News this Week

Science  21 Feb 1997:
Vol. 275, Issue 5303, pp. 1064

You are currently viewing the .

View Full Text

Log in to view the full text

Log in through your institution

Log in through your institution

  1. Astrometry

    Hipparcos Charts the Heavens

    1. Andrew Watson

    The catalog of ultraprecise stellar positions mapped by this European satellite could change our understanding of the history of the universe and the lives of stars

    On 1 June, a new boxed set of six CDs goes on sale—one more disk than the complete Beethoven symphonies. At a price of $400, it might not seem like an impulse buy, but the world's astronomers are likely to see it as the bargain of the century. Contained on the disks is the most comprehensive star catalog ever created, the fruit of a European Space Agency (ESA) satellite called Hipparcos, which was launched in 1989 but is only now yielding results. Michael Perryman of ESA, who leads the Hipparcos team, describes the new list of star positions and brightnesses as “a giant three-dimensional map of the solar neighborhood.”

    Sharpening up the positions of a few hundred thousand unremarkable stars in our galaxy may sound like a workaday task, but the 100-fold leap in accuracy that the Hipparcos catalog represents is sending ripples throughout astrophysics. “The potential impact of Hipparcos is tremendous,” says Berkeley astronomer Ivan King. “It's going to allow you to do a lot of fundamental astrophysics, as well as astrometry,” adds Ken Johnston, director of the U.S. Naval Observatory in Washington, D.C. “I believe it is one of the most important projects for modern astrophysics,” says Malcolm Longair of Cambridge University.

    Scanning eye.

    Colors on this celestial map indicate the number of scans by Hipparcos's telescope, the key to its high accuracy.

    L. Lindegren and F. Mignard

    Investigators involved with the Hipparcos project have had access to its data since August last year. With a few months to go before general release of the catalog, Britain's Royal Astronomical Society held a meeting in London last week to give a taste of the early results. They did not disappoint. Among them is a recalculation of the distances to Cepheid variable stars, one of the standard yardsticks of astronomy, that ups the age of the universe and knocks down the ages of its oldest stars, perhaps easing an apparent contradiction that had puzzled cosmologists. Another headline-grabber for astronomers is a major revision of the Hertzsprung-Russell diagram. A deceptively simple plot of stars' brightness versus their temperature, the HR diagram maps out a kind of evolutionary tree for stars and has become a cornerstone of modern astrophysics. “The new Hertzsprung-Russell diagram is an incredible achievement,” says Longair.

    All this comes from two star catalogs, the product of 3 years of observing stellar positions and an even longer period of data processing (see sidebar). The more accurate of the two catalogs, called Hipparcos, pinpoints 120,000 stars with an accuracy to 1 milli-arc second, 100 times better than existing star catalogs. “You put a golf ball on the Empire State Building and view it from Europe—that's the kind of measurement accuracy we are talking about,” says Perryman, who is based at ESA's technical center in Noordwijk, the Netherlands. The second catalog, called Tycho, contains more than a million stars with a reduced accuracy of between 20 and 30 milli-arc seconds.

    Parallax view

    What makes these position measurements so important to astronomy is that they enable researchers to calculate stars' distances—and hence how big and bright they actually are. To calculate distances, astronomers use a method called parallax: If you plot the position of a nearby star and then wait 6 months until the Earth has moved around to the other side of the sun and plot it again, it will have moved slightly against the background of distant stars. “It's just like looking at a scene first with one eye and then the other. Nearby objects shift when you close one eye and open the other,” says Michael Feast of the University of Cape Town in South Africa. Nearer stars appear to move more than distant ones, so the size of the shift is an inverse measure of distance.

    Of the multitude of stars mapped out by Hipparcos, one group—regular winking stars known as Cepheid variables—is of special interest to Feast and his colleagues. Cepheid variables—the North Star is one—vary in brightness over a regular period that is directly related to the star's intrinsic brightness: The brighter the star, the longer the period.

    Hence, astronomers can use Cepheids to measure distances that are well beyond the range of parallax data. If you wanted to measure the distance of another galaxy, you could look for a Cepheid in that galaxy and measure its period. From the period, you could derive its true brightness; by comparing the true brightness with its apparent brightness as seen from Earth, you get the distance. The challenge is to pin down the brightness-period relationship in the first place, which requires measuring the distance to nearby Cepheids by some other method. “You need to know the distances of at least some Cepheids, so that [you] can calibrate the relation,” says Feast.

    The Hipparcos results have provided the first useful Cepheid parallaxes, which Feast and his colleagues used to recalibrate the Cepheid yardstick. The revised yardstick showed that the Large Magellanic Cloud, a nearby dwarf galaxy, “is 10% farther [away] than was previously assumed,” says Feast. This new measurement is vital to astronomers because the distance to the Large Magellanic Cloud is itself a basic yardstick of astronomy. Hubble Space Telescope (HST) scientists, for example, use this value to determine distances to more distant galaxies using Cepheids.

