# News this Week

Science  03 Jan 2003:
Vol. 299, Issue 5603, pp. 26
1. INTELLECTUAL PROPERTY

# Universities Ask Supreme Court To Reverse Patent Ruling

1. David Malakoff

When Johns Hopkins University scientists met with the school's general counsel to discuss a recent court ruling involving patent rights, the lawyer offered them a disturbing real estate tip. “I told them to buy land in China,” recalls Estelle Fishbein. “Because if this decision stands, Asia is going to be the only place where you can do cutting-edge research without facing oppressive regulation.”

Fishbein's views might be extreme, but she isn't the only university official hitting the alarm button. This week, Johns Hopkins and other academic research powerhouses lined up to ask the Supreme Court to come to their aid. Their target is a federal court ruling on a bitter ownership fight over a laser developed by a university researcher. Critics say the October decision,* in Madey v. Duke University, effectively ends a 170-year-old practice of allowing scientists to freely borrow patented technologies for limited use in basic research that isn't aimed at commercial ventures. The universities are asking the high court to review—and ultimately overturn—the decision by a special patent court, because they believe it will hinder research by forcing scientists to obtain permission before using patented technologies.

“The decision transforms the academic science landscape in a horribly perverse way,” says David Korn of the Association of American Medical Colleges (AAMC) in Washington, D.C., one of the groups leading the charge. “It means that [government] research funds will be diverted into legal and administrative costs.”

Others, however, say the decision simply requires universities to follow the same patent rules as everyone else. And patent attorneys say it is in line with decades of decisions that have narrowed the so-called “research exemption” clause in patent law. Few of those cases have involved universities, however, and that could make the current battle attractive to the high court. “It's a sexy legal question: Why should universities benefit from the [exemption] when they might profit from the fruits of the research, too?,” says attorney Colin Sandercock, a patent specialist with the law firm of Heller Ehrman in Washington, D.C.

The current legal jousting stems from the invention of the free-electron laser (FEL) in the 1970s by physicist John Madey, then at Stanford University. Madey and Stanford won several patents on the device, which generates light that can be tuned to different frequencies, making it a potentially versatile tool for everything from physics research to surgery. After a falling out with Stanford administrators, Madey gained full control of the patents, and in 1988 he moved his Mark III laser into a custom-built laboratory at Duke, in Durham, North Carolina. A decade later, Duke officials removed Madey as head of the lab and he moved again, this time to the University of Hawaii, Manoa. Madey soon sued Duke over his removal and demanded the return of the equipment. His suit also accused the university of infringing on his laser patents by continuing to use the devices (Science, 21 November 1997, p. 1393).

A lower court sided with Duke, ruling in 1999 that the university wasn't infringing because its researchers were using the devices “for experimental, nonprofit purposes only.” That standard is rooted in an 1831 case. But a federal appeals court reversed the decision in October, noting that Duke is a businesslike entity that profited from the use of the lasers. The research “unmistakably further[ed Duke's] legitimate business objectives, including educating and enlightening students and faculty” and helped it “lure lucrative research grants,” wrote Federal Circuit Court of Appeals Judge Arthur Gajarsa.

That language outraged many university research advocates because it implies that the research exemption doesn't apply in an academic setting. “To categorize a research university, with its educational mission, as just another commercial operation borders on ludicrous,” says Sheldon Steinbach, general counsel of the American Council on Education (ACE) in Washington, D.C. It will be “disastrous,” he says, if researchers have to stop and conduct expensive, time-consuming patent searches and make licensing deals every time they want to bring a new technology or technique into the lab.

It also will be difficult for administrators to keep track of which researchers are using patented material, adds James Severson, the new provost for intellectual property at the University of Washington, Seattle. “Academic scientists often don't know, and don't even think about, whether something is protected by a patent,” he says. But the cost of not paying attention could be high, experts say, since alleged infringers could face triple-damages lawsuits.

Madey and some patent attorneys say that the threat of financial punishment is needed in a world where universities increasingly profit from their own patent portfolios—and sue infringers. The decision also follows legal precedent, they add. “What the court said isn't surprising to most businesses, but I guess it's seen as unusual because the case [involved] a university,” says Madey's attorney, Randall Roden of Tharrington Smith in Raleigh, North Carolina. It's been 70 years since a university was involved in a similar, potentially precedent-setting case, other attorneys note.

Duke officials hope the high court will rewrite that case law. The appellate court decision is wrong “on the merits,” they say. AAMC, ACE, and other university groups say they'll support Duke on a request filed this week by the university attorneys, Fulbright & Jaworski in New York City.

If the high court decides to take the case, it likely will be heard sometime after October 2003. If the court declines, the case will go back to a lower court, where Duke might still prevail on other grounds. And some observers predict that Congress may want to have the final word on the right balance between patent holders and the needs of academic researchers. That is, if all the scientists haven't moved to China.

• *Madey v. Duke University, No. 01-1567, Federal Circuit Court of Appeals, 3 October 2002.

2. U.S. CONGRESS

# New Senate Leader no Stranger to Science Policy

1. David Malakoff

A surgeon well versed in the politics of science will be the next majority leader of the United States Senate. Senator Bill Frist (R-TN) was elected to the powerful post last week after Senator Trent Lott (R-MS) announced that he would not seek another stint in the job. Lott resigned after receiving heavy criticism for remarks that appeared to endorse past segregationist policies.

Frist, 50, has served in the Senate for 8 years. A member of one of Tennessee's most prominent families, Frist worked as a heart surgeon and conducted biomedical research. In the Senate, he has become one of the body's leading experts on health care, bioterrorism, AIDS, and cloning. He led the Senate's Science and Technology Caucus and helped write provisions promoting science and math education in a recent bill that calls for doubling the budget of the National Science Foundation (NSF) (Science, 22 November 2002, p. 1537). He has also supported hefty budget increases for the National Institutes of Health and other science agencies. As majority leader, he will play a key role in setting the Senate's priorities and schedule.

Science advocates are generally upbeat about the switch. “He understands science and knows that the agencies that support research are important,” says Sam Rankin, a lobbyist for the American Mathematical Society and head of a coalition that pushes for greater NSF funding.

But some biomedical groups fear that Frist's arrival will strengthen efforts to ban research involving the cloning of human embryos. Earlier this year, he backed legislation that would have banned both cloning to create babies and cloning to produce cells for research and, possibly, medical treatments. Most researchers supported a competing bill that would have outlawed human cloning but allowed regulated research to continue. The issue is still unresolved, and Frist's record suggests that the more restrictive bill could now gain the upper hand.

3. ANIMAL BEHAVIOR

# Orangutans, Like Chimps, Heed the Cultural Call of the Collective

1. Gretchen Vogel

As evening falls in the Kinabatangan forest of Borneo, a careful listener can sometimes hear a loud spluttering sound, a sort of cross between a hoot and a sigh. The call signals that a local orangutan is bedding down for the night. The practice seems perfectly normal in Kinabatangan: Almost every orangutan in the region calls in the same way. But elsewhere on the island, in the Kutai forest, orangutans make their nests without making a ruckus.

The difference is a sign that orangutan groups have at least some hallmarks of what in humans is commonly called culture, says primatologist Carel van Schaik of Duke University in Durham, North Carolina. On page 102, he and his colleagues describe two dozen behaviors that are present in some orangutan groups and absent in others. The practices are apparently learned from other group members and passed from generation to generation. Such observations give biologists richer insights into animal behavior, others say, and they might help researchers understand how human culture evolves.

For many years, culture was thought to be unique to the human species, but evidence has been growing for socially learned traditions elsewhere in the animal kingdom. Behavioral studies have found some signs of social learning in birds, rats, capuchin monkeys, and even fish. Meanwhile, a study suggesting that whales exhibit culturally determined behaviors was met with skepticism (Science, 27 November 1998, p. 1616). But the best evidence until now for nonhuman culture came from chimpanzees: By pooling data from nine long-term sites in different regions of Africa, researchers documented 39 examples of behaviors that were specific to particular groups and did not seem to be determined by the environment (Science, 25 June 1999, p. 2070).

Van Schaik suspected that a similar pattern might be present in Asia's great apes, the orangutans. In the Suaq Balimbing forest in Sumatra, for instance, he and his colleagues had observed these animals using sticks to extract seeds from the Neesia fruit. In Borneo, however, other researchers never see such handiwork, even though Neesia is readily available. Curious to see if the orangutan research community could come up with a list of behaviors similar to that compiled by chimpanzee researchers, van Schaik invited his colleagues to a 3-day meeting to compare notes.

Even van Schaik was surprised by the results. With many of the commonly observed behaviors, he says, “you just assume every [orangutan] does it the same way everywhere.” But as the researchers compared notes—and videotapes when possible—it became clear that many behaviors were strikingly different between orangutan groups.

The list of probable cultural traits is not as long as that for chimpanzees. But Bennett Galef of McMaster University in Hamilton, Ontario, says that nevertheless, the evidence from the orangutan watchers is stronger in some ways. “They were able to document two behaviors that are present in every member of one group and [in] no member of another,” he says, strengthening the case that individuals learn behaviors from the group rather than discover them randomly on their own. Furthermore, orangutans' tendency to interact with their neighbors less than chimps do made the pattern of learning even clearer. The researchers found that groups of more sociable orangutans had larger behavioral repertoires than groups of relatively solitary individuals had, supporting the theory that social contact spreads cultural behaviors.

The observations might help researchers learn more about the roots of human culture by clarifying what makes it distinctive. One critical difference, many researchers note, is that animal groups do not appear to improve upon a previous invention, although humans have been doing so for millions of years. Humans might excel at tinkering because they are great imitators, says psychologist Michael Tomasello of the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany. He and others have shown that human children can imitate a demonstrated skill more readily than captive apes can, an ability that might allow children to acquire complex behaviors more easily from other group members.

Tomasello cautions that the orangutan study is only a beginning. Researchers “talking about their impressions is an excellent way to generate hypotheses, but it's only step one,” he says. Van Schaik agrees that he and his colleagues now need to return to the field with the full list of orangutan behaviors to verify that they're present in some groups and not in others.

