# News this Week

Science  03 Sep 1999:
Vol. 285, Issue 5433, pp. 1089
1. SCIENTIFIC PUBLISHING

# NIH's Online Publishing Venture Ready for Launch

1. Eliot Marshall

It's time to stop talking about a free, public Web site for life science articles and start building one, says Harold Varmus, director of the National Institutes of Health (NIH). This week, Varmus announced that NIH will launch its electronic archive and publication site, now called PubMed Central (formerly E-biomed), beginning in January 2000. It will accept reports—both reviewed and some unreviewed material—comments, and data files from journals and scientific groups and redistribute them on the Internet at no charge. The final plan is described in a notice released by NIH on 30 August (www.nih.gov/news/pr/aug99/od-30.htm).

The Web site also gained one of its first major recruits last week: the American Society for Cell Biology (ASCB). According to ASCB director of publications, Heather Joseph, the society's council voted to share the complete contents of ASCB's journal, Molecular Biology of the Cell, with a 2-month delay after publication, “on an experimental basis.” Joseph explains, “We very strongly support the goal of barrier-free access to the scientific literature” and want to help NIH's “laudable” effort. Commercial journals have not shown any interest in donating text, however, and many nonprofit journals remain ambivalent. Although the plan would allow journals donating material to charge authors a fee, some editors and publishers worry that, if they do, they will drive away authors and ultimately lose revenues. Some, including The New England Journal of Medicine and Science, have published editorials criticizing NIH's plan as monopolistic and not adequately supportive of peer review (Science, 9 July, p. 197).

But Varmus believes the chilly reception does not reflect the views of rank-and-file scientists. In an interview with Science, he said: “Some major journals that have very little to lose here have been incredibly resistant,” although “we've had a productive discussion” about how to make scientific data more widely available. He added, “I have heard less from the scientific community than I am used to, because people in the trenches are used to having their opinions bubble up through their leaders, and [in this instance] some of their leaders have been a little resistant.” Varmus seems convinced that the best way to test the concept is to launch it.

“To focus the next step,” Varmus says, NIH will establish an Internet server called PubMed Central, which will be linked with the popular citation and abstract database, PubMed, run by the National Center for Biotechnology Information. “Any journal that wants to provide its content at any time after acceptance—and it doesn't have to be reviewed acceptance—can do so,” he says. PubMed Central will accept contributions long after publication, “even a year later.” The server will share its content on a daily basis with mirror sites in other countries. Varmus estimates the annual cost to NIH will be very low, $1 million to$3 million.

The only other group that so far has indicated it may be ready to join NIH in distributing reports electronically is the European Molecular Biology Organization (EMBO) in Heidelberg, Germany. Encouraged by EMBO's executive director, Frank Gannon, EMBO and its affiliates have been discussing plans to create a life sciences data center with about a dozen scientific groups. At a meeting in July, they drafted a set of principles for “E-bioscience,” a name they preferred to “E-biomed” because they envisioned a broader site, encompassing all life sciences.

Varmus says NIH has also broadened its proposal to embrace the nonmedical life sciences, a change initiated partly in response to comments from the community. In addition, to allay concerns that a government organization will have too much power over the content of PubMed Central, NIH is taking steps to minimize its editorial role: “Nothing that goes on the server will be determined by NIH,” he says. Instead, NIH will act as gatekeeper, accepting both peer-reviewed work and material that has only been “screened but not formally peer reviewed”—but only from approved groups. At the moment, NIH has some “very stringent” provisional rules limiting the kind of groups that may provide screened-only material. But Varmus is asking the National Academy of Sciences to create an international advisory body that would assume responsibility for deciding who should be allowed in the gate.

The academy could also help in another way—by contributing text from its own journal, the Proceedings of the National Academy of Sciences (PNAS). Nicholas Cozzarelli, editor of PNAS, strongly supports public sharing of PNAS's content and has asked for authority to send material to NIH's new data center. The academy's governing council considered the request last month and deferred to the publications committee, chaired by vision researcher Lubert Stryer of Stanford University. Stryer says that in general, “I favor PubMed Central.” His committee will take up the matter on 13 September.

If the academy decides to donate full-text material from PNAS, Varmus will have won an important symbolic battle in his effort to establish the credibility of PubMed Central.

2. CELL BIOLOGY

# Introducing Proteins Into the Body's Cells

1. Evelyn Strauss

To pharmaceutical chemists and basic researchers, proteins are a bit like protégés who never quite fulfill their potential. Despite their wealth of biochemical talents, they generally lack the one skill scientists need to put those talents to work: the ability to make their way through the fatty membrane that surrounds cells. One answer is to coax the target cells to make the protein themselves, by inserting the corresponding gene, but so far no one's figured out how to deliver nucleic acids efficiently to cells in animals or humans. Now researchers may have hit on a powerful strategy: fusing foreign proteins to a segment of another protein, derived from the AIDS virus, that has an unusual ability to cross cell membranes.