    Indeed, the revision may have a profound effect on cosmology. Estimates of the age of the universe from the HST have put it between 9 billion and 13 billion years old, but the oldest stars in the universe, those in the clumps of stars called globular clusters, are thought to be as much as 15 billion years old. Using the new Hipparcos data and revised Cepheid distance scale, Feast's team has derived a new distance scale for globular clusters in the Large Magellanic Cloud and the Andromeda galaxy. The estimated age of the clusters depends on the actual brightness of their stars, which has to be inferred from their apparent brightness and their distance. The distance “turns out to be a good deal greater than previously determined and reduces the age of globular clusters from about 15 billion years to about 11 billion years,” says Feast.

    At the same time, the revision of the distance to Cepheids in the Magellanic Cloud also leads to a 10% reduction in the value of the Hubble constant—the universe's expansion rate—as measured from more distant Cepheids. A lower expansion rate implies that the universe has had a longer time to slow down since the big bang and hence is older. Feast offers 13 billion years as his best estimate. “An increase of the expansion age and a decrease of the star ages bring things into much better agreement,” he argues.

    Some astronomers, however, say that it is too early to start redrawing length scales for the entire universe. “My personal feeling is to say wait a while before drawing those sorts of conclusions,” says Floor van Leeuwen of Britain's Royal Greenwich Observatory in Cambridge. “The chemical composition of the Magellanic Clouds is somewhat different from what you find in general in galaxies,” says van Leeuwen, which could affect the brightness-period relation of the Magellanic Cloud Cepheids. As a result, he says, “direct implementation of these calibrations involves some risks.”

    To support his argument, van Leeuwen cites Hipparcos data on the Pleiades, the familiar cluster of nearby stars known as the Seven Sisters. Measurements of the cluster “are not at all as expected,” says van Leeuwen. Based on the distances measured by Hipparcos, he says, “the stars of the Pleiades cluster appear to be fainter than the average star in the solar neighborhood of the same temperature.” The Pleiades were always assumed to be normal stars that fitted well into accepted theory of stellar structure. “The sort of difference we observe there should serve as a warning sign: Don't draw too far-reaching conclusions too quickly.”

    Starry eyes

    Astronomers who study stellar evolution are getting the same warning from Hipparcos, which promises to have a major impact on the Hertzsprung-Russell diagram. This plot of star temperature along the horizontal axis and luminosity marked vertically divides stars into different “species,” with cool red supergiants in the upper-right corner and the relatively dim but hot white dwarfs at the lower left. Separating these is the long, diagonal stripe of the “main sequence” stars, running from hot, bright ones at one end to dim, cool ones at the other. Our sun sits roughly in the middle of the main sequence. “[The HR diagram] is one of the keystones of astronomy for doing things like stellar evolution and looking at the life and death of stars,” says the Naval Observatory's Johnston. A revised HR diagram based on Hipparcos data will be released along with the catalogs in June.

    The precise positions in the Hipparcos catalog are also aiding studies of stellar evolution by enabling astronomers to correlate optical measurements with those made at other wavelengths. Johnston, for example, is merging two different kinds of data to study stars' surface layers, the source of the light that makes stars visible. “I'm interested in overlaying optical images and radio images, and one of the things Hipparcos will do is give me a very good optical reference frame,” he says. With optical positions accurate to a milli-arc second and a radio frame good to 100 micro-arc seconds, “I can then overlay my optical image with my radio image and attempt to match these things up,” he says.

    Patrizia Caraveo and Giovanni Bignami of Italy's Institute of Cosmic Physics in Milan are merging other kinds of data in an effort to understand the pulsar Geminga, the only neutron star known to emit gamma rays but no radio waves. Neither Hipparcos nor HST alone can give a good fix on Geminga's position: Hipparcos is too insensitive to its faint optical signal, and the HST has too small a field of view. But by combining data from both, together with results from ground-based telescopes, Caraveo and Bignami have pinpointed Geminga to 40 milli-arc seconds. “To be able to tell whether the single photons coming from the object are actually in phase with the rotation period, you need to know very accurately the position,” says Bignami. And once astronomers can clock how Geminga's spin is slowing down, they may gain insights into its mechanism, he explains.

    Dozens of other studies are benefiting from Hipparcos data, including an effort by Martin Barstow of the University of Leicester to determine the absolute sizes of white dwarfs, burned-out stars that are among the oldest objects in the galaxy. Because the size of the dwarfs affects how fast they cool, the measurements are critical for efforts to determine the age of our galaxy from the temperature of these stellar embers. Barstow's best estimate for the galactic age based on white-dwarf temperatures is now between 8 billion and 10 billion years.

    Astronomers readily concede that the Hipparcos data proper are short on glamour. But with the long-awaited results starting to flow, Hipparcos scientists hope the project will soon win the recognition it deserves. Like phone books, the Hipparcos catalogs may lack excitement, but they are essential if you want to dial up a whole universe.

