age, but in a stop-and-go universe, as suggested by the high-z data that John Tonry presented, it could go either way. If the slowing down were more important, the universe would lie younger than 14 billion years, and if the speeding up were more important, the universe would be older.

By coincidence, if Qm = 0.3 and £2A = 0.7, which is a good representation of the supernova data, including our new points at red-shift one and beyond, then the slowing down and the speeding up just about balance, and the elapsed time from the Big Bang to now, is just 14 J + 1.6 billion years for a Hubble constant of 72. So, after all this lucubration, including cosmic deceleration followed by cosmic acceleration, it looks like the answer you get using third-grade arithmetic is the right answer for the age of the universe. And that answer is in good accord with the ages of objects in the universe. So far, so good.'

But if going to redshift 1.2 is good, wouldn't it be better to go even farther into the past? The effects of deceleration would be larger, the contrary effects of pixie dust would be larger, and the difference between them would be even more impressive evidence that we were seeing cosmological effects based on the history of cosmic expansion, not illusions caused by stellar ages, chemistry, or absorption. But John Tonry had already led us pretty close to the limit of what can be done from the ground: we were using the world's largest telescopes at the world's best sites. The next step needed to be taken above the Earth's atmosphere.

Although the Hubble Space Telescope is not good for a wide-angle search, it has no peer for peering deeply into a little patch of sky. In 1995, Bob Williams, then Director of the Space Telescope Science Institute in Baltimore, promoted a project to stare at an otherwise blank and uninteresting piece of sky with the Hubble Space Telescope. He consulted widely to be confident that this "Hubble Deep Field" had broad community support. There is just one space telescope and, though the director is responsible for setting the scientific program, and can nominally do what he thinks best, in practice the "director's discretion" time is limited, and most of the space telescope's observing program for every year is decided in a bruising peer review by a Time Allocation Committee.

The previous Director, Riccardo Giacconi, had used his discretion to deal with scientific opportunities that cropped up between cycles, or to rectify injustices caused by the Byzantine rules of the time allocation process. But Bob Williams wanted to concentrate his Director's time on a single spot to drill the deepest well into the past that technology would allow. Personally, I thought it was a dumb idea.

In an expanding universe, the contrast of galaxies with the sky should fade out in proportion to (1 + z)4, so once you got beyond zof one, you were losing a factor of 24, which is 16, and you were running the risk of seeing nothing much. Why invest so much valuable telescope time on this potentially futile effort when you were turning down good proposals (including some of mine) every year?

Fortunately, Bob received many opinions, not just mine, and he went ahead. The Hubble Deep Field observations produced a gusher of information on the past—especially the past history of star formation in galaxies. The distant galaxies were not only visible against the night sky (because they were lumpy—-who knew?) but they were just within reach of tile Keck telescopes for spcctra, so the Hubble Deep Field is not just a knockout screensavcr, but a powerful window into the history of star formation in the universe at redshift 1, 2, 3, and beyond.

In 1996, Ron Gilliland of the Space Telescope Science Institute and Mark Phillips, from Ccrro Tololo, applied for time to revisit the Hubble Deep Field. A second look would allow them to detect things that had changed. A second look would allow them to find supernovae at high redshift. When Mark described what they had planned, I was lukewarm. A back-of-the-envelope calculation showed that their chances of finding anything were not very good, and even if they did find something, without an extensive follow-up program, they weren't going to learn very much. Without a light curve, they wouldn't know if the supernova was going up or down. Without a spectrum, they wouldn't know the supernova type or the galaxy redshift. And they were looking in the wrong place for light from a very distant supernova: a supernova beyond redshift 1.5 would have its emission shifted out into the infrared, l^eyond 1 micron, where the CCD detector on HST was completely indifferent to light. On the other hand, there wasn't much harm in trying, and they might get lucky. Fortunately, I wasn't on the Time Allocation Committee that year and they got the time.

In data from the repeat observation, which comprised 18 orbits exposed between 23 and 26 December 1997, scrupulous subtraction by Ron Gilliland showed there was a Christmas present for Ron and one for Mark. There were two definite dots. SN 1997ff and SN 1997fg. Ron and Mark, together with Peter Nugent, wrote up their discovery for The AstrophysicalJournal.

On one hand, I was right. They didn't have enough information to do much with this detection, and it didn't add much to the cosmo-logical story that was unfolding. On the other hand, they were right—they showed that HST can be used to find very faint new objects. And more than anyone knew at the time, we had all been very, very lucky because SN 1997ff was about to be observed again and again in just the right way to add to the story of the accelerating universe.

Earlier in 1997, when astronauts rode the Space Shuttle up to HST, they carried two new instruments—a much improved spectrograph called STIS, and an infrared camera called NICMOS. While having the space tclescopc in a low Earth orbit creates huge headaches for planning observations, it does make it possible to bring up new instruments for a telescope that had been designed in the 1970s. NICMOS was a little infrared array, something like a CCD, but with a light-detecting ability that extended out to 2.5 microns in the infrared, roughly five times the wavelength of visible light. Infrared emission comes from cool places and infrared light is not obscured by dust as much as visible light, so NICMOS was a powerful tool for probing the cool, dusty placcs where stars arc being born.