On page 1569, molecular biologist Steven Dowdy of Washington University School of Medicine in St. Louis and his colleagues report that such tagged molecules can infiltrate all the tissues of living mice. “This is an entirely novel and apparently powerful approach for introducing proteins into the brain and throughout the body,” says Raymond Bartus, a neuroscientist at Alkermes Inc. in Cambridge, Massachusetts.

If the method works with other proteins, it might be used to combat inherited diseases and other conditions caused by a malfunctioning or absent intracellular protein. Researchers might, for example, introduce a tumor suppressor gene into cancer cells to help stop their abnormal growth or to add back the enzyme that's defective in the hereditary neurodegenerative disease, Tay-Sachs disease. “It really is intriguing and unexpected … that you can get proteins so pervasively into cells,” says Bert Vogelstein, a cancer geneticist at Johns Hopkins University School of Medicine in Baltimore. Still, Bartus cautions, “a lot of the details have to be worked out, and it will take some time before [the method] is harnessed for therapy in humans.”

To devise the method, Dowdy and his colleagues exploited the 10-year-old discovery that an AIDS virus protein known as TAT (for trans-activating protein) enters cells without aids such as cell surface receptors. Researchers don't know how TAT does that, but in 1994, investigators at Biogen in Cambridge, Massachusetts, showed that it could ferry other proteins into cells. They chemically attached a bacterial enzyme called β-galactosidase to a large piece of TAT that included its “protein transduction domain” (PTD), a stretch of 11 amino acids that helps TAT traverse the cell membrane. When they injected the cross-linked protein into mice, they detected hints of its presence in several tissues. “[The method] was inefficient, but it did work,” Dowdy recalls. “We thought to ourselves, ‘This has tremendous merit’ and picked up the literature trail.”

To try to improve the efficiency, the group took what Dowdy calls a “biochemically blasphemous” approach. Unfolded, “denatured” proteins lose their activity. But reasoning that a partially unfolded protein would have more of its oily interior amino acids exposed and might therefore slide more easily through the lipid-rich cell membrane, the researchers denatured test proteins that carried the TAT PTD before incubating them with cultured cells. As the group reported last December in Nature Medicine, denatured PTD-containing proteins enter cells more efficiently than do the native versions. “Other molecules in the neighborhood don't go in, and nothing appears to leak out,” says Dowdy. But with the denatured protein and its attached PTD, “it's like the parting of the Red Sea. No one knows how it happens.”

The group has used this strategy to transport over 50 proteins ranging widely in size into a variety of human and mouse cell types in culture. Once inside, they regain their activity, presumably because they can access the cell's normal protein-folding machinery, says Dowdy. Now the team has extended the method to live animals.

Steven Schwarze, a postdoc in Dowdy's lab, engineered a protein that contains the PTD from the TAT protein attached to β-galactosidase. After partially unfolding the protein, the team injected it into the abdominal cavities of mice, while control animals got a version without the TAT sequence. Four hours later, the researchers found little or no detectable β-galactosidase in the tissues of the controls. But the protein joined to the TAT PTD showed up throughout every tissue they looked at—blood, spleen, liver, kidney, heart, lung, and even brain—and it had regained its enzymatic activity. “Not only do you know that the whole protein got in, but you know it refolded properly,” says Joan Brugge, a cell biologist at Harvard Medical School in Boston.

The technique should give basic researchers an extremely efficient way of introducing proteins into cultured cells to see how they affect cell function. And ultimately it might be used in treating human diseases as well. But as Bartus and others point out, there are potential pitfalls. The PTD could elicit an immune response or the method could produce other toxic effects, although no signs of problems have appeared yet. And the very efficiency of the method could cause trouble. “One important issue is that if there's a spill, the aerosol could be taken up by the lungs and then spread quickly in the body,” Brugge says. “So for experimental use, investigators have to be really careful.”

Dowdy says that it probably won't be possible to target proteins carrying the TAT PTD to particular cells, but the group has already begun to cope with the delivery system's promiscuity by designing proteins to act only in certain cellular environments. As scientists tune the basic scheme, they'll no doubt find many ways to help proteins reach their full potential.

3. SCIENTIFIC COMMUNITY

# DOE Polygraph Plan Draws Fire

1. David Malakoff

The Department of Energy (DOE) has moved a step closer to subjecting up to 5000 researchers and other employees at its three nuclear weapons laboratories to lie detector tests. The long-awaited proposal, published in the 18 August Federal Register, has triggered protests from opponents—including a petition by 165 Los Alamos scientists—who say that the devices aren't reliable and that testing could damage morale and recruiting efforts. While DOE has scheduled hearings on the plan, both sides say that expanded use of the polygraph seems to be an inevitable consequence of allegations that China has obtained secrets about the U.S. nuclear arsenal.

Critics “can write all the petitions they want, but [polygraphs are] coming,” says DOE counterintelligence chief Ed Curran. However, he notes that the current proposal is the product of negotiations with lab managers and staff and that DOE has no desire “to force testing down people's throats.” Chris Mechels, a retired Los Alamos computer scientist and vice president of the Los Alamos National Laboratory Employee Rights organization, also doubts the tests can be derailed. With Congress backing the plan, opponents can make “obligatory protestations, but I think it's a done deal,” he says.