  2. Astrometry

    A Mapmaker That Was Nearly Lost

    1. Andrew Watson

    Hipparcos—a tortured acronym for High-Precision Parallax Collecting Satellite—is living up to hopes that it would be the modern successor to Hipparchus, the ancient Greek astronomer who drew up the first accurate star map. But the European Space Agency (ESA) satellite was very nearly stillborn.

    After a perfect launch on 8 August 1989, an onboard motor failed to fire, and instead of finding itself in a stable geostationary orbit, Hipparcos wound up in a highly elliptical orbit—arcing up to 36,000 kilometers above Earth, then swooping down to 500 kilometers above the surface. Engineers quickly set about rewriting the satellite's control software and enlisted the help of additional ground stations around the globe to download data. “This was certainly the most significant setback we had to face,” says Michael Perryman of ESA, the project leader.

    Steady gaze.

    Hipparcos's 29-centimeter main mirror looks in two directions at once, allowing it to compare stellar movements in different parts of the sky.


    But there were more to come. The orbit plunged the satellite into Earth's radiation belts twice every 10.7-hour orbit, and collisions with particles took a toll on the solar panels. Eventually, in June 1993, Hipparcos fell silent. In that time, however, its wide-field telescope had collected 1000 gigabytes of data on stellar positions.

    Because of the way the data had been collected, it had to be processed as a single giant block—not a single star position could be determined until the whole process was finished. “This is the biggest data-handling problem ever undertaken in astronomy,” says Perryman, which explains the 3-year wait from the end of the mission to the first release of data last year. “All of that [data] had to be put together into one global astronomical jigsaw from which you extract the positions of each of the stars.”

  3. Biochemistry

    Src Structure Crystallizes 20 Years of Oncogene Research

    1. Carol Featherstone

    Just over 20 years ago, researchers discovered the first cancer-causing gene, or oncogene, in cells. In Nobel Prize-winning work, they realized that a normal cellular gene, when modified by a virus or mutations, can suddenly trigger wild cell growth—and cancer. The first such Jekyll-and-Hyde gene was called src for the sarcomas it caused in chickens, and the notion of “the enemy within,” as Nobel laureates Michael Bishop and Harold Varmus called it, was so powerful that almost every aspect of src has been under intense scrutiny ever since. For 2 decades, researchers in academe and at pharmaceutical companies have sought the controls for the Src protein's on-off switch. Now, the first descriptions of the crystal structure of this protein and one of its close relatives offer dramatic new evidence on how Src is regulated.

    In last week's issue of Nature, Wenqing Xu, Stephen Harrison, and Michael Eck of Children's Hospital and Harvard Medical School in Boston describe the structure of inactive Src to a resolution of 1.7 angstroms; in a companion paper, Frank Sicheri, Ismail Moarefi, and John Kuriyan at Rockefeller University in New York City report a similar structure for a related protein called Hck. Experts on Src and its relatives are thrilled by the structures. Biochemist Sara Courtneidge of Sugen Inc. in Redwood City, California, who has built her career around the protein, describes the Src structure as “very beautiful. … It's exciting and satisfying to see it finally.”

    Turned off.

    In the inactive form of the Src protein, the SH3 domain (yellow) binds to a helix (red) and helps to force part of the kinase's active site (dashed line) closed.

    Source: Xu et al.

    One reason the structure is so satisfying is because it supports previous research on the function of various parts of the protein. But it also reveals some surprising new twists, showing how in the benign form of Src, the active site of the protein is kept closed by the regulatory segments. The structure also paves the way for rational design of drugs that might keep Src and its relatives harmless, and thus be useful in cancer therapy. “The structures will be useful for refining inhibitors to make them potent and specific,” predicts another longtime Src researcher, Tony Hunter of the Salk Institute in La Jolla, California. And because both Src and Hck, a blood-cell protein, have quite similar structures, the results may be applicable to the whole Src family, which includes proteins involved in everything from immune responses to bone development.

    The Src protein is a tyrosine kinase, an enzyme that attaches phosphate groups to the amino acid tyrosine in certain proteins, thereby triggering a functional change in the recipients. For their structure, Xu, Harrison, and Eck worked with a large fragment containing 85% of Src's 536 amino acids, including four key substructures. These business ends of the molecule are the kinase domain that does the catalytic work, plus three regulatory regions: the so-called Src homology (SH) domains, SH2 and SH3, and the carboxyl tail of the protein.

    The interactions among these four domains had long been a source of speculation. The tail region, in particular, has proved critical to Src's oncogenic abilities, for the loss of a single tyrosine amino acid here can keep the protein active all the time and trigger uncontrolled cell growth.

    The new structures now reveal for the first time how these regions interact. Both proteins appear to be held in a benign, inactive state by a “belt and braces” architecture that keeps their catalytic sites locked up. Src's tail, with its crucial tyrosine, is the belt, bending back and winding around the bottom of the SH2 domain. That aspect of the structure was no surprise, as it had previously been predicted by retrovirologist Hidesaburo Hanafusa of Rockefeller University.