Chimp researchers, meanwhile, are meeting this month in Leipzig, Germany, to refine their list of possible cultural behaviors. The meeting will include representatives from three additional study sites not included in the first survey.

Chimp and orangutan researchers have little time to waste. Both species are gravely threatened by habitat loss and poaching. Illegal logging and civil unrest have taken a toll: One of the sites van Schaik canvassed for the current study is now “pretty much gone,” he says, with its unique cultural traditions wiped out as well.

4. JAPAN

# Science Fares Well in a Tight Budget

1. Dennis Normile

TOKYO—Japan's scientific community last week got a pleasant surprise. Despite belt-tightening that will hold total growth in government spending to just 0.7% in the fiscal year beginning on 1 April, the administration's budget includes a 3.9% rise, to $10.3 billion, for science. And a separate economic stimulus package, which will be appropriated during the last 3 months of the current fiscal year, will provide a hefty dollop of funds to upgrade research facilities. The Cabinet approved both budgets on 24 December, and the legislature is expected to endorse them in the next few weeks. “Compared to other fields, [science] has been treated very favorably,” says Toichi Sakata, deputy director-general for research promotion at the Ministry of Education, Culture, Sports, Science and Technology. The few grumbles within the community focus on worries that the spending is emphasizing quick results at the expense of long-term scientific health. Particularly notable jumps are in store for competitive grant programs. Funding for peer-reviewed grants for scientific research, the largest source of support for individual academic researchers and small groups, will rise 3.6% to$1.5 billion. Several large-scale physics projects are also getting significant increases. Funding for a high-intensity proton accelerator being built in Tokai, northeast of Tokyo, will jump 54% to $41 million to support a ramp-up in construction. And support for neutrino studies—set to grow by 23% to$18.8 million—includes money to begin fully restoring the Super-Kamiokande neutrino observatory. An accident late last year destroyed two-thirds of the facility's 11,000 photomultiplier tubes (Science, 11 January 2002, p. 247). The facility recently restarted observations using the remaining sensors. Super-Kamiokande project director Yoichiro Suzuki says that “we benefited from a tailwind,” in the form of publicity surrounding the award of a share of last year's Nobel Prize in physics to Masatoshi Koshiba for his work on neutrinos using the original Kamiokande observatory.

The supplementary budget, intended to stimulate the economy primarily through public works spending, will also benefit science. The education ministry's share of that pie includes $1.8 billion for upgrading university research equipment and facilities. Indeed, several programs facing reductions in the ordinary budget, including space science and earthquake research, managed to cover the cuts with funds from the supplementary budget. One area that was not so fortunate is nuclear power, where the budget will drop 3.4% next year, to$2.6 billion. A spokesperson for the education ministry's nuclear power research division says most of the cuts will be covered by efficiencies resulting from merging two major research labs. But he admits there will be a yet-to-be-defined impact on research.

Finance Minister Masajuro Shiokawa emphasizes that the government gave priority to science and technology, among other fields, because it is expected to help revitalize the economy. Those expectations make some researchers nervous. One institute head, who asked to remain anonymous, says there is “too much of an emphasis on short-term results rather than long-term benefits.”

5. HUMAN SUBJECTS

# Crossing a Frontier: Research on the Dead

1. Jennifer Couzin

In what appears to be a first, a prominent university medical school has crafted a new policy for self-regulation in a rarely discussed research area: studies on the dead. The University of Pittsburgh School of Medicine and its hospital, well known for their work in the organ transplant field, have approved two studies and are considering a third on individuals who have been declared brain dead but whose cardiopulmonary systems continue to function, thanks to life-support equipment. Pitt has established a special ethics committee to oversee such research.

Research on brain-dead patients and cadavers, which also fall under the Pittsburgh policy, is legal but unregulated. Many medical schools have guidelines that apply to cadavers, but not to those who are brain dead. Pitt chose to draft a more extensive policy after the panel that normally screens clinical trials, the institutional review board (IRB), declined to consider a trial involving brain-dead subjects, according to ethicist and critical-care specialist Michael DeVita. The IRB said its role was limited to live patients. The investigators in the Pittsburgh case, which DeVita declines to describe, were uncomfortable going forward without ethical review and contacted DeVita for help. He and several colleagues drafted the new rules and helped establish the review panel, which gained momentum following a separate Pitt study that removed hearts unsuitable for donation and kept them beating outside the body.

Ethicists and physicians are far from unanimous on the acceptability of such experiments. On the one hand, they agree that studies on the brain dead could deliver valuable information that might not be obtainable in clinical research without grave risks. On the other, they recognize that the experiments raise ethical questions, can be logistically complex, and may be emotionally difficult for families and even medical staff.

“You need a policy to make the research happen, and I think the research should happen,” says Arthur Caplan, a prominent bioethicist at the University of Pennsylvania in Philadelphia, who has pressed for years to offer dying patients and their families the option of participating in research after death. However, he adds, “most institutions are nervous or, more accurately, horrified … by the subject. … People worry about a public relations disaster.”

The Pittsburgh policy requires informed consent from the patient before death or from a surrogate, normally the next of kin, after death. It also stipulates that experiments not jeopardize organ or tissue procurement and that protocols on dead subjects be evaluated for their science and their ethics.

To its creators' surprise, the new Committee for Oversight of Research Involving the Dead has turned out to be far busier than expected, says Clifford Schold, the committee's chair and assistant vice chancellor for clinical research at the University of Pittsburgh's Schools of the Health Sciences. Since late September, it has evaluated nine submissions, including the three involving brain-dead subjects. The committee's members—numbering fewer than a dozen, including a transplant surgeon, a philosopher, and a nun—have approved some and sent others back to the researchers for revisions.

Pitt is hardly the first to consider or pursue research on brain-dead subjects. Officials at AbioMed, the Danvers, Massachusetts, company that created an artificial heart, debated first testing the device on individuals who were brain dead because it was deemed so novel and risky. But “people were frankly somewhat mixed and ambivalent in their comfort level,” says Edward Berger, AbioMed's vice president for strategic planning and policy, and the company rejected the idea. Caplan notes that the first artificial heart, the Jarvik-7, was quietly tested on individuals who were brain dead.

Another institution currently supports two trials on brain dead and newly dead subjects—the M. D. Anderson Cancer Center in Houston, Texas (Science, 15 February 2002, p. 1210). But, unlike at Pitt, the IRB reviewed and approved both trials. A top M. D. Anderson official can't understand why Pitt didn't do the same: “It's no time for the IRB [at Pittsburgh] to get cold feet—they've got to step up to the plate,” says Leonard Zwelling, M. D. Anderson's vice president for research administration at the hospital. Philip Troen, the chair of the University of Pittsburgh IRB, says his group simply “made a decision that jurisdictionally it wasn't appropriate.” He declined to comment further.

Pitt officials are confident that they made the right decision in establishing a mechanism to oversee this research. DeVita and Schold anticipate more studies on this group of subjects in the coming months and years, particularly in the drug metabolism and organ transplant fields. “The obvious merits of performing experiments in dead individuals are going to make this a far more common endeavor,” says DeVita. “[It] requires some oversight.”

6. CHINA

# With Help From Overseas Chinese, a Western-Style Institute Takes Shape

1. Jeffrey Mervis
1. Yang Jianxiang*
1. Yang Jianxiang writes for China Features in Beijing.

BEIJING—Its new home outside Beijing is almost finished. But not a single researcher has been hired, and negotiations over who will be calling the shots are still under way. Welcome to China's latest attempt to create a world-class life sciences institute.

Sitting somewhere between an idea and reality, the National Institute of Biological Sciences (NIBS) already has a compelling history. Two years ago, Premier Zhu Rongji, on a visit to Singapore, heard about an idea being put forward by a group of Chinese-born scientists that China should create a biomedical research institute run along Western lines. Its director would set the research agenda, hiring senior staff and allocating resources, and a science-savvy board of trustees would look after the institute's long-term health, raise money, and be a bridge to the authorities. That management structure would set it apart from the typical Chinese institute, which is kept on a short leash by administrators who often have little scientific training. “There is nothing like this in China,” says Cornell University plant biologist Ray Wu, an adviser to the project. The premier liked the idea and asked the Ministry of Science and Technology (MOST) to work out the details.

MOST, which was already planning an institute for agricultural biotechnology, agreed to broaden its scope to encompass the health sciences and to invest $25 million over 5 years in operating costs. Beijing municipal authorities volunteered to host the institute, putting up almost$50 million for land and a well-appointed research laboratory at the new Zhongguancun science park north of the city. “It's as attractive and well-equipped as any U.S. institute,” says Wu about the five-story, 18,580-square-meter facility, complete with an underground garage, that is expected to be finished this winter. The State Development Planning Commission is also chipping in \$12.5 million for the project.

With construction under way, discussion turned to how the new institute would be run. The Singapore group, facing uncertain job prospects there because of the merger of two institutes, saw it as a way to return home under more favorable conditions. But the group bowed out after negotiations with MOST broke down earlier this year. “All we can say is that we have made our contributions and are very proud of ourselves,” says Li Peng, a principal investigator at the National University of Singapore's Institute of Molecular and Cell Biology. “We wish the best for the new institute.” Their departure cleared the way for Wu and other prominent overseas Chinese scientists to make their case for a truly independent institute, with leaders chosen by a blue-ribbon panel of scientists rather than by career bureaucrats and given the authority to set the institute's course.

“The Chinese system likes to keep control, but that's not healthy,” says Wu. “We're still negotiating the charter, and I think that they are slowly coming around to the idea.” Chinese officials have already agreed to add three scientists to the 11-person oversight board, he notes, adding expertise to a body that otherwise consists of representatives from each of the ministries involved in the project, including health, agriculture, and education. “We all think that the scientists should be in charge of the science.”

Although the structure of the institute hasn't yet been settled, Chinese officials are running classified ads in scientific journals (including Science) for a director, deputy director, and 25 principal investigators and senior researchers, at salaries pegged halfway between local and Western levels (http://www.cncbd.org.cn/). An 18-member panel of Nobelists and overseas and domestic Chinese scientists has been assembled to review the candidates, according to MOST officials, who requested anonymity.