DOE chief Bill Richardson outlined the polygraph plan in April as part of a suite of security measures designed to calm members of Congress alarmed by allegations of Chinese espionage at Los Alamos, Lawrence Livermore, and Sandia national laboratories (Science, 26 March, p. 1986). Before the plan is put in place, however, the agency must collect public comment on how it plans to treat the 20,000 nonfederal employees who make up most of the lab staffs. They work for the University of California, which operates the Los Alamos and Livermore labs, and for the Lockheed Martin Corp., which operates Sandia.

The Federal Register notice (www.access.gpo.gov/su_docs/fedreg/a990818c.html) fills in the blanks. It specifies eight groups of employees and job seekers who would be “eligible” for periodic testing, including those involved in counterintelligence work and research that requires access to classified data. Overall, more than 10,000 lab employees would fall into one of the groups, but DOE officials say most lab scientists, whose work is not classified, would not be included. The proposed rule notes that the exams are voluntary, but that those who decline them could face “consequences,” including loss of their security clearance and transfer to a less sensitive position. The plan also gives test-takers 48 hours notice, but does not allow a lawyer or witness to observe the questioning. DOE officials say examiners will be limited to asking four “yes or no” questions related to spying and sabotage. “We did everything we can do to give the advantage to the person taking the polygraph,” Curran adds.

In the notice, DOE “acknowledges that some individuals consider polygraph examination results to be generally unreliable,” but contends that there are “no scientific studies” that cast doubt on their value “as an investigative tool.” Indeed, the agency claims polygraph results “are superior to random interviews” but should not “constitute the sole basis for taking any action against an individual.” The notice also disputes critics who say testing could drive away researchers, and Curran told Science that less than half of those eligible will probably be tested.

Such assurances are small comfort to lab employees. Some are signing up to speak next month at a series of public hearings on the proposal, while others are signing petitions. The resisters include 165 members of Los Alamos's X-Division, which does top-secret work on nuclear weapons. “We are opposed to unwarranted blanket polygraphing of Q-cleared personnel,” the petitioners wrote to Richardson last week, referring to a security clearance that gives holders access to classified information on a “need to know” basis. DOE officials say such blanket exams are not under consideration.

At Livermore the alarm is being raised by the Society of Professional Scientists and Engineers, an employees organization. “If thousands of workers are tested, as DOE proposes, some will surely be falsely accused of lying, with devastating effects on their careers,” says computer scientist Patrick Weidhaas. It is “unthinkable,” he says, that a “research institute with top scientists is supposed to undergo testing using a machine that a lot of experimentalists would not want to have in their lab due to its lack of accuracy.”