    But in an unexpected finding, the structure reveals that the SH2-carboxyl tail complex doesn't directly block the catalytic site—in defiance of “almost every model that was ever drawn for Src,” says biochemist Joan Brugge of ARIAD Pharmaceuticals in Cambridge, Massachusetts, who helped discover the Src protein. Instead, SH2 works with SH3 to form a brace that keeps the active site of the catalytic domain locked in a viselike grip. And in perhaps the biggest surprise, the SH3 domain plays a crucial role in preventing the catalytic site from functioning.

    The SH3 domain had been something of a mystery, because it was known to bind to a particular kind of helical protein structure called a polyproline helix, but no one had ever predicted such a helix in the Src protein. Now, both structures show that Src proteins do indeed have one such helix. The contortions of the SH2 domain and the tail force this helix against the catalytic domain, where the SH3 domain latches onto it. This deforms the catalytic domain and is responsible, perhaps more than anything else, for keeping the kinase inactive and harmless. So, contrary to all expectation, “the SH3 domain is closest to the point of action,” says Kuriyan.

    These interactions serve dual purposes, explains Eck. “This very remarkable arrangement serves not only to turn off the kinase, but also to sequester the SH2 and SH3 domains” so that they cannot bind to other molecules, he says. The bonds holding the SH2 and SH3 domains in place are relatively weak, however. So, in a normal cell, the two domains can be readily released by other proteins that bind to them more strongly, thereby allowing the Src kinase to be switched on. Thus, the structure allows precise and economic regulation that keeps the potentially dangerous protein on a short leash, says molecular biologist Giulio Superti-Furga of the European Molecular Biology Laboratory in Heidelberg, Germany.

    From these structures, it's possible to make an informed guess about what happens to make the oncogenic Src spin out of control, says Kuriyan. When the crucial tyrosine in the tail is missing, it no longer binds to the SH2 domain. As a result, the polyproline helix is not pushed near the catalytic domain, and the binding of SH3 does not interfere with the catalytic site—leaving the protein free to phosphorylate wildly.

    The next major goal is the structure of a Src protein with its regulatory domains in an active conformation, says Sugen's Courtneidge. That could help answer questions such as whether altering any one of the three regulatory interactions can cause the whole protein to pop open, or whether the kinase can be activated if only one interaction is disrupted. And the structure of an active Src might help to understand the regulation of many kinases with SH2 and SH3 domains, says Superti-Furga. For now, biochemists and molecular biologists can at last interpret their mutational studies in the light of a real structure, he says: “It's harvest time for us now.”

  4. Physiology

    A New View of How Leg Muscles Operate on the Run

    1. Elizabeth Pennisi

    You probably think of the muscles in your legs as motors, contracting and relaxing to drive legs up, forward, and down when you run. But new results reported on page 1113 by comparative physiologist Thomas Roberts of Northeastern University in Boston and his colleagues suggest that what actually propels you is a set of springs, consisting mainly of the tendons that attach the muscles to the bones. Often, the role of the muscles, they found, is to hold the ends of these springs rigid so that they can store energy as you land after each step.

    The findings, from an experiment in which the researchers directly monitored a leg muscle in turkeys as they ran on a treadmill, apply only to running on level ground. But they challenge the long-standing assumption that muscles always do actual work, shortening against resistance, to propel an animal forward—a view based on studies of isolated muscles or analyses of high-speed films of animals in motion. Instead, sensors implanted in the turkey leg showed that the muscle shortens slightly before the foot is planted and then simply exerts the force needed to keep the tendon stretched while the foot is on the ground, storing energy for the next stride. “The muscle is not doing any work,” says physiologist Robert Full from the University of California, Berkeley.

    Turkey trot.

    Monitoring the leg muscle reveals how turkeys—and perhaps other animals—run.

    Tom Roberts

    That finding helps explain an observation that has long puzzled muscle physiologists: Larger animals expend less energy per pound to propel themselves forward than small animals do. Physiologists tended to assume that the muscles and tendons in the larger animals simply use energy more economically as they do work. But the turkey studies indicate that the amount of work done has little to do with this difference.

    Instead, the answer lies in the fact that muscles that develop force more slowly require less energy, notes Peter Weyand, who did this work with Roberts at the Harvard lab of the late Richard Taylor: Larger animals take bigger steps and their feet stay on the ground longer, buying their leg muscles extra time to generate the necessary force on their tendons.

    For the current experiments, Roberts worked with turkeys because their tendons are better suited than other animals' to the technique he and his colleagues used for measuring the force generated by muscle contraction. To do this, they glued tiny strain gauges, which are commonly used to detect bending in bone or stressed steel, to each side of the tendon attaching the animals' gastrocnemius muscles (the equivalent of the human calf muscle) to the heel. The tendons of most animals are so elastic that the gauges would “pop right off” when a muscle pulled on them, says Roberts. But in the turkey, a large part of the gastrocnemius tendon is calcified and stiff. Consequently, the gauges stay mounted and can measure the stress exerted as the muscle contracts.