Compared to the CCDs on the Space Telescope, the NICMOS array is very modest—it has only 65,000 pixels, compared to 2.5 million for the visible-light camera. It covers a tiny patch of sky smaller than 1 arcminute on a side, while the CCD array covers an area H times larger with finer pixels. But it had one powerful new property—with sensitivity in the infrared, NICMOS looks at wavelengths where very distant galaxies and supernovac arc brightest. Observing from spacc gives sharp images, but even more importantly for infrared observations, there is much less emission from the sky. As a result, NICMOS can knock the socks off 10-meter telescopes on the ground at the job of measuring infrared light from distant stars and galaxies.

Ordinary stars in high redshift galaxies emit visible light that gets rcdshifted by cosmic expansion into the infrared. So it seemed to the NICMOS team like a good idea to follow up on the success of the Hubble Deep Field, which had been done in visible light, by using NICMOS to pound away for 100 orbits on a tiny patch to sec what would show up in the infrared. After a test exposure on 26 December 1997, the NICMOS team started their observations in earnest on 19 January 1998. Without the intervention of human intelligence, by pure good luck, SN 1997ff was in the corner of their small field of view, like a hummingbird in a family snapshot on the fourth of July.

In science, as in life, it is good to be lucky! To build up a deep field, the NICMOS team returned again and again to the same place, slowly accumulating more and more data to beat down the noise and to allow the faint galaxies to be seen. Over a period of 32 days, HST accumulated many exposures of the same place. And almost every one had an infrared image of SN 1997ff, building up the material for a beautiful light curve for this object. But nobody knew that. The data went into the STScl Archive, where they aged like a fine claret from Bordeaux, just as Bev Oke had placed his supernova spectra in his Caltech desk to ripen for the opportune moment.

Last year, I was visiting the Space Telescope Science Institute as a member of one of the myriad committees that provide sage advice to NASA on how to proceed. We were discussing the Next Generation Space Telescope. NASA had learned how to put a man on the Moon: by using checklists. Their confidence in the efficacy of paperwork has now been transmogrified into a worship of land-scape-format Powerpoint presentations. To escape from the blizzard of charts during a col fee break, I walked down the hall to see my onetime student Adam Riess.4

Things were going well for Adam. After getting his Ph.D. at Harvard, he had gone on to become a Miller Fellow at Berkeley. His thesis on SN la had won the Trumpler Award for the best recent Ph.D. He had married Nancy Schondorf. He was the first author of our paper on the accelerating universe. He now had a real tenure-track job at the Space Telescope Science Institute. He won Harvard's Bok Prize, awarded to one of our astronomy department graduates for outstanding work before the age of 35. Alex Filip-penko and I risked indictment for perjury in writing Adam incandescent recommendations for the American Astronomical Society's Warner Prize. He won that, too. His picture was in Time Magazine, pleasing his mother no end. The Warren, N.J., Echo-Sentinel ran a front page story under the headline, "Local Boy Does Well in Astrophysics." He was buying a house. And now, he had a really great result to show me.

"Don't tell anyone," he told me, carefully closing his office door. "I'm still working on this. Wait 'til you see!"

Inwardly, I was chuckling. Adam was the gossip, not mc.

While the meeting droned on down the hallway, Adam showed me the graphs and pictures that told the story. Given the quality of my advice to NASA over the years, my absence from the meeting may have been a net benefit to society. Adam had scoured the HST archive and dredged up the NICMOS data for SN 1997ff. The rule is that you have one year for proprietary use of your own data, but then everything becomes public. STScI had built an excellent archive and encouraged people to exploit it. The statute of limitations had run out for the NICMOS team. Anyone could do what Adam was doing. Mayl>e somebody else was.

"Don't tell anyone."

"Get on with it."

With the Hubble Deep Field itself as the "before" image and the discovery data from Gilliland and Phillips at visible wavelengths, the repeated NICMOS observations made a fantastic data set. Supernova 1997ff was a bigger, fatter dot on the infrared images than in the CCD data. Adam was working with a whole squad of competent people who knew the details of the Hubble Deep Field and of NICMOS, including Rodger 'iTiompson, leader of the NICMOS team. Adam was putting it all together into an amazing picture for SN 1997ff. They had a great light curve. They had color measurements. They had good data about the properties of the galaxy in which the supernova had gone off. It looked like an elliptical—the kind that has only SN la and little dust. The observations also yielded a red-shift for the galaxy estimated from its colors. Adam showed that you could, independently, get a redshift estimate for the supernova from its colors. The two methods agreed. The redshift was about 1.7. This was what we had all been dreaming about doing—and the data had already been gathered without any planning or filling out of forms.

The payoff was to see whether this supernova from deep in the past could tell us whether we had made a colossal mistake in drawing the inference of an accelerating universe. Adam got slower and slower in flipping through the figures. He was enjoying the suspense. Hvcn more, he was enjoying the turned tables, How many times had I revised his manuscripts and been the one holding the authority to sign off on a result, to approve his thesis? No more. A student had becomc a collcaguc, and he was relishing it.