4. SCIENTIFIC COMMUNITY

1. Jon Cohen

LA JOLLA, CALIFORNIA—Three years ago, the appointment of cell biologist Thomas Pollard to head the Salk Institute for Biological Studies here ended a long and tortuous search for a leader to help put this scientifically rich, but endowment poor, institution on a more solid financial footing. Pollard has done that—Salk's endowment more than doubled, to over $100 million, during his tenure—but the leadership vacuum returned last Monday, when Pollard announced that he would step down as the Salk's president next year. The reasons are murky, although trouble for Pollard was evident as early as February, when the institute split his job, leaving him as president but giving the CEO responsibilities to board chair Frederick Rentschler. “Being president is tough at a place like the Salk, where you have to administer it and raise money,” says Stephen Heinemann, chair of the academic council. Pollard, he says, “actually has accomplished a number of things,” but Heinemann stresses that “what makes Tom most enthusiastic is when he's doing his science.” Pollard himself says he wants to spend more time in his lab. “If there were 36 hours in every day, it would have been easier,” says Pollard, who plans to stay on as a Salk faculty member. The Salk has a stellar research faculty but lacks alumni, an attached medical school, or a famous president—all aids to fund raising. Under Pollard, however, the institute raised a record$25 million last year alone, and the stock market increased the value of its investments. Salk will soon receive another bolus of more than 16 million if a pending deal with Merck goes through. This would allow the pharmaceutical giant to buy a biotechnology company in which the Salk has major stock holdings, Sibia Neurosciences. Although Salk is in better financial health than ever before, supporting three endowed chairs for the first time, Heinemann and other Salk researchers say the institution would like a much larger endowment still, so that, ideally, all 56 faculty members would have endowed chairs. “A lot of faculty are struggling,” says Heinemann, who has one of the three current chairs. The Salk board recently brought in an independent consultant to review the organization and make recommendations for restructuring it. Faculty member Fred Gage and others note, however, that the review specifically did not address whether Pollard should stay on as president. Last week's decision opens a new period of uncertainty for Salk, which has had trouble finding a leader ever since Frederick de Hoffman, its head for 18 years, stepped down in 1988. The problem, time and again, has been finding a topflight researcher who is willing to devote enough time to fund raising and administrative issues—precisely the fix that the Salk is in once again. A six-member search committee has begun scouting for Pollard's successor. “Pollard played an important role,” says Heinemann. “Now we're going to look for a new type of leadership.” If past is prologue, expect the Salk to try and find a Nobel laureate who is winding down his or her lab activities. 5. PLANT GENETICS # A Paternity Case for Wine Lovers 1. Michael Hagmann In vino veritas—the Romans had it right. In the more than 5000 years since humans began making wine, plenty of secrets have tumbled from lips it has loosened. Now, wine grapes themselves are spilling some intimate secrets—about their own parentage. Using DNA fingerprinting techniques akin to those used to solve crimes and settle paternity suits, scientists at the University of California (UC), Davis, have shed some light on the fiercely disputed pedigree of a number of the world's most renowned grapevine varieties, or cultivars. As they report on page 1562, 18 varieties long grown in northeastern France—including Chardonnay, the “king of whites,” and reds such as Pinot and Gamay noir—prove to be close relatives. Indeed, 16 of them turned out to be the offspring of a single, highly prolific pair of parents: Pinot, the very epitome of a fine Burgundy, and, surprisingly, Gouais blanc, an obscure white variety that was widespread in the Middle Ages but was banned several times in France, most recently in the 1950s, due to the poor quality of its wine. James Luby, a fruit geneticist at the University of Minnesota, St. Paul, says the results “show the power of using genetic markers to clear up a conundrum that has been speculated about for decades, if not centuries.” What's more, the findings are likely to cause a stir in the wine community. “This is quite a shocker,” says grape geneticist Bruce Reisch of Cornell University's New York State Agricultural Experiment Station in Geneva. “No one would have imagined that all [these varieties] are from the same parents.” The viticultural detective story has its roots in the early 1990s when Mark Thomas, a grape geneticist at the Commonwealth Scientific and Industrial Research Organization in Adelaide, Australia, and his colleagues developed a system to distinguish grape cultivars based on their so-called microsatellites. These consist of simple, repetitive sequences of DNA that vary in length between unrelated individuals, creating a genetic “fingerprint” that can tie crime-scene evidence to suspects and establish paternity—in both humans and wine grapes. Thomas and others originally demonstrated the power of this technique by unraveling the parentage of Müller-Thurgau, the most widely cultivated white variety in Germany, developed around the turn of the century. Then, in 1997, plant geneticist Carole Meredith and her colleague John Bowers at UC Davis for the first time identified the parents of a traditional cultivar. They showed that Cabernet Sauvignon—undoubtedly the world's greatest and most successful red wine cultivar—is a progeny of two other classical Bordeaux varieties, Cabernet franc and Sauvignon blanc. Encouraged by their success, Bowers and Meredith, in collaboration with Jean-Michel Boursiquot of the Ecole Nationale Supérieure Agronomique and Patrice This of the Institut National de la Recherche Agronomique (INRA), both in Montpellier, France, decided to see if they could find the hidden family ties among 322 French grape varieties, some of which—such as Gouais blanc—are not even cultivated anymore and had to be retrieved from the INRA plant preservation collection. Aided by a computer program Bowers developed for spotting microsatellite patterns shared by the cultivars, the researchers were able to construct what amounted to a Burgundy family tree, tracing the relationships of 18 varieties from that region. Pinot and Gouais blanc appeared to be the likely founders of the Burgundian line—a conclusion buttressed by statistical analysis, which showed that a Pinot-Gouais lineage is at least 1012 times more likely than any other combination for all 16 progeny varieties. “This makes it a pretty safe bet for Pinot and Gouais blanc,” says Thomas. Because most, if not all, of the 16 siblings predate the times of deliberate grapevine breeding by hundreds of years, Meredith says the various crosses must have occurred spontaneously—and independently—by cross-pollination between Pinot and Gouais vines, most likely somewhere in northeastern France. For wine purists, especially in France where new hybrid grape varieties are legally excluded from bearing the prestigious designation, “Appellation d'Origine Controllée” (AOC), the findings might be rather disconcerting. Says Alain Bouquet, a grape breeder at INRA, “The AOC system from 1934 was based on the assumption that varieties obtained by crossing are inferior to the traditional varieties. I think this is erroneous in the case of crosses between old European varieties, as it is now proven that the two best varieties in the world, Cabernet Sauvignon and Chardonnay, are derived from such crosses.” Indeed, Bouquet and his colleagues are now performing their own Pinot-Gouais crosses to see if they can recreate the successes of the original, spontaneous crosses. He expects the first miniharvest in 4 to 5 years. “Grapevine selection is a very long and costly process,” he explains. The new results also show that the best grapes don't necessarily make the best parents. “It doesn't take two great varieties to produce a great progeny,” says Meredith. Or as Luby puts it, “Even a scruffy bull can sire good offspring.” 