    At the same time that the researchers were measuring force, they also wanted to see how much the muscle was contracting as the turkeys ran. For this measurement, Richard Marsh from Northeastern sewed two piezoelectric crystals, one about 2 centimeters higher than the other, onto the gastrocnemius muscle. When the investigators sent a small electric current into one crystal to make it vibrate, the other crystal sensed the vibrations after a lag that depended on the distance separating the two. How much the muscle shortened as it contracted could then be determined from changes in the travel time.

    The results of these measurements were surprising. For one, the researchers observed that the relaxed muscle itself acts as a spring at times, lengthening as the leg swings forward, then recoiling as the foot heads toward the ground. While the turkey's foot is on the ground, the muscle generates force but shortens by just about 7% of its length. This small change translates into very little work output. Instead, “the force the muscles are generating in the stance phase were largely isometric,” comments biomechanicist Andrew Biewener of the University of Chicago. “They are simply acting as force generators.”

    And the strain gauges indicated that muscle exerts strong tension—100 newtons—on the tendon, particularly as the turkey plants its foot. The force exerted by the muscle enables the tendon and the sheath covering the muscle to stretch and store gravitational energy, just as a pogo stick stores energy each time its rider lands. The release of that energy is what propels the bird forward, Roberts says. “Most of the work is done passively in the elasticity [of the muscle and tendon],” Biewener adds.

    “At first glance, that's counterintuitive,” notes Thomas Daniel, a biomechanicist at the University of Washington, Seattle, because it implies that a running animal can get by on gravitational energy alone, without any additional energy input—as if a ball could bounce indefinitely. In fact, as Biewener points out, once a turkey or other running animal is in motion, it doesn't need to do much work to keep itself going. The small amount of shortening Roberts and his colleagues measured while the foot is planted is apparently enough to make up for friction and other inefficiencies.

    The situation changed, however, when the turkeys ran on an inclined treadmill. Their gastrocnemius muscles shortened more, working harder than they did on level ground. In addition, to produce the same force as it did on level ground, the amount of muscle used had to triple. “A lot more muscle is having to get recruited to get more force when the animal is [moving] uphill,” notes R. McNeill Alexander, a biomechanics specialist at the University of Leeds in the United Kingdom.

    Not everyone is certain that Roberts's experiments will be the last word in muscle function. Muscle physiologist Robert Gregor from the Georgia Institute of Technology worries that the piezoelectric crystals may not reliably estimate length changes. The speed at which the signal travels from one crystal to another may vary, depending on the properties of the tissue between the two crystals, he notes.

    Roberts, however, thinks the variability in signal travel time is too small to be a major concern, although he agrees that studies need to be done in other muscles to see if they work the same way. But to many biomechanics researchers, the familiar machinery of running already has a whole new look. Comments Daniel: “It uses an ingenious set of experimental methods, and the results are incontrovertible.”

  5. Molecular Evolution

    Did Birds Sail Through the K-T Extinction With Flying Colors?

    1. Ann Gibbons

    Even children's books can tell you that a giant asteroid or some such catastrophe killed off the dinosaurs 65 million years ago and cleared the way for birds and mammals to inherit the Earth. But paleontologists have long debated which other species—and how many—died out at the Cretaceous-Tertiary (K-T) boundary. The prevailing view has been that most archaic birds and mammals went the way of the dinosaurs, and that the few survivors rapidly hatched today's diverse array of birds and mammals (Science, 3 February 1995, p. 637). Now, a report on page 1109 combines molecular and fossil data to come to a different conclusion: that a diverse flock of birds flew through the cataclysm unscathed.

    At least 21 different avian lineages survived the devastation, according to molecular evolutionist Alan Cooper of Oxford University and theoretical biologist David Penny of Massey University in New Zealand, who used differences in the DNA of modern birds as a “molecular clock” to determine the age of each lineage. Although other molecular researchers raise questions about the analysis, they generally agree with the conclusions, which match other molecular studies. “This makes the big disaster at the K-T boundary a much less dramatic event,” says molecular evolutionist Svante Pääbo of the University of Munich in Germany. “It could change the textbook view that almost all the groups of birds died out.”

    Many paleontologists already think the fossil record shows that several groups of modern birds got an early start and made it through the extinction. But others point to fossil evidence suggesting that many other groups did suffer catastrophic extinction (Science, 22 November 1996, p. 1303). They also say the Cretaceous fossil record for modern birds is spotty, yielding little evidence for their presence before the K-T boundary. Ornithologist Alan Feduccia of the University of North Carolina, Chapel Hill, argues that the only truly modern birds who lived in the Cretaceous were a small group of transitional shorebirds. In his view, these were also the sole avian survivors of the extinction, and after the dust settled, they rapidly gave rise to other modern birds.