“I think that the government is ready for a Western-style institute,” says Wu, noting that he has helped set up similar institutions in Taiwan and India. “But it will take a lot of work.”

7. GERMAN SCIENCE

# Stem Cell Project Wins Final Approval

1. Sabine Steghaus-Kovac*
1. Sabine Steghaus-Kovac is based in Frankfurt.

FRANKFURT, GERMANY—Neuropathologist Oliver Brüstle received an early Christmas present on 23 December: His application to work on human embryonic stem (ES) cells finally cleared the last legal hurdle. Brüstle, who works at the University of Bonn Medical Center, is the first German researcher to navigate a complex approval process for research on ES cells. It took more than 2 years.

The derivation of stem cells from human embryos is banned in Germany by the Embryo Protection Act. But in January 2002, after an intense debate, the Bundestag passed a law allowing researchers to import ES cells in exceptional cases. They must prove that the work is “outstanding” and that it cannot be done with alternatives such as human adult stem cells or animal cells, and they can import only ES cells derived before January 2002 from surplus embryos created for in vitro fertilization. The law gives the Robert Koch Institute (RKI) in Berlin the authority to determine that those conditions are met.

Brüstle had applied in 2000 for funding from the Deutsche Forschungsgemeinschaft (DFG), Germany's main research agency, to extend his work on mouse stem cells to human ES cells. The ultimate goal is to develop therapies for neurodegenerative diseases such as Parkinson's, Huntington's disease, and myelin disorders such as multiple sclerosis. DFG postponed a decision three times to allow a broad political discussion of the topic before announcing last January that it would fund Brüstle's project if it meets the conditions laid out in the new law. Last week, RKI gave its stamp of approval.

Brüstle will import 12 cell lines from the Rambam Medical Center in Haifa, Israel. RKI's decision, says Brüstle, sends an important signal: “Researchers in Germany now have the opportunity to develop this promising field together with their colleagues inside and outside Europe.” Three other German groups have also applied for special authorization by RKI.

8. SCIENTIFIC MISCONDUCT

# More of Bell Labs Physicist's Papers Retracted

1. Robert F. Service

In 2001, Jan Hendrik Schön, the former physics prodigy at Bell Laboratories in Murray Hill, New Jersey, cranked out papers at the astounding average rate of one every 8 days. Now, in the wake of a 25 September Bell Labs report that concluded Schön had committed widespread misconduct (Science, 4 October 2002, p. 30), the retractions are coming almost as fast. In November, Science published retractions of eight papers by Schön and colleagues. Nature posted linked warnings to its electronic versions of Schön's papers and said it would soon issue retractions in print. In the latest round of backpedaling, officials at the American Physical Society (APS) and the American Institute of Physics (AIP) announced last month that they were issuing retractions for 12 papers Schön and co-authors had published in their journals.

According to AIP journals publisher Thomas von Foerster, all of the authors agreed to retract six papers in the AIP journals: five in Applied Physics Letters and one in the Journal of Applied Physics. Martin Blume, editor-in-chief of APS's journals, says that all authors agreed to retract two articles in Physical Review B (PRB) that the Bell Labs committee had red-flagged as likely to contain “substituted data.” But for four papers in PRB and Physical Review Letters that the committee did not review, Schön did not agree to his co-authors' decision to retract. The text of the retraction will reflect that disagreement, Blume says. APS also plans to post notices alerting readers to two controversial papers that are not being retracted because the authors continue to stand behind them. In one of those papers, Schön is the only author.

Blume says APS editors felt it was important to review all the papers that Schön co-authored since 1998, not just those flagged by the Bell Labs report. The journals will keep the retracted papers on their Web sites but will link retractions to the online papers and publish errata in the printed journals. “We do not want to tamper with the archive of published papers,” says Blume. “It will say in red, ‘retraction.’ It's like a scarlet letter.”

9. DEVELOPMENTAL BIOLOGY

# Newborn Neurons Search for Meaning

1. Marcia Barinaga

Some brain regions are replenished with new cells throughout life, but researchers aren't sure what the newcomers contribute to the behaviors those brain regions control, which include singing, smelling, and learning

A neuron is a precious thing. Indeed, until recently, brain neurons were thought to be irreplaceable. Adult brains, we were taught, are unable to make new neurons, so those with which we are endowed at birth must last a lifetime. Major cracks in that dogma began to appear in the 1980s with the discovery that some songbirds—warm-blooded vertebrates like ourselves—give birth to waves of new brain neurons seasonally. Then researchers observed the birth of new neurons—a process called neurogenesis—in the brains of adult mammals, and the old view came crashing down.

Interest in newborn neurons surged, and researchers quickly discovered conditions that turn the birthrate of new neurons up or down. Stress and depression, for example, suppress neurogenesis, whereas exercise, stimulating environments, and antidepressant drugs all give it a boost. A paper in this issue adds pregnancy to the list of neurogenesis-enhancing conditions. But a key piece of the puzzle remains missing, says Larry Squire, a neuroscientist at the University of California (UC), San Diego: “We really don't know anything at all yet about what the function of these new neurons might be.”

That fact can get lost in the hubbub surrounding the field. The notion of pools of new neurons repopulating brain areas is “so seductive,” says Sarah Bottjer, a neuroscientist at the University of Southern California in Los Angeles, “that people get sucked in and lose their perspective.” To many observers, it seems that boosting neurogenesis simply must increase brain power, and that idea has been translated into notions that exercise, stimulating environments, or even sex will improve one's memory or ability to learn. “A lot of what is being said is more speculation than fact,” cautions Tracey Shors, a neuroscientist at Rutgers University in Piscataway, New Jersey. “It is too simplistic to think that if you have more cells you are going to be smarter.”

## The dogma dissolves

The first challenge to the “no new neurons” dogma came in 1965. After treating rodents with markers for dividing cells, Joseph Altman and Gopal Das of the Massachusetts Institute of Technology saw cells that appeared to be newly born neurons. But the techniques available at the time couldn't rule out that the cells were glia, support cells in the brain that are known to be replenished, and the results were largely disregarded.

Nearly 20 years later, Fernando Nottebohm and then-graduate student Steven Goldman at Rockefeller University in New York City found evidence of newborn neurons in canaries' high vocal center (HVC), a brain area that helps produce their song. Postdoc John Paton found that the young neurons integrate into the circuitry of the HVC and fire in response to sound. “That gave credibility to the idea that in warm-blooded vertebrates, you could have ongoing neurogenesis in the adult brain,” says Eliot Brenowitz, a neuroscientist at the University of Washington, Seattle. Neurons were continually added to the HVC, but the bulk appeared during times when male canaries learn new song elements. That led Nottebohm to propose that the new neurons were important for song learning.

Nottebohm's team then discovered neurogenesis in the hippocampus, a brain area critical for spatial learning, in black-capped chickadees. These birds hide seeds in the fall to retrieve later. Postdoc Anat Barnea trapped adult birds, injected them with a label for newborn cells, banded their legs, and released them. She recaptured the birds 6 weeks later and looked for new neurons in the hippocampus. She found new neurons throughout the year, but their numbers peaked when the birds were storing seeds, suggesting that neurogenesis might aid the birds' memory during that critical time.

Once it was established that neurons were born in the brains of adult birds, some researchers argued that birds were a special case. Their need to be light and energy-efficient might have driven them to evolve a system to lose neurons when they don't need them and regrow them when they do.

Then researchers found neurogenesis in the brains of adult mammals. In 1992, Sam Weiss and Brent Reynolds of the University of Calgary in Alberta, Canada, isolated neural stem cells—which can differentiate into neurons—from the subventricular zone of mouse brains. And in 1994, Arturo Alvarez-Buylla and Carlos Lois, then at Rockefeller University, showed that in adult brains these cells form neurons that travel to the olfactory bulb, the brain area that receives sensory information from the nose. Alvarez-Buylla's group, now at UC San Francisco, went on to learn that the new neurons connect to the established circuitry in the olfactory bulb, and Jonas Frisén's team at the Karolinska Institute in Sweden found that they are activated when the bulb responds to smells.

Meanwhile, Fred Gage and his colleagues at the Salk Institute in La Jolla, California, showed that the mammalian hippocampus—as in birds, an important site for memory—also sprouts new cells. His team found neural stem cells in the adult rat hippocampus and subsequently showed that new neurons are born in the hippocampus of adult mice and humans.

Despite all this progress, and some encouraging clues and circumstantial evidence about the role of neurogenesis, researchers studying both birds and mammals are still working furiously to answer the same question: What do these new neurons do?

## Neurogenesis on the wing

“The avian system is a little further along” toward an answer, says Alvarez-Buylla, because researchers can work with a well-characterized behavior that varies between species. Male canaries sing elaborate songs during the breeding season and learn new song elements every year. During the nonbreeding season they sing less, their song gets less precise, and their HVC shrinks in size. But when the time comes to embellish their songs again, the HVC enlarges through the addition of new neurons. Zebra finches, on the other hand, learn one song during adolescence and never change it. They also differ from canaries in that they sing and court mates throughout the year. Their brains reflect this difference: Zebra finches add large numbers of neurons to their HVCs only when they are young.

The correlations between neurogenesis and song learning inspired Washington's Brenowitz and then-grad student Tony Tramontin to test another species. Like canaries, song sparrows' singing virtuosity peaks seasonally. But song sparrows are like zebra finches in that they sing the same song throughout life. If neurogenesis in the adult HVC were specifically involved in learning rather than performing a song, Brenowitz and Tramontin reasoned, then song sparrows, like zebra finches, would not add new neurons each season.

Tramontin trapped wild song sparrows, injected them with a marker for new neurons, recaptured them, and found, surprisingly, that the sparrows added lots of new neurons to their HVCs seasonally, as canaries do. The results “argue against the hypothesis that the changes evolved specifically to enable these kinds of seasonal changes in learning,” says Brenowitz, but they don't rule out the possibility that, in seasonal song-learners like canaries, the new neurons aid learning.