6. ASTRONOMY # Stellar Small Fry, or Runaway Planet? 1. Govert Schilling* 1. Govert Schilling is an astronomy writer in Utrecht, the Netherlands. Dark objects each the size of a dozen Jupiters could lurk in nearby space, a new discovery suggests. Maria Zapatero Osorio of the Canaries' Institute of Astrophysics in La Laguna, Tenerife, along with colleagues there and at the University of California, Berkeley, has found a mysterious object, dubbed S Ori 47, which defies easy classification: It may be too light to be a brown dwarf, the smallest kind of star, and could even be a giant planet drifting alone through space. “This is the lowest mass object [beyond our solar system] ever imaged by astronomers,” says Zapatero Osorio, who descibes the finding with her colleagues in a paper to appear in Astrophysical Journal Letters. S Ori 47 may be just 1.5% of the mass of the sun, or 15 times the mass of our own giant planet Jupiter. Whatever it is, it could be only one of many, as it is visible only because it is still glowing after its fiery birth. “There may be tens of them within 30 light-years from the sun,” says Zapatero Osorio. “It's a very important discovery,” says Kevin Luhman of the Harvard-Smithsonian Center for Astrophysics, who himself is hot on the trail of extremely low-mass objects. “It's discoveries like this that are developing an empirical picture of substellar objects.” But he doubts that the objects are plentiful enough to account for the galaxy's “dark matter,” the mysterious missing mass that seems to pull on the visible stars and gas. Zapatero Osorio and her colleagues found S Ori 47 when they were observing a young star cluster in the constellation Orion. The stars in the cluster, 1100 light-years away, all formed just a few million years ago, so S Ori 47 is still glowing with the heat generated as interstellar gas collapsed to form it. The team studied it in detail using an infrared camera mounted on a 1.5-meter telescope at Teide Observatory on Tenerife in the Canary Islands, measuring its luminosity (0.2% of that of the sun) and surface temperature (some 1700 degrees Celsius). They estimated its mass by plugging these measurements into theoretical models of how quickly objects of different masses should cool and fade after their formation. The models are uncertain, says Zapatero Osorio, so the mystery object could be anything from 10 to 20 Jupiter masses. That mass range straddles the dividing line between brown dwarfs and giant planets, which many astronomers put at about 13 Jupiter masses. Objects just above that mass—brown dwarfs—are not massive enough to ignite the hydrogen fusion furnace in their cores, but at some point during their lifetime they do burn deuterium (heavy hydrogen). However, anything less than 13 Jupiter masses is thought to be incapable of burning even deuterium and is considered a planet. Planets are also generally thought to form in the disk of material around a star, while stars can form directly out of a collapsing gas cloud. S Ori 47 may have formed as a solitary object, but it is equally likely to be an ejected planet. Last year, Susan Terebey of the Extrasolar Research Corp. in Pasadena, California, claimed to have discovered an ejected planet on a Hubble Space Telescope photo. Although “there has been much skepticism about this particular claim, there's no reason why it couldn't happen,” says James Liebert of the University of Arizona, Tucson. According to Liebert, the highly elliptical orbits that some extrasolar planets seem to follow around their parent stars can only be explained through the gravitational interactions of a third body in the system. “This can easily result in the ejection of a Jupiter-mass planet from the solar system altogether,” he says. Regardless of its true origins and nature—low-mass brown dwarf or rogue planet—S Ori 47 appears to be no astronomical oddity. “Currently, we're observing much fainter candidates in the same cluster,” says Zapatero Osorio. Because all cluster members are roughly the same age, the fainter ones are probably even less massive. If the cluster is typical for the galaxy at large, space could be heavily populated with such objects. But their true numbers may never be known, because most are expected to have faded to blackness. 7. NEUROSCIENCE # A Machine With a Fly's-Eye View 1. Sunny Bains* 1. Sunny Bains is a scientist and journalist based in the San Francisco Bay area. Watch a house fly dart through a kitchen, navigating around obstacles at top speed, and you may not be surprised to learn that this blight of homemakers is a favorite of neurobiologists. For more than 30 years, they have measured brain activity and movement in tethered living flies to learn how the tiny fly brain processes fast-changing visual information and turns it into flight commands. More recently, they have used computer simulations to try out their theories on how these processes work. Now, scientists have a new tool that may help them unravel the secrets of a fly's agility: an analog electronic circuit that models a key part of the fly's visual system and is built into a rudimentary robot so that it can interact with the real world. The robot fly was developed at the California Institute of Technology (Caltech) in Pasadena by Christof Koch and graduate student Reid Harrison, who says it may turn out to be a better probe of the fly visual system than experiments with live flies or computer simulations. “By building a model that interacts with the real world in real time, one can easily expose the model to complex stimuli that might be difficult to simulate,” says Harrison. The robot fly's eye may also benefit robotics, because its analog design is fast, very stable, and uses little power. “[These experiments] show that much can be learned from biology for robotics,” says fly vision expert Martin Egelhaaf of the University of Bielefeld in Germany. Harrison and Koch, whose results will soon be published in Neural Computation and Autonomous Robots, based their visual system on three layers of structure in the fly's eye. The top layer of photoreceptors each send a signal to a cell called an elementary motion detector, which detects motion by working in concert with neighboring EMDs. When a photodetector picks up a signal, its EMD compares that signal with what its neighbors are seeing. If a neighbor detects an identical but delayed signal, then the EMD decides that they are coming from a single source moving in that direction and sends a message to the next layer of cells. These are the large “horizontal system” (HS) cells, which collect the outputs from all the EMDs and sum them to produce a signal that drives the fly's motor system to react to the detected motion. The researchers duplicated this structure in a single integrated computer chip. Each electronic EMD consists of a light detector and a simple circuit to compare signals and generate an output signal—all covering an area 60 by 200 micrometers. The chip contains 144 such devices arranged in a 6-by-24 grid. Each EMD in the array sends its signals to another circuit—the equivalent of the HS cells—that sums the outputs to produce a response to the observed movement. To see if this vision system would successfully respond to real-world stimuli, the researchers mounted it on a wheeled Lego robot. The robot had two powered wheels with very different gear ratios, so it naturally ran in tight circles. The vision system would therefore see a lot of horizontal motion and produce a large signal. The team wanted to use the signal to get the robot to compensate for the motion and move in a straight line. So the Caltech team used