    Given the uncertainties of the fossil record, Cooper and Penny turned to the genes of living birds for clues about their ancestry. Theirs is the latest in a series of studies that depend on the simple premise that as two species evolve, their genes gradually accumulate different mutations. The more differences between two species, the more time has passed since they diverged from a common ancestor. By measuring the genetic differences between lineages, researchers can construct a family tree that shows the order in which species branched from a common ancestor.

    In the new work, Cooper and Penny looked at 16 orders of birds, using two small fragments of genes—600 base pairs of a protooncogene in the nucleus, and a 390-base pair region from the mitochondrion, a cellular organelle that has its own DNA. First, they measured the differences in the same snippet of gene between a pair of closely related birds, such as rheas and ostriches. They dated the split between these two using the earliest known fossil of each pair. For example, the oldest rhea dates to more than 60 million years ago, so the researchers reason that it took at least 60 million years to accumulate the 57 mutations that distinguish the rhea and ostrich genes. They analyzed another pair of closely related birds, such as loons and shearwaters, the same way, then averaged the mutation rates of the two pairs. Then they counted the additional mutations between the two pairs—38 in this case—and, using their average mutation rate of about one per million years, concluded that the last common ancestor of all four birds must have lived more than 98 million years ago, in the mid-Cretaceous.

    Using this so-called quartet method developed by Penny and a new family tree, the team found that many lineages diverged long before the K-T extinction. These results point to an Early Cretaceous origin for modern birds as a whole and suggests that at least 21 lineages, including parrots, wrens, and penguins, survived the extinction (see diagram). These findings support other molecular genetic studies, notably one of both birds and mammals by evolutionary biologist S. Blair Hedges and colleagues at Pennsylvania State University, which appeared in the 16 May 1996 issue of Nature.

    Ancient origins.

    Many bird groups may predate the extinction.

    Bentoo Penguin and Loon: VIREO; Emu: Adrienne T. Gibson/Animals, Animals; King Parrot: Patti Murray/Animals, Animals

    But because the molecular clock has to be calibrated with fossil dates, there is plenty of room for disagreement. Feduccia, for one, flatly dismisses the study as “a gross misrepresentation of the fossil record. The results are flawed because the molecular calibrations are based on erroneous identifications of the Cretaceous and early Paleocene fossils.” For example, he says that a Cretaceous fossil identified as a 70-million-year-old loon, used to date the loon-shearwater split, may not be a loon at all. And the dates used for the loon and other bird fossils, such as the duck, are too old, he says. But others have more faith in the data used: “The fossil record is accurate,” says vertebrate paleontologist Luis Chiappe of the American Museum of Natural History. “The dates are good.”

    Still, even some molecular systematists, including Hedges, Joel Cracraft of the American Museum of Natural History, and David Mindell of the University of Michigan, raise concerns about the study's methods and the calculated divergence dates for various birds, as well as the details of the family tree. Hedges says that the snippets of genes used are just too small to serve as reliable molecular clocks. And he'd like to see a statistical test to show that the divergence dates aren't being skewed by lineages with unusually fast or slow rates of evolution. Cooper and Penny say that they have done such statistical tests, and their work stands up. They add that they are already analyzing sequences of the entire mitochondrial genome to boost their data set.

    In fact, they are so sure of their method that they extended it. Using other researchers' sequence data and fossil records, they conclude in their paper that at least 100 terrestrial vertebrate lineages survived the boundary, including primates, rodents, and other mammals. What this means—to Cooper, at least—is that doomsday scenarios of the K-T extinction are wrong: It was a selective culling and may not have been catastrophic for all terrestrial creatures, he says. But given the weight of fossil evidence to the contrary, it may take more data—and more debate—before most scientists are persuaded that the K-T mass extinction scenario is for the birds.

  6. Fullerenes

    Trapped Buckyball Turns Up the Amp

    1. Alexander Hellemans

    In the dozen years since their discovery in 1985, the soccer-ball-shaped molecules of 60 or more carbon atoms now known as fullerenes have displayed a dazzling variety of tricks. Although real-world applications are still a way off, researchers have coaxed these “buckyballs” to become superconductors at low temperatures, emit light and carbon ion beams, and form many other compounds with different properties. Now, two European researchers have added something new to this list of talents: They have created an electromechanical amplifier from a single buckyball. “Nothing like this has been done before, and the experimental know-how to be able to do this is highly impressive,” says Daniel Colbert of the Center for Nanoscale Science and Technology at Rice University in Houston.

    The little squeeze.

    Compressing the C60 molecule with the tip amplifies the voltage in this circuit.


    This demonstration came about through a piece of serendipity. In 1995, Christophe Joachim of the CNRS Laboratory for the Study of Materials and Structures in Toulouse, France, and James Gimzewski at the IBM Research Laboratory near Zurich, Switzerland, tried to measure the electrical resistance of the basic fullerene molecule, C60, using a scanning tunneling microscope (STM). This instrument can map out details of a surface with atomic accuracy by passing a current from an ultrafine tip to the surface and detecting changes in the current when the tip scans over it. The researchers used the STM tip to hold down a single C60 molecule on a metallic surface so they could measure the current through it. But they noticed, to their surprise, that the apparent resistance of the molecule changed drastically when they squeezed and deformed it with the STM tip. “We found this funny,” says Joachim. “Why shouldn't we try to use it as an amplifier? we wondered.”