Another twist came in 2000 when Nottebohm postdoc Constance Scharff, along with Harvard neuroscientist Jeffrey Macklis and other colleagues, simulated in zebra finches the loss and rebirth of neurons that occurs seasonally in canaries. The team selectively killed neurons in the HVC of zebra finches and found that new neurons migrated into the HVC, apparently replacing those that had died. The birds' song was “markedly degraded” after the loss of the neurons, Nottebohm says, but some of the birds' songs returned just as new neurons were added to the HVC. The songs were restored to different degrees but, true to their species, the birds did not sing anything new. Together with Brenowitz and Tramontin's work, that suggests that neurogenesis in the HVC is vital to the birds' ability to sing, but there remains, says Nottebohm, “no direct … evidence that the new neurons are involved with learning.”

## Sniffing for a function

As in the bird's song system, researchers have a good idea where new neurons fit into the circuitry of the rat's olfactory bulb, but they are just beginning to get clues about what function the neurogenesis may serve. The new neurons join the ranks of so-called granule cells, which are activated by the neurons that respond to odors. Their primary role is to blunt the activity of neighboring cells.

This blunting helps sharpen the pattern of neural activity evoked by an odor, according to behavioral research done by Pierre-Marie Lledo, Gilles Gheusi, and their colleagues at the Pasteur Institute in Paris. They reported in 2000 that a strain of mutant mice with deficient olfactory neurogenesis, and therefore relatively few granule cells, had difficulty discriminating between paprika and cinnamon, a distinction normal mice can make easily. The mutant mice seemed able to detect and remember odors normally, Lledo says, suggesting that they didn't have an overall deficit in their sense of smell. The study did not distinguish the value of newborn granule cells from those that might have been there all along, however, an issue that the Lledo team is now addressing.

The sense of smell is essential for behaviors more fundamental than distinguishing cinnamon from paprika, including a mother rodent's ability to recognize and nurture her young. That led Calgary's Weiss to wonder if rodent mothers would benefit from an extra influx of new neurons into the olfactory bulb to hone their olfactory discrimination. If so, he proposed, pregnancy might boost olfactory neurogenesis.

A team from Weiss's lab led by postdoc Tetsuro Shingo, working with the lab of their Calgary colleague Jay Cross, injected pregnant rats with a chemical to label new neurons. Neurogenesis rates jumped during pregnancy by 65%, they report on page 117, peaking around the seventh day of a 21-day gestation and again after delivery. They also saw a subsequent increase in new neurons integrated into the olfactory bulb.

The team identified prolactin, a hormone that increases during pregnancy, as the neurogenesis trigger and discovered that mice deficient in prolactin receptors have only half the surge in neurogenesis during pregnancy. Weiss notes that other groups have shown that the receptor-deficient mice tend to ignore their young, a behavior that could be due to an olfactory deficit. But, he cautions, “you have to take that with a big grain of salt, because transgenic animals may display unusual behaviors.”

Pregnancy isn't the only condition that increases prolactin production, Weiss adds: “In males and females one hormone is increased immediately after sexual activity, and that is prolactin.” That fact, coupled with his new report, has led some people to joke that sex must be good for the brain. That's “going off the deep end,” Weiss laughs. “We don't know the relevance that has to human neurogenesis at all.”

## Memory's neurons

Such titillating observations aside, most people are less concerned about their olfactory system than they are about the other mammalian brain area where neurons are known to be born: the dentate gyrus of the hippocampus. The hippocampus, which shrinks with age, disease, and depression, is an area that every aging human has an interest in preserving and rejuvenating if possible.

So when the Salk's Gage and his colleagues found neurogenesis in the hippocampus of adult rats, and subsequently showed that housing rats in an enriched environment filled with challenging toys more than doubled the survival rate of new hippocampal neurons and tripled it in aged rats, they generated a lot of excitement. Decades earlier, researchers at UC Berkeley and the University of Illinois, Urbana-Champaign, had shown that exposing animals to enriched environments could increase their mental power and brain size; the new work suggested that at least part of that result might be due to new neurons.

Further correlations between neurogenesis and behavior kept rolling in. In 1999, Rutgers's Shors, with Elizabeth Gould of Princeton University and their colleagues, reported that a challenging learning task, in which rats learned to associate a sound with a later shock to the eyelid, enhanced the survival of newborn neurons in the dentate gyrus. Gage and postdocs Henriette van Praag and Gerd Kempermann found that running in an exercise wheel increased rats' dentate neurogenesis by a whopping 80%. Meanwhile, other groups, including Bruce McEwen's at Rockefeller University, showed that depression or stress diminish neurogenesis in the dentate, and Ronald Duman and his colleagues at Yale University discovered that antidepressant drugs give neurogenesis a boost.

Out of this work a picture was emerging of three crucial steps in the life of new neurons: birth, differentiation into neurons, and survival. And different manipulations variously affected all three steps.

To better understand what the new neurons might be doing, researchers wanted to see them in action. Most experiments on neurogenesis and neuron survival had used a label that cannot reveal whether new neurons become integrated into the brain's circuitry. To get around this problem, van Praag, Gage, and their colleagues injected mice with a fluorescent marker that could be seen under a microscope while the neurons were still alive. With this method, “you can take brain slices and find when the cells functionally integrate into the circuit,” says Gage. Van Praag reported in February that 1 month after injecting the label, she saw newly born neurons linked up to the hippocampal circuitry and functioning. “The new cells actually [fire],” says Gould. “That really suggests that they are doing something.”

A number of experiments have suggested that the new neurons might boost learning. The evidence has been correlational, however, and it doesn't all agree. The early data “seemed to fit with the idea that more neurons enhanced learning and fewer new neurons would be detrimental to learning,” says Gould. “But now there are also reports that suggest the opposite or have found no correlation between the number of new neurons and learning.” For example, Eberhard Fuchs and his colleagues at the German Primate Center in Göttingen recently reported that stress decreases dentate neurogenesis in adult tree shrews, but the stressed shrews were better at a spatial navigation task than controls. Gage's team found that a maze-learning task did not enhance neurogenesis. And Michela Gallagher of Johns Hopkins University in Baltimore, Maryland, reported at the Society for Neuroscience meeting in Orlando, Florida, in November that although cell proliferation in the dentate declines with age in rats, some aging rats perform as well as young rats on memory tests. “The state of the field now reminds me of when you go to clean your office,” says Gould. “It gets a lot worse before it gets better. We are in that stage where we've pulled everything out of the desk, and it's a big mess.”

Some of the confusion, Gage says, may arise because many studies have simply measured the birth of new cells without verifying that those cells become neurons, a practice that can miss shifts in the percentage of new cells that become neurons. Beyond more rigorous identification of neurons, “what is needed are some good techniques for selectively stopping neurogenesis,” to get to the issue of cause and effect, says UC San Diego's Squire, and everyone in the field seems to agree. Last year, Shors and Gould took a stab at stopping adult neurogenesis. They treated rats with a drug called MAM that blocks neurogenesis by killing dividing cells. They used a dose that didn't make the rats overtly sick but depleted dentate neurogenesis by 75% and found that the treatment diminished the rats' ability to perform the eye-blink test. Other groups have used x-rays to block neurogenesis and have found an impact on learning.

But such treatments might have other unknown effects on the brain. The studies “are really confounded [because] if you see a decrease in performance, it may be due to the fact that the animal is sick, or you have affected some other brain region, or some other process in that same brain region,” says Gould. Much better, researchers agree, would be transgenic animals in which researchers could selectively thwart neurogenesis in particular regions at particular times.

That is something “we dream of doing,” says UC San Francisco's Alvarez-Buylla, and many groups are working hard to make the dream a reality. In anticipation of a model system that can provide clear answers, people in the field are already lining up to place bets on whether adult neurogenesis plays an essential role in learning or other brain functions. “I'm on the optimistic side,” says Gage. “I think it does, and we are going to find out how.”

10. MANTLE DYNAMICS

# Plumes From the Core Lost and Found

1. Richard A. Kerr

The first clear seismic images of deep, rising magma plumes support, in part, a theory under fire

SAN FRANCISCO—For a 30-year-old unproven hypothesis, mantle plumes have shown remarkable vigor. Most geoscientists assume that the plumes, columns of hot rock rising 2900 kilometers to the surface from the very bottom of the rocky mantle, explain volcanic hot spots such as Hawaii and the great magmatic outpourings of the geologic past called flood basalts. Some have even posited the possible evolutionary effects of plumes spewing such huge eruptions.

No more the cozy comforts of ignorance. Increasingly detailed seismic probing of Earth's interior is forcing geologists to confront some cold, hard truths about these elusive phenomena. Last month, at the fall meeting of the American Geophysical Union (AGU) in San Francisco, scientists reported that not all hot spots have plumes. Some insisted that plumes more than a few hundred kilometers in depth could not form because of the mantle's physical properties. Yet in the face of this assault on the status quo, the meeting also featured striking new evidence of how some hot spots are fed from the deepest reaches of the mantle.

“There was a lot of plume bashing going on” at the AGU meeting, says seismologist Göran Ekström of Harvard University. Don L. Anderson of the California Institute of Technology (Caltech) in Pasadena gave three invited talks and co-authored two more presentations, none of which missed an opportunity to put down plumes. Anderson, a pillar of the geophysics community, believes that the mantle's physical properties preclude the formation of narrow, buoyant plumes in the lower mantle and in fact seal off the mantle below about 1000 kilometers. Instead, he argues, chains of volcanoes like the one anchored at the Hawaiian hot spot could form along a crack in the plate that lets hot mantle rock a few hundred kilometers down rise to the surface and melt.

A case in point seems to be Yellowstone, one of the largest continental hot spots. Two groups of seismologists—Jason Crosswhite and Eugene Humphreys of the University of Oregon, Eugene, and Brian Zurek and Kenneth Dueker of the University of Wyoming, Laramie—reported that they can see no sign that the Yellowstone hot spot is fed from deeper than 200 kilometers, even though they can probe the mantle down to 660 kilometers.