the output from the vision system to boost the power driving the low-geared wheel and reduce the power driving the high-geared wheel. Hence the more the robot turned, the bigger the signal, causing the small wheel to run faster and the big wheel to run slower and hence slowing the turn. This optical feedback worked well, and, with the vision system installed, the robot moved in near-straight lines despite the huge asymmetry in its wheels. Insect vision specialists told Science that the success of the robot is an important validation of current theories of fly vision. Alexander Borst of the University of California, Berkeley, believes the way the Caltech vision array feeds straight into the motor system, and hence gets automatic feedback, mirrors biology well. “The big advantage that I see is that the visual system is automatically in a closed loop, which neuroscientists usually forget when thinking of the computational tasks of the visual system,” he says. The nature of the circuits may also hold lessons for roboticists. The vision system contains no digital electronics, which would involve a complex computational recipe containing thousands of operations, but instead uses analog circuits, processing light signals by employing the innate physics of the device—a technique pioneered by Caltech's Carver Mead. This makes its power consumption extremely low—just 5 microwatts for the array. Harrison compares this with the charge-coupled device imagers on the Sojourner rover used in the recent Mars Pathfinder mission. These used 0.75 watt just to acquire images, and much more power to process the data. The Caltech researchers are now continuing to refine their vision controller system, while at the same time giving their robot fly a smaller and more biologically inspired body that they hope might one day take flight. In particular, they are in the process of designing micromechanical “halteres”: vestigial wings that assist a real fly with attitude control and stabilization. 8. COMPUTER SCIENCE # Internet Security Code Is Cracked 1. Alexander Hellemans* 1. Alexander Hellemans is a writer in Naples, Italy. A popular encryption tool for keeping credit card numbers and other information secret on the Internet has been cracked. Last week, scientists announced at a press conference in Amsterdam that, using a sophisticated mathematical algorithm and cleverly written software, they had broken the RSA-155 code, which protects credit card transactions and secure e-mail in Europe. The number-crunching legerdemain suggests that anyone trafficking in confidential information on the Internet may soon have to switch to more sophisticated encryption software. Using RSA-155, one party can send a secure message to another by using the recipient's “public key”—a 155-digit product of two large prime numbers—to transform the original message into ciphertext. Decoding the message, however, requires the two prime numbers, known only to the recipient. For a long time this encryption was considered secure. Factoring a 155-digit number was thought to be beyond the scope of practical computations. Two years ago, however, a group led by Herman te Riele of the Centre for Mathematics and Computer Science (CWI) in Amsterdam succeeded in factoring a monster 180-digit number belonging to a special set, called Cunningham numbers, that are easier than ordinary numbers to factor. After improving the software and the algorithm—called Number Field Sieve—used for pinpointing likely prime numbers, Te Riele's team, including researchers from CWI and from Microsoft and Sun Microsystems, devoted 5 months on 300 personal computers and a Cray 916 supercomputer to finding the two prime factors of a 155-digit number. “Our aim was to show that in principle this can be done,” Te Riele says. For the moment, he says there's little reason for European users to worry that someone will snoop on their Internet credit card purchases—cracking the code still takes too much computing firepower and expertise. He figures it won't be long, however, before such code-cracking becomes common enough to threaten ordinary users. “The situation can become unsafe in 2 or 3 years,” he says. One of the inventors of the RSA code says he had already reconciled himself to someone breaching the code. “I and fellow cryptographers have been recommending for a long time that keys of that size are too short,” says Ronald Rivest, a cryptographer at the Massachusetts Institute of Technology. All Internet commerce, says Te Riele, may soon have to move to the more intractable codes—involving 232 digits—that are now standard in the United States or the even longer codes of 309 digits used for government and military transactions. At the current rate of progress, says Te Riele, even his group would not succeed at breaking such codes for at least another 25 years. 9. NEUROSCIENCE # Probing Alcoholism's 'Dark Side' 1. Robert F. Service NEW ORLEANS—Alcoholics, it seems, may drink not just to feel good, but also to avoid feeling bad. At least that was the message delivered by new results on the brain chemistry of rats presented at the American Chemical Society meeting here last week by neuroscientist George Koob of The Scripps Research Institute in La Jolla, California. Work by his team shows that in animals, brain levels of a neurotransmitter associated with stress responses skyrocket during withdrawal from alcohol. The finding buttresses a long-standing view that addicts take their drug of choice as a form of self-medication to prevent depression and stress. Enoch Gordis, head of the National Institute on Alcohol Abuse and Alcoholism in Bethesda, Maryland, calls the finding “very important.” Knowing that this stress neurotransmitter, a small protein called corticotropin-releasing factor (CRF), is involved in alcoholism may provide a new target, he says, for gene hunters who are looking for the genetic changes that make members of some families more susceptible to the condition than others. In addition, it could help pave the way for novel medications that combat the anxiety associated with alcohol withdrawal. Over 15% of people who try alcohol wind up becoming addicted. But treating the disease has proved difficult because biochemically alcoholism is anything but simple. Alcohol affects several families of neurotransmitters, initially stimulating the release of dopamine in the amygdala and the nucleus accumbens, the brain's reward centers. The levels drop to normal with continued drinking, however, and medicines that target these centers—modulating the effects of dopamine—“have been something of a disappointment” in treating alcoholism, says Koob. That suggests that other factors are involved in dependence, leading the team to explore what Koob calls “alcoholism's dark side”—the reinforcement that alcohol could provide by eliminating the withdrawal symptoms of anxiety and depression. In the mid-1990s, Koob and his colleagues—who include Floyd Bloom, editor-in-chief of Science—began focusing on CRF. The brain peptide works both through the pituitary gland and directly in the brain to trigger the release of hormones and other changes that spark arousal, vigilance, and mood changes. These responses are healthy when dealing with, say, an attacking predator, but over the long haul they can produce chronic anxiety—symptoms that are also linked to drug dependence. One study from Koob's group that looked at marijuana withdrawal showed that CRF levels jumped threefold when mice habituated to marijuana were injected with a compound that countered its effect (Science, 27 June 1997, p. 1967). Similarly, alcohol-dependent rats forced to go on the wagon show classic anxiety signs, such as avoiding unfamiliar places. And in one early study, when Koob and his colleagues injected CRF-blocking peptides into the rats' brains, they found that the stress responses fell dramatically. So in the current study, Koob and his colleagues decided to take a closer look at CRF. Bert Weiss and other Scripps team