    So the team constructed a simple circuit based around the STM to demonstrate C60's ability to amplify an electrical signal. The key to the setup is the piezoelectric crystal that controls the distance of the STM tip from the surface. Such a crystal expands when a voltage is put across it. One loop of the circuit controls the voltage to the crystal, while a second loop passes a current down through the tip and the C60 molecule to the metal surface. Upping the voltage in the crystal circuit by 20 millivolts expands the crystal and moves the tip 1 angstrom (10−10 meters) closer to the surface, compressing the buckyball by about 15%. The resulting resistance change in the buckyball changes the voltage in its circuit by 100 millivolts. Hence, the input voltage to the crystal has been amplified by a factor of 5.

    The researchers are now investigating other molecules in search of the same properties. “Perhaps we can expect a similar effect when you deform a nanotube; to my knowledge, nobody has tried this yet,” says Gimzewski. They are also looking for other ways to compress the molecule. Possibilities include tiny actuators similar to the bimetallic strips used in thermostats, or molecules that deform in response to light, an effect known as photochromism. “The active element can consist of just one molecule,” says Gimzewski.

    Gimzewski's most recent work takes a step in this direction: He and his team have created a monolayer of bianthrone molecules on a copper substrate. These molecules change shape when irradiated by light. The team then attached C60 molecules to that layer and repositioned these molecules using an STM tip without destroying the monolayer—demonstrating for the first time that it is possible to move buckyballs around on top of a molecular monolayer. “Our research is now moving toward supramolecular systems, and this is one example of such a system—one type of molecule interacting with another type of molecule,” says Gimzewski.

    Practical applications are still far off, but “in a number of years, conventional microelectronics could run out of steam, and it is important to start now to look at possibilities from other directions,” says Gimzewski. Colbert agrees: “The import is as a demonstration of things to come. We are going to look back in 5 or 10 years and consider these things as important demonstrations of our beginning to play in this playground.”

  7. Electron Microscopes

    Electron Mirror Gives a Clearer View

    1. Erik Stokstad

    Optics researchers since Newton have known that any lens, no matter how well ground, suffers from flaws that will blur an image and add a rainbow fringe to its edges. Even the lenses in electron microscopes, which use beams of electrons instead of light to create images, can't escape these kinds of aberrations. Researchers long ago came up with corrective measures for light-focusing lenses. Now, a fix is in sight for electron microscopes as well, an Oregon team reports in the current issue of Microscopy and Microanalysis.

    The fix—a corrective mirror developed by physicist Gertrude Rempfer and her colleagues at Portland State University—could increase the resolution of some high-powered electron microscopes by a factor of 5 or more. That's potentially enough to distinguish receptors on a cell membrane. “It's a technological tour de force,” comments Jon Orloff, who works on electron optics at the University of Maryland.

    Looking sharp.

    An electron image of the same screen, with (bottom) and without correction.

    G. Rempfer

    It has been more than 250 years since Englishman Chester Moor Hall learned how to correct for the foibles of glass lenses. Glass brings different wavelengths of light, or colors, to focus at slightly different points, creating a blur of colors called chromatic aberration. Even light of just one wavelength won't focus perfectly, because rays passing through the edges of a lens bend too much, in what is known as spherical aberration. In 1732, Hall hit on the idea of passing the light through another lens that had the opposite defects and canceled out the aberrations.

    Rempfer and her colleagues looked for a similar solution to the aberrations introduced by the lens of an electron microscope, which consists of metal plates or magnetic or electric fields. Like a glass lens focusing light, electron lenses refract electrons of different energies by different amounts, causing the equivalent of chromatic aberration, and they suffer from a kind of spherical aberration as well. The Oregon team's solution is to send their beam of electrons into an electron “mirror,” consisting of a hyperbolic electric field that bends the paths of the electrons even as it repels them. Because higher energy electrons penetrate deeper into the electron mirror before they are reflected, the field has more time to bend their paths, introducing a pattern of aberrations that exactly cancels those introduced at the lens.

    “This is the answer to a problem that has plagued electron microscopy for the past half-century,” says University of Wisconsin physicist Brian Tonner. While it could ultimately benefit the two most common kinds of electron microscopes—scanning and transmission—its most immediate application will be for so-called photoelectron emission microscopes (PEEMs), says O. Hayes Griffith, a physical chemist at the University of Oregon, who collaborated with Rempfer.