The Oregon group used a standard imaging technique, in which the arrival times of seismic waves from distant, large earthquakes are recorded by scores of seismometers spread across the region. Waves that pass through hotter rock beneath Yellowstone are slowed relative to those encountering only average temperatures. The technique combines the delays of all the waves crisscrossing beneath the hot spot, the way computed tomography forms an image of the body. Although their instrumentation would have allowed them to detect anomalously high temperatures down to 400 kilometers, there were none detectable deeper than about 200 kilometers.

The Wyoming group used a different technique to look deeper than 400 kilometers. Any deep plume would slice through the two levels at which mantle minerals undergo phase changes—marking the “transition zone” between upper and lower mantle—whose exact depths depend on temperature. If pierced by a deep plume, the transition zone should thin. But although the pair reported that the seismically determined depths to these phase changes undulated across Yellowstone, they found no such thinning. Such results constitute “the first conclusive evidence of the nonexistence of a plume under a classic hot spot,” says seismologist Richard Allen of the University of Wisconsin, Madison.

Yellowstone isn't the only place where deep-plume hunters are coming up empty- handed. Jeroen Ritsema of Caltech and Allen have applied the tomographic technique to a set of global seismic observations, paying particular attention to the mantle beneath 45 hot spots on most people's lists. “The relation between hot spots and plumes has been implicit in many people's minds,” says Allen, but “it seems clear there is not a plume beneath every hot spot.” They have identified only eight plumes going deeper than 200 kilometers.

So where do plumes actually span the mantle? After 4 days of widespread plume bashing at the meeting, seismologist Tony Dahlen of Princeton University delivered an hourlong invited lecture that shored up conventional wisdom on the topic. The work, which involves sharpening up global tomographic images, offers evidence of at least a dozen deep, continuous plumes rising beneath major hot spots worldwide. “People recognize the first images … that are actually convincing,” says Dahlen's Princeton co-worker, Guust Nolet. Allen just calls it “the most exciting thing I saw at AGU by a long way.”

A key to the Princeton plume imaging was to think of a seismic wave's behavior in terms of a hollow banana rather than just a thin line. Seismic waves actually ripple away from an earthquake in all directions, but for the purpose of analysis, seismologists traditionally consider seismic waves to be a collection of lines or “rays.” In conventional analyses, when a ray path passes through a hotter blob of rock, the full slowing of the ray is assumed to be recorded when the wave eventually reaches a seismometer. But, at least in the case of lower-frequency waves passing through skinny blobs like a plume, a ray begins to “forget” its slowing as seismic energy radiates into the ray path from adjacent parts of the wave. Nolet and Dahlen, working with former Princeton postdoc Shu-Huei Hung of the National Taiwan University in Taipei, concluded that a more useful representation would be a hollow banana: most sensitive around a curving ray path (all seismic waves are curved by the deep Earth) but insensitive at its center.

Graduate student Raffaella Montelli of Princeton used the new technique in analyzing a relatively small but high-quality set of 87,806 seismic recordings assembled by seismologist Guy Masters of the Scripps Institution of Oceanography in La Jolla, California. In Princeton's final global image, the features beneath the classic hot spots of Hawaii, Tahiti, and Easter Island “really are deep mantle plumes,” said Dahlen. Some hot-spot plumes, such as those rising to Réunion in the Indian Ocean and the Azores in the Atlantic, actually branch off one of the two huge “superplumes” rising into the lower mantle beneath the South Pacific and Africa (Science, 9 July 1999, p. 187).

Not every hot spot has a deep plume in the Princeton tomography, however. Yellowstone is “iffy,” says Nolet, and nothing deep feeds Europe's shallow Eiffel plume or Africa's Tibesti hot spot. Absent plumes might reflect patches of sparse data, says Nolet, but “there are a lot of things we call hot spots and associate with plumes that may be shallow.”

“It was very impressive,” says seismologist Yang Shen of the University of Rhode Island, Narragansett. From the Princeton presentations and his own work with Hung on Iceland data, he finds that the hollow-banana approach improves plume images substantially, up to 100% in the upper mantle beneath Iceland. Seismologist Adam Dziewonski of Harvard was more cautious after hearing the rapid-fire presentations. “I'm usually pretty skeptical when people say they get images of plumes,” he says. In the Princeton case, he wonders if they haven't somehow smeared signals from shallow hot rock down into the lower mantle. He's waiting for the Princeton group to complete its testing of the tomography.

Plumes spanning the mantle would have a stimulating effect on a range of earth science. They could clarify how cooling of the interior drives mantle churning. Geochemists would have a better idea of where to locate the mantle's five compartments that store material for up to billions of years. Geologists might better understand the massive flood basalts—thought to spill from the bulbous heads of rising plumes—that dot the globe and are speculated to have overheated climate and triggered extinctions (Science, 6 December 1996, p. 1611). Plumes may even shatter supercontinents. Now that would be true vigor.

11. PHYSICS

# Researchers Race to Put the Quantum Into Mechanics

1. Adrian Cho is a writer in Grosse Point Park, Michigan.

Machines that make the slightest possible motion could lead to wild new technologies and help reveal why the weird rules of the microscopic realm don't apply to our everyday world

Like fidgety 3-year-olds, tiny objects simply cannot sit still. Atoms, molecules, and other minuscule particles must constantly flit about because of a law of nature that says if you know precisely where something is, you can't know where it's going, and vice versa. The Heisenberg Uncertainty Principle is an unavoidable nuisance; experimental physicists have observed countless times that the smallest bits of stuff in nature wriggle whenever they try to pin them down. However, no one has directly observed the ineluctable quantum quivering—or zero-point motion—of a larger, humanmade object.

That may soon change. Exploiting recent advances in nanotechnology, physicists are racing to fashion vibrating gizmos that can make and measure literally the slightest possible motion. At least four groups hope to reach the quantum limit of motion within months. The feat could open the way for tiny, fingerlike force detectors with the highest possible sensitivity, says Andrew Cleland of the University of California (UC), Santa Barbara. Such detectors might enable researchers to quickly decode DNA and other large molecules, and someday they might serve as the guts of superfast quantum computers.

Quantum machines might even help solve a conundrum as old as quantum mechanics itself: Why can a tiny object like an electron be in two different places at once, whereas a big thing like a pencil or a person cannot? “We don't see quantum behavior in our macroscopic world, so in some sense we're protected from quantum mechanics,” says Miles Blencowe, a theoretical physicist at Dartmouth College in Hanover, New Hampshire. “What protects us?” To find out, he says, experimenters might try putting progressively bigger mechanical devices into here-and-there “superpositions” to observe what, if anything, goes wrong.

First, though, physicists must reach the quantum limit of mechanical motion. That will require overcoming serious technical challenges, says Michael Roukes of the California Institute of Technology (Caltech) in Pasadena: “This is just damned hard stuff to do.”

## A subtle vibe

The biggest hurdle is heat. Thermal energy makes large objects wiggle, and at any achievable temperature those vibrations overwhelm the zero-point motion. For example, according to quantum mechanics, a tuning fork can gain or lose energy only in discrete dollops whose size is proportional to the fork's frequency of vibration. Because the frequency is low (440 cycles per second for concert-pitch A), each quantum of energy is so small that the fork contains billions of them even at a degree above absolute zero. To suck out enough of them to see the zero-point motion, the fork would have to be cooled to a few billionths of a degree.

Or an experimenter could increase the size of each quantum of energy by cranking up the frequency of the fork. Shrinking the thing would work, as a smaller fork will ring at a higher frequency just as a violin produces higher notes than a double bass. That's essentially what physicists are doing by fashioning tiny vibrating beams only dozens of nanometers thick and a few micrometers long out of materials used in microchips, such as gallium arsenide, silicon, and silicon nitride. With masses of several millionths of a nanogram and containing just a few billion atoms, the tiny beams vibrate hundreds of millions of times a second: Roukes and colleagues have just punched through the billion-cycle-per-second barrier. That means researchers might see the zero-point motion of the devices if they can cool them to a few thousandths of a degree—a routinely obtainable temperature.

Of course, such tiny devices move far less than an ordinary tuning fork. The zero-point motion of a typical beam would be less than a thousandth of an atom's width. To track such small movements, most researchers employ an exquisitely sensitive electrical valve known as a single-electron transistor. Electrons can hop one by one from the input to the output of the device, but only if a certain voltage is applied to a controlling “gate” electrode. By having the motion of the beam change the gate voltage, perhaps through an electric field emanating from another electrode on the beam, researchers can take their observations by monitoring the current through the transistor (see figure on p. 36).

Cobbling together such tiny motion sensors is no mean feat, says Robert Knobel of UC Santa Barbara. “We're definitely leveraging every bit of semiconductor and lithography technologies,” he says.

## Spin doctors

Reaching the quantum limit of motion could open new avenues of technology, says Chris Hammel of Ohio State University in Columbus. That's because tiny beams and fingerlike appendages called cantilevers make force detectors whose precision is limited by how well researchers can follow their motion. A cantilever with a magnetic tip might even be able to sense the undulating magnetic field of a single twirling atomic nucleus—the ultimate in nuclear magnetic resonance. “If you can look at each nuclear spin in a molecule,” Hammel says, “you can take a single molecule and determine its structure.”

More immediately, quantum-limit measurements might lead to improved techniques for doing nuclear magnetic resonance on small samples, such as films, or probing the architecture of microchips. Much farther down the road, individual magnetic nuclei might serve as the qubits—bits that can be set to 0, 1, or 0-and-1 at the same time—that will enable quantum computers to crunch many different numbers at once. If so, tiny nanomachines might serve to read out those qubits.

Nanomachines themselves might serve as qubits, says Robert Blick of the University of Munich, Germany. Even as he pushes to reach the quantum limit with a single device, Blick has a plan to string together several machines as qubits. If such fanciful schemes pan out—and that's a big if—the quantum computers of the future might have as much in common with the adding machines of yesteryear as with personal computers.

Most speculatively, machines in quantum motion might help explain why large objects never appear in two-ways-at-once superpositions. Many physicists think that quantum mechanics itself provides the solution to the puzzle. An individual atom can persist in a superposition, they say, because it interacts only very weakly with its environment. A big object, on the other hand, inevitably feels the effects of its surroundings much more strongly, so its quantum states entwine with the quantum states of the rest of the universe in a process called decoherence. The environment tips a large object into one possible condition or the other, eliminating the this-and-that superposition, says Wojciech Zurek, a theorist at Los Alamos National Laboratory in New Mexico.