members started by implanting tiny tubes in the brains of rats to allow them to monitor the concentrations of the peptide in the animals' cerebrospinal fluid over a period of weeks. They found that to start with, the levels were similar in both normal controls and alcohol-dependent rats, but when the dependent rats were forced to go on the wagon, their CRF levels shot up 10-fold. The researchers are currently studying what happens when they start drinking again. Koob says his team is not sure why CRF levels increase during abstinence, but it may be because alcohol suppresses another brain neurotransmitter called glutamate, which in turn is thought to spark the release of CRF, and once that suppression is gone, CRF levels soar. But the rise in CRF apparently isn't the only brain neurotransmitter change underlying alcohol dependency. In another recent rat study, Koob and his colleagues showed that brain levels of the reward-inducing dopamine drop to half their normal amount during withdrawal and return to normal levels when alcohol-dependent rats drink. Together, the low dopamine and high CRF levels may provide a powerful stimulus to drink. “It's a double whammy,” says Koob. “When you're dependent, you're drinking to restore your brain's reward system to its normal balance.” That, he adds, could help explain why during their first year of treatment only about half of alcoholics are able to kick their drinking habits. 10. CANCER RESEARCH # New Type of Blood Vessel Found in Tumors 1. Marcia Barinaga An army's supply lines are often its weak point. Some researchers think that the same goes for cancer. By cutting off angiogenesis —the growth of the new blood vessels that tumors need to nourish themselves—they hope to block tumor growth. But in a new twist that may require some retooling of that approach, a research team at the University of Iowa College of Medicine in Iowa City has found that at least some aggressive cancers have developed an unexpected way of ensuring a blood supply: Rather than inducing normal blood vessel formation, they apparently make new blood supply channels themselves. New blood vessels are normally formed by a special type of cell called endothelial cells. But Iowa cancer biologist Mary Hendrix and her colleagues Andrew Maniotis and Robert Folberg report in the current issue of the American Journal of Pathology that highly malignant uveal melanomas, which develop in the eye, are full of networks of blood channels made by the melanoma cells themselves. The team's results suggest that this happens because melanoma cells, like other cancer cells, lose some of the characteristics of the tissue from which they were derived and acquire the ability to switch identities, turning on genes expressed by other cell types—in this case including key endothelial cell genes. Avraham Raz, a cancer researcher at the Karmonas Cancer Institute in Detroit, finds the new discovery “very exciting.” If the finding is correct, he adds, “then [the melanoma] cells don't have to rely on growth of endothelial cells, because they can [form blood vessels] by themselves.” The discovery may also help explain why the melanomas containing the channels tend to spread aggressively. To form tubes, cancer cells must be capable of burrowing through tissue, a trait also needed for metastasis. What's more, the blood vessels would provide the cells with direct access to the bloodstream. And if other types of cancer cells have similar capabilities, as preliminary work suggests, it would imply that tumors might be able to circumvent some antiangiogenesis drugs now under development, which are directed at endothelial cells. However, the work may also help researchers find genes needed for both types of vessel formation, which could provide new drug targets. The discovery arose from a collaboration between Hendrix, whose lab has studied the metastatic properties of melanoma cells for many years, Folberg, a pathologist who specializes in uveal melanomas, and Maniotis, who recently moved to Iowa from the lab of Harvard angiogenesis researcher Judah Folkman. Folberg had identified fine weblike patterns of blood vessels in the melanomas and found that patients whose tumors had those patterns were very likely to develop metastases and die. To find out why, Maniotis compared the blood vessel-forming capabilities of cells cultured from the aggressive and nonaggressive tumors, using an assay developed in Folkman's lab at Harvard. Maniotis wanted to see if the cells from aggressive tumors could induce endothelial cells in the cultures to form blood vessels. Instead, the researchers were “absolutely shocked,” Hendrix says, to find that aggressive melanoma cells could form channels even in the absence of the endothelial cells. Maniotis repeated the test with a variety of uveal melanoma cell lines, and under a variety of culture conditions. And always, says Hendrix, the cells “formed a pattern that completely recapitulated what was seen … in the patients.” On closer examination of tumors taken from patients, Folberg found that their vessels, like those formed in the culture dishes, were devoid of endothelial cells and instead were completely coated with melanoma cells, easily identifiable by their melanin-containing pigment bodies. To look for clues to how the melanoma cells had picked up this new skill, the team then collaborated with cancer geneticists Paul Meltzer and Jeff Trent, of the National Human Genome Research Institute in Bethesda, Maryland. Using microarrays that can survey the expression of thousands of genes simultaneously, they found that a host of genes not normally active in melanocytes, the skin cells that give rise to melanomas, are turned on in the tumor cells. “The highly aggressive cells … have become embryonic-like,” says Hendrix. “They are expressing genes that are characteristic of many different cell types.” And counted among those genes were many that are characteristic of endothelial cells and may contribute to those cells' ability to form blood vessels. Other researchers, such as tumor biologist Meenhard Herlyn of the Wistar Institute in Philadelphia, had already found that melanoma cells express some typical endothelial cell genes, but this new work goes further, Herlyn says, by showing that melanoma cells actually act like endothelial cells and form vascular channels. What's more, Hendrix says the team has found the channels in the more common skin-derived form of melanoma, and “we have preliminary evidence that this happens with aggressive prostate tumor cells [and] with aggressive glioblastoma,” a brain cancer. If further work confirms that the phenomenon is widespread, says Raz, “we may have to rethink our strategy of combating vessel formation.” For example, some current antiangiogenesis efforts may target characteristics of endothelial cells not shared by the tumor cells and so might miss blood vessels made from the actual cancer cells. But Folkman, whose lab at Harvard has pioneered antiangiogenesis drugs, notes that endothelial cell growth must occur at the sites where the tumor vessels hook up to the surrounding vessels, and blocking that growth might be enough to starve a cancer. What's more, he points out, some of the antiangiogenic treatments home in on target proteins that have been identified in the vessel-forming melanoma cells and so may work against those cells as well. Even if the drugs currently in development don't act against both types of vessel formation, the Iowa team's finding opens the door to identifying such common targets. Now, Herlyn says, the Iowa team can “sort out which genes are the most important ones” for blood vessel formation. So in time, researchers may be able to turn the devious adaptability of tumors back on themselves. 11. MILITARY RESEARCH # Pentagon Agency Thrives on In-Your-Face Science 1. David Malakoff The Defense Department's2 billion research agency does things differently. But its unorthodox management style seems to work, and it's winning converts