    A PEEM bombards surfaces with intense visible or ultraviolet light to spur the emission of electrons, then focuses the electrons into an image. The emitted electrons come in a wide range of energies, however, resulting in strong chromatic aberration. One solution has been to restrict the wavelength of light, eliciting electrons at just a single energy—but also producing a dim image. By eliminating the aberrations with the mirror, “you could make a bright, quick picture without losing resolution,” says Martin Kordesch, a physicist and electron microscopist at Ohio University. The mirror could also increase the ultimate resolution of these microscopes from 70 angstroms to 10 or fewer, roughly the scale of atoms, says Griffith.

    Surface scientists, who study the composition and chemical behavior of surfaces, would be the most immediate beneficiaries. But the corrected optics could also aid biologists. For example, researchers examining receptors on a cell surface now have to label them with fluorescent markers. “That's like seeing the headlights of a car far away,” says Griffith. With a corrected PEEM, “you could see membranes right down to the cellular proteins, like looking at a landscape in daylight.”

    That's still in the future, though. The Oregon group has not yet attached their mirror to a working electron microscope. Meanwhile, Tonner's research group at the Wisconsin Synchrotron Radiation Center is trying to build a prototype corrected PEEM, and in Europe, a consortium of German groups has mounted a multimillion-dollar effort to build a corrected microscope. But in the race for higher resolution, says Tonner, Rempfer and her electron mirror have “made a quantum leap.”

  8. Mathematics

    How to Play Platonic Billiards

    1. Barry Cipra

    SAN DIEGO—Billiards is a game of geometry. Expert players delight in setting up incredible “trick” shots based on careful calculation of angles and distance. Now, Matthew Hudelson has added a new dimension to this pastime, as he reported here last month at the joint meetings of the American Mathematical Society and the Mathematical Association of America. With a little help from a computer, the Washington State University mathematician has demonstrated some amazing three-dimensional shots for a cue ball bouncing around inside three equal-sided, or Platonic, solids—the eight-sided octahedron, 12-sided dodecahedron, and the 20-sided icosahedron. Each of Hudelson's shots hits each side of the pertinent solid and returns to its exact starting point and direction of travel.

    Anyone for a game?

    Mathematicians have mastered return shots within five Platonic solids.

    Rhonda Roland Shearer; Photo by: Lee Boltin

    That's no trick on a 2D billiard table, provided it has a regular shape. Just start the ball at the midpoint of one side and aim for the midpoint of an adjacent side. But add a dimension and the problem gets far more interesting. When Hudelson took up the challenge, mathematicians had worked out the required trajectories for only the two simplest Platonic solids. Hugo Steinhaus gave the answer for the cube in the 1950s, and John H. Conway and Roger Hayward independently solved the tetrahedron in the early 1960s. Both cases were described in a 1963 Scientific American column by Martin Gardner, who says he is “quite impressed” with Hudelson's extension. Conway, now at Princeton University, agrees: “It's a very nice result.”

    Plumber's nightmare.

    Part of a chain of icosahedra.

    M. Hudelson

    In principle, solving the problem is just a matter of algebraic bookkeeping—tracking the equations of the straight lines the ball follows as it bounces around. A standard theoretical approach is to glue together a sequence of “reflected” copies of the shape or solid at the sides in the order in which you expect they will be hit. If there is indeed a trajectory that hits the sides in the prescribed order, then there will be a straight line that stays inside the glued-together construction, because each time the ball hits a wall, the mirror image of the ricochet is a straight line that continues the incoming shot.

    What makes things difficult is the sheer number of possibilities that have to be investigated, particularly for the dodecahedron and icosahedron. Even for the octahedron, there are hundreds of different ways to glue together eight copies of the shape, each corresponding to a different order in which the ball hits the walls. That's a lot of algebra.

    Hudelson took up the problem last summer after hearing it mentioned in a geometry seminar at Washington State. He started with the octahedron. First, he fashioned a likely arrangement of stuck-together octahedra made with cardboard and tape. The resulting “plumber's nightmare,” as Hudelson calls the tube, told him that there would be trajectories that hit the eight walls in the order he had guessed. It was then a relatively simple exercise in computer algebra to identify one path that returned the ball to its starting point and its original direction of travel.

    For the dodecahedron and icosahedron, Hudelson had the computer do all the work. To get started, he wrote a program that generated random initial trajectories and followed them for the first 12 or 20 bounces. Running the program 100,000 times for each solid, he got about 50 trajectories that hit all 12 sides in the dodecahedron and about five that hit all 20 sides in the icosahedron. Each of the successful trajectories hit the walls in the same order, which suggested that there is essentially only one solution to the problem for each solid. He then went on to identify the one trajectory that took the ball back to its starting point and direction.

    Hudelson doesn't see immediate applications for these virtuoso shots: “It just seemed like a hole that needed filling.” However, he notes that theoretical physicists, who use billiards on odd-shaped tables as a model of the behavior of atoms jumping chaotically between energy states, may turn out to be avid players (Science, 20 December 1996, p. 2014). If and when physicists make the leap from two to three dimensions, Hudelson's Platonic shots will be ready to show them a trick or two.

Stay Connected to Science