Other theorists, however, suspect that the story is stranger than that. Roger Penrose of the University of Oxford, U.K., theorizes that a large object that's in two places at once interacts with itself through gravity in a way that tugs it to one place or the other. Philip Pearle of Hamilton College in Clinton, New York, speculates that the universe is filled with a strange “noise” that jostles a big object one way or the other. If there is something beyond quantum mechanics, “then there's a huge fish to find out there,” says Keith Schwab, a physicist at the National Security Agency (NSA). To find the fish, Schwab says, you have “to take these devices and make them bigger and bigger and bigger.”

Working at NSA's Laboratory for Physical Sciences in College Park, Maryland, Schwab has devised a scheme to do just that. He envisions putting a tiny dab of superconducting metal called a “Cooper pair box” on a vibrating beam. Electrons in superconductors travel in pairs, and individual pairs can jump into or out of the box. Electrons in the box will feel a pull, thanks to the electric field between the beam and the gate electrode of the single-electron transistor. As a result, when an extra pair of electrons hops into the box, the beam will draw slightly closer to the gate. Schwab plans to put the box in a superposition of two states that differ by one pair of electrons. If all goes well, he says, the pull of the field should force the beam to occupy two different positions at the same time.

But before researchers can attempt such ambitious stunts, they have to prove that they can perceive the slightest movement. Speeding their progress is a friendly and almost familial rivalry: Schwab, UC Santa Barbara's Cleland, and Munich's Blick all worked as postdocs in Roukes's lab at Caltech. With so many challenges before them and so much competition between them, these researchers have to keep moving—just like the tiny machines they study.

12. CLIMATE CHANGE

# Consensus on Ecological Impacts Remains Elusive

1. Mari N. Jensen*
1. Mari N. Jensen is a science writer in Tucson, Arizona.

Two big new studies strengthen the case that global warming is causing biological effects, but critics say even the additional data fall short of proof

Working Group II had a problem. The group, part of the Intergovernmental Panel on Climate Change (IPCC), reviewed 44 studies showing that more than 400 species of plants and animals across the globe had shifted their ranges or changed behaviors such as the timing of egg laying. To the biologists on the committee, this was a strong signal of climate-induced effects on a variety of biota. They wanted to give the finding a very high level of confidence, 95%. But the nonbiologists, mostly economists, advocated a confidence level of 33% to 67%, and no more.

Ultimately, IPCC's consensus document listed a high confidence level, 67% to 95%. But despite that paper compromise, the group remained split on how certain it was that global warming caused the observed biological changes. The issue: which data should be considered in such an analysis.

“We all think we know how to analyze data, so you'd think there wouldn't be any disagreements,” says one of the IPCC authors, Camille Parmesan, a population biologist at the University of Texas, Austin. “But we would look at the same data, … and one person says, ‘So what?’ and the other person says, ‘Wow, look at that!’”

Richard Tol, an environmental economist at Hamburg University in Germany, for instance, has questioned whether the data set represented a “fair sample.” He points out that biologists tend to do studies in regions where impacts of climate change are expected; in addition, he said, studies showing no effect are unlikely to be published. Others, including economist Gary Yohe of Wesleyan University in Middletown, Connecticut, said that high confidence was unwarranted because the analysis simply showed a correlation, not cause and effect.

In an effort to persuade the skeptics, Parmesan teamed up with Yohe, also an IPCC author, to reanalyze these and other data. Another IPCC author, Stanford University ecologist Terry Root, independently embarked on a similar study. The two papers, published in the 2 January issue of Nature, are being touted as the most comprehensive meta-analyses to date of the biotic effects of global warming. To many, they clinch the case. But Tol and other working group members maintain that the two studies don't provide any greater level of confidence than before that global warming is causing the observed biological changes.

As part of her joint project with Yohe, Parmesan sorted species into four categories: those that changed their ranges or behaviors in accord with global warming predictions, those that did the opposite, species that did not change, and species showing changes that couldn't be ascribed to global warming. She found that 87% of 484 species analyzed changed their timing as predicted by models of global warming. Distributional shifts were consistent with predictions for 81% of 460 species. Such changes would occur by chance less than one time in 10 trillion—an airtight case, argued Parmesan.

Not so, said Yohe, pointing out that the result was still based on correlations. He then developed a probabilistic model that could use Parmesan's data. A key variable controlling the confidence level was the likelihood that a species' observed change was properly attributed to climate change. The model's result: an estimate of medium confidence, or 33% to 67%.

Parmesan then focused on effects she calls “sign switching”; these can only be explained by a temperature increase or decrease—for example, spring signals such as flowering that happen earlier during warmer decades and later during cooler decades. Between 80% and 100% of the 294 species examined switched as predicted by temperature flip-flops.

Yohe found these tests to be a compelling demonstration of causation, even though the data sets were smaller. “To the degree to which sign switching occurs, it's very convincing,” he says.

In a separate study, Root attempted to strengthen the literature review started at IPCC by gathering a larger data set. Her team analyzed studies in which either positive or negative changes (such as range shifts or earlier flowering) were observed and linked to temperature, but, unlike Parmesan, it did not include studies that showed no effects. In all, the team analyzed 143 studies covering 1468 species of plants and animals around the globe; of those, 81% exhibited changes consistent with those predicted for global warming. When Root's team focused on timing specifically, it found that spring events shifted an average of 5.1 days earlier per decade.

“What we're finding is a correlation,” says Root. “But it's such a robust correlation that the probability it's by random chance is minuscule.”

Harvard University biological oceanographer James McCarthy, co-chair of Working Group II, says, “Before these analyses, [the idea of] consistent response of organisms to temperature was just a hypothesis. Now we know that organisms are responding globally—on all continents, across a large range of organisms.”

But economists say that the biologists have not yet proved their case. Tol says Parmesan and Yohe try to deal with problems of study selection and bias, but they fail because the bias is inherent in the nature of published literature. “The Root et al. paper does not try,” he adds. Even so, he says the papers will stimulate other groups to work on the problem, and therefore he finds the papers “a useful first step.”

Yohe still won't assert with 95% confidence that global warming is causing biotic changes. However, as more evidence accumulates that he can apply in his probabilistic model, he expects to come around.

13. # The Milky Way's Restless Swarms of Stars

1. Robert Irion

Violence is routine in globular clusters, where suns crash, binaries burn, and the whole cluster teeters on the brink of collapse

The night sky would look spectacular from a planet near the core of a globular cluster, one of the Milky Way's compact swarms of stars. Instead of the paltry few stars we see within several light-years of our sun, cluster aliens would face a vista of 100,000 stars or more. Astronomers would have crisp views of binary partners that whip around each other in hours or minutes, pulsars that spin nearly 1000 times every second, and perhaps a nest of neutron stars or the event horizon of a sizable black hole at the center of it all.

There's just one problem: Such a planet almost certainly doesn't exist. Stars in the hearts of globular clusters interact so closely and so frequently, in astronomical terms, that planetary systems can't survive the chaos. Indeed, stars themselves are not immune. Many get banished to the cluster's outskirts or ejected into deep space after intense gravitational encounters. Some stars even collide, making bizarre new objects that are at least a billion times more likely to arise in globular clusters than elsewhere in the galaxy.

It's no wonder that scores of observers and theorists peer beyond our sedate galactic neighborhood to study these angry swarms of stellar bees. “Globular clusters are the breeding grounds for stellar exotica,” says astronomer Jonathan Grindlay of the Harvard-Smithsonian Center for Astrophysics in Cambridge, Massachusetts. “They really are weird and wonderful places,” adds astronomer Michael Shara of the American Museum of Natural History in New York City. “And it's all because stars get very close and do very unpleasant things to each other.”

## Stellar ecology

The Milky Way's 150 known globular clusters live within the galaxy's halo, a sparsely populated sphere of old stars that cocoons the disk and central bulge. They dart on vast swooping orbits; a time-lapse simulation of clusters in motion creates the impression of moths fluttering around a street lamp. Many of these tight knots form attractive targets in small telescopes, and they spread gloriously into separate stars when the Hubble Space Telescope (HST) stares at them.

Detailed surveys by both HST and the Chandra X-ray Observatory have given globular clusters a new cachet. Cluster cores, the wellsprings of the exotica Grindlay talks about, are no longer impenetrable walls of light. “They've become transparent to both HST and Chandra,” says astronomer Adrienne Cool of San Francisco State University in California. “We see galaxies behind the clusters. Even an incredibly dense core is spread out on the sky.”

Such surveys have revealed profound differences among clusters. For example, the nearby object 47 Tucanae teems with millisecond pulsars—old neutron stars revived and spun up to hundreds of revolutions each second, presumably by pulling gas from a newly captured companion star. Indeed, 47 Tucanae might contain about 100 millisecond pulsars, according to an analysis by Grindlay and his colleagues in the 10 December 2002 Astrophysical Journal. By contrast, the galaxy's largest cluster, Omega Centauri, has none. An explanation might come from new Chandra and HST images, now under scrutiny by teams that Cool is leading. It's a tall task: HST's Advanced Camera for Surveys required a span of nine exposures to capture 1.7 million stars in Omega Centauri, just half its population, because the cluster sprawls across such a large chunk of the sky.

Other answers about the distinctions among clusters are coming from new computer simulations of how they evolve. The simulations rely on GRAPE boards, ultrafast processors built by astrophysicists at the University of Tokyo to compute gravitational interactions (Science, 13 July 2001, p. 201). “I think we are very close to doing a globular cluster from birth to death,” says astrophysicist Simon Portegies Zwart of the University of Amsterdam, the Netherlands. Already, GRAPEs simulate the internal motions of a small cluster of 100,000 stars over a breathtaking sweep of time scales: from binary stars that orbit in mere hours to the lifetime of the cluster itself, spanning 10 billion years.