Was the supernova dim, showing we had been fooled, or bright, pointing the finger at A? Adam kept the final chart facc down.

"If this thing turns out dim, Adam, you'll have to give back the Trumpler Award and the Bok Prize. Nancy will have to decide for herself what to do with you."

I was joking, but I was burning with impatiencc to see him flip over that ace in the hole.

"Look at this."

Supernova 1997ff was extra bright for its redshift, the way it should be if the universe was decelerating at first, then accclerating.

"Adam, this is really good "

"You are really, really lucky—the NICMOS guys could easily have choscn another spot to observe."

"Adam, this is really good and it is really important. The NASA press machinery will lap this up. But don't believe everything you read."

On 25 June 2001, Mike Lemonick at Time wrote up the story of SN 1997ff and, in a week without a terrestrial disaster, they ran it on the cover: "How the Universe Will End." Since they already had a file photo of Adam, they used it in a photo sequence his mother loved even more. It showed Einstein, then Hubble, then Zwicky, then Penzias and Wilson, then Adam Guy Riess, explorers of the cosmos. I told Adam they were arranged in decreasing order of importance but increasing lovability.

This result is too important to rest on just one object, but SN 1997ff points toward a universe that is genuinely a mixture of dark matter and dark energy. Further observations with HST will reveal more of those very distant supernovae and show more clearly whether we live in a universe that was slowing down before it began to acceleratc. That will be the smoking gun for A. But SN

1997ff is a test that the accelerating universe could have failed. And it did not fail that test.

Supernovae are the only direct evidence for acceleration, but shortly after the first supernova results in 1998, we began to combine the supernova data with observations of the ripples in the cosmic microwave background that can determine the cosmic geometry. Martin White, as a post doc at the Center for Particle Astrophysics at Berkeley, again when he was at Illinois, and later as a colleague on the Harvard faculty before he defected back to Berkeley, had repeatedly pointed out to me that when experiments measure the angular scale of the freckles on the microwave background, they measure the cosmic geometry. You could learn the total ii. When combined with the supernova observations, these measurements pin down how much dark energy and how much dark matter the universe contains.

Here's the way that works. The era when the universe was opaque ended about 300,000 years after the Big Bang. So the biggest scale of temperature variations that result from variations in the matter ought to have a size of something like 300,000 light-years. This is similar to the biggest variations in water level you can make in a bathtub—the longest waves are the size of the tub. You can try this at home. If anybody objects to the mess, you just say you are studying the formation of acoustic waves in the early universe. We view those ripples at a distance of 14 billion light years. Now, the angle covered by an object of known size (300,000 light-years) at a known distance (14 billion light-years) depends on the geometry of the universe. And Finstein tells us that matter and energy, the total S3, that is £2m plus £2A, determines the curvature of space. If the universe has the geometry of a sphere (Q greater than one) the freckles of the CMB will cover a larger angle than if the universe has the geometry of a saddle (£i less than one) or has the flat geometry predicted by inflation (£2 exactly equal to one).

The supernova data gave a value for - £iA, and the microwave background gives a value for Q.ni + iiA. Even a person whose theoretical credentials were fabricated in Photoshop could see that this would let you measure both fiN1 and £2A. If you know that the sum of the ages of Becky and Boh is 79 and you know that Bob's age minus Becky's is 26, that's enough to tell how old each one is. If you have measurements from supernovae and from the microwave background, you can learn how much dark matter and how much dark energy makes up the universe.

In 1998, measurements of the fluctuations in the microwave background were just beginning to provide credible measures of the total ft. Back in 1992, the COBE satellite had shown that there were fluctuations, but the COBE map of the sky was a blurry one, averaging over patches about 7' on a side. Small-scale freckles would be smoothed into patches of tan by those measurements. The recent balloons and ground-based systems were designed with the acuity to see variations on the scale of one degree or less.

Ihe experimenters were working like demons to get their measurements before results from the next satellite, Microwave Aniso-tropy Probe (MAP), dominated the field. MAP was specifically designed to make a fine angle map of the microwave emission. But as with any satellite, between the design and the launch, technology marches on. The detectors in MAP were conservative designs when they were thought out, over 7 years before the measurements. Agile balloonists, dwellers in high deserts, and Antarctic adventurers use more recent technical developments to detect the microwave signals. Though their observing sites have to contend with more interference from the Earth's atmosphere, if they are very clever and somewhat lucky, the little mice can sometimes do very well compared to a lumbering elephant of a space project.

In 1998, the situation was confusing, but hopeful. There were many measurements that suggested roughness in the cosmic background at an angular scale of about one degree. But some measurements disagreed with others and there was no single set of measurements that showed by itself that this signal was there. For outsiders to the field, like our team, it was hard to know how to use this information. Stepping up to the challenge, some energetic workers set themselves up as knowledge brokers, combining the data from various experiments to extract something reliable from the conflicting evidence. The supernova data and the CMB data were complementary. They defined two lines, perpendicular to one another.

the smoking gunj

Was this article helpful?

0 0

Post a comment