The work reveals that the dynamics of the cluster and the evolution of its stars are inextricably linked. Stars are so close that any changes in their mass or size influence how they interact with neighbors. Portegies Zwart and his colleagues use the phrase “stellar ecology” to convey this point. “We borrowed the term from biology,” he says. “Everything goes together into one big pot of interactions, and then something beautiful comes out.”

## Binary molecules

The pot's most critical ingredients are binary stars. “A globular cluster's energy is locked up in these stellar molecules,” says astrophysicist Piet Hut of the Institute for Advanced Study in Princeton, New Jersey. “It can be liberated if they interact with other single or double stars, and that can heat the whole cluster.”

If Hut's words sound like thermodynamics, that's intentional. Scientists draw a parallel between heating a gas and exciting the stars in a cluster with the motions of its binaries. “The particles in the gas are stars instead of molecules, but it's the same physics,” says astrophysicist Frederic Rasio of Northwestern University in Evanston, Illinois.

A globular cluster's binary stars respond to the density of the overall cluster. If the cluster is puffed up and stars aren't jammed too close together, there are fewer encounters between binary systems and other stars. But when gravity pulls the cluster together more tightly—as it inexorably does —the close flybys skyrocket. A third star can swing past a binary and, via a complex gravitational dance, gain kinetic energy. That forces the binary stars into a tighter orbit around each other, a process called binary “burning.” In some cases, the third star can replace one of the two binary members and fling it to the cluster's outskirts. Such exchanges puff up the globular again. “It's like having a bunch of tightly coiled springs right at the center,” says Shara.

This give-and-take carries a cost. Many of the gravitational boomerangs are so extreme that stars flee the cluster entirely, kicked out by their kin into the galaxy. Such “evaporation,” as astronomers call it, is the main threat to a cluster's long-term survival. Moreover, the binary fuel supply runs low as the orbits in each pair grow tighter after repeated interactions. A cluster might last a billion years, 10 billion years, or more, depending on its initial mass and how many binaries it contains. But ultimately, it's condemned to dwindle from a grand swarm of stars into a nubbin of old survivors.

No one knows how easy it is for binary stars to arise when globular clusters are born. However, calculations show that it doesn't take many of them to affect the cluster's evolution. If just 5% to 10% of a cluster's stars are in binaries, they impart enough “heat” to the rest of the stars to prevent the cluster from imploding on itself. Indeed, just a handful of binary stars can yield more energy than the gravitational energy that binds a whole cluster together.

Still, catastrophic collapses do occur. A surprising portion of the Milky Way's globular clusters—as much as 20%, by some reckonings—have undergone “core collapse.” In that extreme state, the density of stars keeps rising all the way to the center. The core becomes as crowded as the finite number of stars will allow, perhaps 10,000 times more densely packed than a normal cluster.

At this point, says Northwestern's Rasio, observations and theory conflict radically. Models predict that a core-collapsed cluster should immediately rebound as the innermost binaries—forced into overdrive by frequent encounters—expel stars to the outskirts once again. According to that scenario, core collapse is such a fleeting state that astronomers should see no such systems today. “After core collapse, we really don't understand what happens,” Rasio says. “Anyone who claims to have a viable model is hiding something.”

Somewhat better understood are the weird objects that arise when stars near a cluster's core smash together. Once again, binary stars are the catalysts. Two stars orbiting each other create a huge volume of space that a third star can enter and—if the gravitational duel turns deadly—spiral into a collision. “It's a colossal magnification effect,” says Shara. “It makes stars millions or even billions of times more likely to interact or collide in globular clusters than anywhere near the sun.”

Astronomers have found signs of crashes in virtually every cluster they've studied. Called “blue stragglers,” the objects appear younger, hotter, and more massive than the cluster's other stars, all of which were born at about the same time in the galaxy's infancy.

Simulations by theorist Alison Sills of McMaster University in Hamilton, Ontario, and others predict that, by definition, a new blue straggler doesn't become a star for thousands of years after its violent rebirth, because nuclear fusion takes time to resume in its riled-up core. Further, the merged object rotates so fast after the collision that it nearly flies apart. “It may throw off some mass, but we really don't know how to make the spin subside,” Sills says. Insights might come from detailed HST observations of nearly 100 blue stragglers, in the final stage of analysis by a team including Shara and astronomer Rex Saffer at Villanova University in Pennsylvania.

## Getting at the core

Another cluster mystery lies at the center of a swirling debate: What lurks at the innermost core? NASA alluded to this in a press conference in September touting a major discovery by HST: “intermediate-mass black holes” at the centers of two clusters, M15 in our Milky Way and G1 in the neighboring Andromeda galaxy. Weighing in at thousands of times the mass of our sun, these black holes were a putative steppingstone between smaller black holes formed by supernovas and titanic ones at the cores of galaxies. Today, astronomers are standing by their claim for a black hole in G1, but the M15 result has eroded into statistical insignificance.

To identify the suspected midsize black holes, astronomers used HST to measure the orbital speeds of stars at the clusters' cores. Rapid orbits suggested that the gravity of some dark, hidden mass was pulling stars into unexpectedly fast whirls around the centers. But there's a catch: Other compact objects at the center could produce the same dynamics—for example, thousands of neutron stars within a small volume of space. Such nests are improbable in most places. However, they could arise in globular clusters, where massive objects quickly settle to the center like heavier molecules at the base of Earth's atmosphere.

For M15, the HST astronomers relied on an earlier estimate of the number of neutron stars in the cluster. The relatively low figure allowed them to state with confidence that neutron stars alone could not explain the star motions they saw. Only later did the researchers determine that the figure they had adopted was three to five times too low. “We could no longer rule out the no-black-hole case” with the extra neutron stars, says astronomer Karl Gebhardt of the University of Texas, Austin.

That backtracking didn't surprise Hut, Portegies Zwart, and their colleagues, who conducted GRAPE simulations of an M15-like cluster with 128,000 particles. In the 1 January issue of Astrophysical Journal Letters, the team reports that M15 probably contains a dense warren of small black holes and neutron stars, all orbiting near the center. For an intermediate-mass black hole to arise, they claim, a cluster's giant stars must merge quickly in their youths, when they are bloated and more likely to crash. Once the stars blow up after about 10 million years to forge black holes and neutron stars, assembling a bunch of them becomes much more difficult. “Compact objects are so tiny that they almost never collide once they form,” Portegies Zwart says. “They kick each other out of the cluster rather than merge into something bigger.”

The debate is far from over. Gebhardt has data in hand from a ground-based survey of star motions in about 15 other Milky Way clusters, most of them easier to study than the crowded heart of M15. His analysis, forthcoming within a few months, suggests that intermediate-mass black holes may yet prove to be the best explanation. “I'm getting much more convinced now that this is a reality,” he says. “But I realize it will be an uphill battle for me to convince others.”

Whether or not they harbor black holes, globular clusters are unendingly strange to those who study them. “If the universe didn't make globular clusters, nobody would insist that they should be there,” says Stephen Zepf of Michigan State University in East Lansing. Fortunately, they do exist—giving astronomers their best clues about what happens to stars when the going gets dense.

14. # Breaking Up Is Easy To Do

## R.I.

Life as a globular cluster is no picnic. In an ongoing threat to a cluster's survival, gravitational mayhem deep in the core ejects its own stars (see main text) and whittles away its mass. Clusters are threatened from outside, too; our Milky Way snacks on them like finger foods. Over time, the galaxy can shred a compact cluster into ghostly streams of stars. “Once a globular cluster forms, it immediately starts dying,” says astronomer Stephen Zepf of Michigan State University in East Lansing. “It's just a matter of how long it takes.”

The galaxy's appetite stems from its fierce tides. Just as the moon's gravity raises swells of water on opposite sides of Earth, the Milky Way puffs up stars on both sides of a globular cluster by tugging harder on the cluster's nearer side than its farther side. This process sets a “tidal radius,” the boundary where a star feels equally drawn to the cluster and to the galaxy. Beyond this radius, stars drift into space—spreading along the cluster's orbit in both directions to form faint but persistent “tidal streams.”

When a globular cluster loops near the galaxy, the gravitational field grows, and the tidal radius shrinks. Each encounter shocks the cluster and forces it to cough up more stars. The most dramatic case is Pal 5, a small cluster caught in the act of dissolving by the Sloan Digital Sky Survey (Science, 14 June 2002, p. 1951). “Pal 5 is the paradigm of a tidally disrupted cluster,” says astronomer Steven Majewski of the University of Virginia, Charlottesville. His team has observed nodules of stars within Pal 5's tidal streams, marking each time the doomed cluster dove through the Milky Way's disk, compressing the tidal radius and freeing a burst of stars.

Astronomers think they see tidal streams emerging from as many as 20 other globular clusters, but the dim trails are hard to identify and trace. Still, it's clear that the Milky Way's outskirts are laced with the stretched-out remnants of vanished star systems. “The outer halo is almost completely networked with these streams,” Majewski says.

Consequently, today's assembly of about 150 clusters is a shadow of some grander swarm in the past. However, no one knows how many clusters dotted the skies in the Milky Way's youth. Some models by Zepf's colleague at Michigan State, astrophysicist Enrico Vesperini, suggest that the Milky Way originally had about 350 clusters; others peg the initial population at 1000 or more. Moreover, the survivors aren't typical; certain qualities helped them endure, most notably their higher masses and more-distant orbits. “There may have been a collection of fragile, fluffy things that we have no representatives of now, because they've been completely wiped out,” Zepf says. Studies of the types of clusters around young galaxies elsewhere offer clues about the Milky Way's globular heritage (see p. 63).

Our galaxy's gluttony actually adds a few clusters here and there. When dwarf galaxies spiral into our own, gravitational tides shred them as well. This is happening now to a loose agglomeration of stars called the Sagittarius dwarf (Science, 7 January 2000, p. 62); indeed, Majewski's group sees tidal streams from Sagittarius wrapping clear around the Milky Way. However, several of the dwarf's globular clusters will outlast this dissolution for billions of years. The Milky Way's largest globular cluster, Omega Centauri, might be the core of a dwarf galaxy whose outer parts were stripped off long ago, because it consists of an odd mix of stars with different ages and compositions. Our galaxy, it seems, can't quite finish every morsel it tries to consume.

15. # Outside Sources Frame a Galactic Puzzle

1. Govert Schilling*
1. Govert Schilling is an astronomy writer in Utrecht, the Netherlands.

The Milky Way doesn't have a monopoly on globular clusters. Powerful new telescopes have brought other galaxies' clusters into astronomers' view—and into the thick of their efforts to fathom the origin of a familiar but mysterious type of galaxy

When astrophysicist Ivan King signed a copy of his 1976 textbook The Universe Unfolding for his student François Schweizer, he wrote in it: “Make as much of this obsolete as you can, François!” Schweizer followed his teacher's advice. In 1987, 13 years after finishing his studies at the University of California, Berkeley, he proposed a radical new theory about the cosmic objects known as globular clusters.

Conventional wisdom held that the clusters—huge, dense balls of hundreds of thousands of stars—are among the oldest objects in the universe, born early in cosmic history when enormous clouds of gas collapsed under the pull of their own gravity, triggering huge bursts of star formation. Schweizer suggested that those primeval clusters are only part of the story: New globular star clusters, he proposed, are still forming where galaxies collide, deep in space.

“Back then, Ivan thought it was a crazy idea,” says Schweizer, now at the Carnegie Observatories in Pasadena, California. Theorists still heatedly debate whether newborn clusters could survive the turmoil of a galactic merger. It's a pivotal question, because the stellar archipelagos hold an important key to understanding the birth and evolution of galaxies. The chemical compositions revealed in their light are starting to yield important clues to the history of elliptical galaxies, common astronomical objects so enigmatic that astronomers aren't sure whether they are much older than our Milky Way galaxy or much younger. New results are being published almost weekly, and Schweizer's ideas and their rivals are in the thick of the fray. “This is an incredibly exciting time,” says Keith Ashman of the University of Missouri in Kansas City.

## Great balls of stars

Globular clusters long ago proved their astronomical value closer to home. About 150 of them are known to swarm around our own Milky Way galaxy in a 130,000-light-year-wide, more or less spherical halo that surrounds the flattened spiral disk of the galaxy. Well over 10 billion years old, more than twice the age of our sun, they consist mainly of slowly evolving, reddish dwarf stars. For decades, astronomers have been convinced that globular clusters are the natural byproducts of the formation of galaxies —the fossil remnants of the prehistory of the cosmos. As a result, Ashman says, the study of globular clusters “has had a huge impact on understanding how the Milky Way formed.”

Other galaxies, however, have globular clusters, too. Although most galaxies are so far away that even the largest telescopes can't pick out their individual stars, globular clusters stand out as bright pinpoints within the haze of light. Using large ground- and space-based telescopes, astronomers can study such “extragalactic” clusters and use them to trace the major star- forming episodes in a galaxy's history. “This is really hard work, for which you need the largest telescopes on Earth,” says Duncan Forbes of Swinburne University in Hawthorn, Australia.

For most extragalactic globular clusters, the relatively small light-gathering power of the 2.4-meter Hubble Space Telescope is insufficient. “Without the new generation of 8- and 10-meter telescopes such as Keck, Gemini, and the European Very Large Telescope, none of this would be possible,” says Stephen Zepf of Michigan State University in East Lansing.

Most studies of extragalactic globular clusters have focused on elliptical galaxies, which lack the flattened disks of spirals like the Milky Way. Because they consist mainly of old stars and have used up most of the interstellar gas from which new ones form, astronomers used to believe that elliptical galaxies are older than spirals are. Now, however, many think they assumed their present form relatively late in life, as a result of cataclysmic events.

Which view proves correct might hinge on its proponents' ability to explain two puzzling observations. First, elliptical galaxies—especially the giants that sit in the centers of dense swarms of galaxies—harbor huge numbers of globular clusters. One massive elliptical galaxy, called M87, is surrounded by at least 14,000 of them. Second, the star clusters in most elliptical galaxies seem to belong to two different populations, one slightly redder than the other. Sensitive spectroscopic observations of globular clusters have shown that the “red” clusters are younger than the “blue” ones are and contain higher concentrations of metals—astronomical jargon for elements heavier than hydrogen and helium that are produced by nuclear fusion inside stars. “This suggests that there have been distinct epochs and/or different mechanisms for the formation of globulars,” says Jean Brodie of the Lick Observatory in California.

What were those mechanisms, and when exactly did they occur? Most astronomers agree that the metal-poor blue clusters are ancient—older even than the galaxies they inhabit. When it comes to the red clusters, however, opinions differ sharply.

## Four models

Ashman and Zepf promote Schweizer's idea that most metal-rich globular clusters in elliptical galaxies result from galaxy mergers. According to this model, two spiral galaxies violently collide and merge into a giant elliptical. The original globular clusters of the colliding spirals, which were born before stars had time to create heavy elements, combine to form the low-metal blue clusters of the resulting elliptical. Meanwhile, shock waves in the interstellar gas of the two colliding spirals— now enriched with heavy elements— produce huge bursts of star formation and new massive star clusters with a higher metal content and a redder color.

“If you make spirals, you have one episode of globular cluster formation,” says Ashman. “If you smash spirals together, you have another one. In my opinion, that's where the evidence points.” Schweizer agrees: “I think it's the main process, even though there are still people who have trouble believing that elliptical galaxies can be young.”

Indeed, some astronomers still hold that many or most elliptical galaxies formed early in cosmic history. “The merger model has lots of very severe problems,” says Brodie. “Mergers obviously happen, and they do produce a few globular clusters, but they are the frosting on the cake.” To create the enormous numbers of globular clusters observed in many giant ellipticals, Brodie says, would take at least a dozen mergers, each churning out clusters at a furious rate. Collisions between galaxies, however, are so rare that few galaxies are likely to have experienced more than a handful of them. Schweizer acknowledges the difficulty. “I am at a loss to explain how these large numbers form,” he says.

Brodie and Swinburne University's Forbes believe that the vast majority of red globular clusters in elliptical galaxies are much older than the merger model predicts. Instead, Brodie suggests that all globular clusters, both red and blue, date from the early days of the universe. “Galaxies may have formed from primordial gas in a two-stage collapse,” she says. The oldest clusters—the bluer ones—formed during the initial collapse of protogalactic clouds, which contained hardly any heavy elements. After 1 billion or 2 billion years, during which the first generation of hot, luminous stars temporarily halted the initial collapse and enriched the interstellar gas, a second collapse phase produced the metal-rich red clusters and the bulk of the stars in the new galaxy. But the model is still a rough sketch, Brodie acknowledges: “We don't have detailed analyses yet. It's all rather handwavy.”

Forbes favors a variation on that theme, in which the second cluster population formed more gradually by a process known as hierarchical galaxy formation. In this scenario, galaxies grow over a few billion years through mergers of larger and larger gas clouds. Most cosmologists invoke hierarchical models to explain the origin of galaxies, Forbes says. “Unfortunately, the globular-cluster community so far hasn't talked too much to the galaxy-formation people,” he adds.

Still another model, put forth by Patrick Côté of Rutgers University in Piscataway, New Jersey, suggests that elliptical galaxies—and their store of globular clusters—have grown over many billions of years by consuming very large numbers of neighboring dwarf galaxies, each with its own handful of clusters. Such galactic cannibalism certainly takes place: Our own Milky Way is currently gobbling up the Sagittarius dwarf galaxy, along with its small retinue of four globular clusters. But William Harris of McMaster University in Hamilton, Ontario, says that Côté's “accretion scenario” can't explain the thousands of metal-rich globular clusters in elliptical galaxies such as M87. “There are not enough low-mass dwarf galaxies around in the real universe to make the numbers work,” Harris says. “My guess is that most of the red clusters arise through hierarchical merging of gas clouds in the early universe,” he adds, “but different processes could be at work in one galaxy. The question is, how much of the story does each of them tell?”

View this table:

## Age will tell

More-accurate age determinations of extragalactic globular clusters—particularly the metal-rich red ones—might put the competing models to an acid test. Schweizer, Ashman, and Zepf's merger model predicts that each elliptical galaxy will have clusters in just a few narrow age brackets; these will vary from galaxy to galaxy, depending on when the mergers that created each galaxy took place. Brodie's two-stage collapse model predicts that all red clusters will have more or less the same, very old age. In Forbes's hierarchical model, the red clusters are also ancient, but their ages span a few billion years. Finally, if Côté's accretion scenario is correct, red clusters in each elliptical galaxy will range widely in age.

“Determining ages precisely is one of the keys,” says Forbes. Thanks to the new generation of large ground-based telescopes, outfitted with sensitive spectrographs and infrared detectors, that is finally becoming possible. Early results appear to give some encouragement to merger enthusiasts. Over the past few years, for example, a number of intermediate-age red clusters have turned up in some elliptical galaxies, including the giant galaxy Fornax A. Those discovered so far are between 3 billion and 7 billion years old—easily young enough to have been formed in galactic collisions, Ashman says. “There have been hints for the existence of intermediate-age globular clusters for a few years,” he says, “so this is an exciting result, even though we don't know yet how common they are.”

Meanwhile, Schweizer and his co-worker Paul Goudfrooij of the Space Telescope Science Institute in Baltimore, Maryland, have found young globular clusters in so-called merger remnants: misshapen galaxies that bear the scars of a relatively recent merger but haven't yet settled down to the quiescent state of an elliptical. “This proves that globulars that form during a merger can indeed survive,” says Goudfrooij. However, Forbes warns that many more observations are necessary, because the galaxies studied so far might not be typical. “Galaxies are like people,” he says. “The more you get to know them, the more peculiar they turn out to be.”

Astronomers are eagerly awaiting a conference in Santa Barbara, California,* where presentations are expected to include results from sensitive wide-angle detectors and spectrographs that have recently come online on some of the new large telescopes. But no one expects one theory to land a knockout blow anytime soon, and some astronomers suspect that a combination of models will ultimately prevail. “If different processes are at work, it's of course more difficult to interpret,” says Harris, “but also more interesting. These are still early days.”

• *Globular clusters: formation, evolution, and the role of compact objects, 27 to 31 January.