The exploding landmines were so close that Gary Settles could feel the shock waves as the blasts gouged meter-deep craters in the earth and sprayed deadly shrapnel into the air. The Pennsylvania State University, University Park, mechanical engineer, safely sheltered inside a concrete bunker on a Florida military base, was getting a reality check of sorts. He and other researchers had been brought to the base by their funder, the Defense Advanced Research Projects Agency (DARPA), to get a closeup look at the pernicious power of the explosives and to instill a sense of urgency to a $25 million effort to invent radically new mine detectors that mimic a dog's keen sense of smell. The experience was “certainly out of the ordinary,” says Settles, who usually can be found in his lab, studying gas and liquid flows with lasers and high-speed cameras. At DARPA, however, “out of the ordinary” is standard operating procedure. DARPA is not your typical federal research agency. Whereas the bigger—and more mainstream—agencies, like the$3.5 billion National Science Foundation (NSF) and the $16 billion National Institutes of Health (NIH), typically use peer-review panels to pick grant winners and then stay out of a researcher's way, the Pentagon's leading research funder takes a decidedly different approach. For 41 years, it has given a small group of program managers extensive power to direct high-risk, and sometimes wacky, research projects. Right now, for instance, the$2 billion agency is funding work on hopping robots that could scout battlefields, software that could instantly translate any Web page into English, and beetles that might be trained to home in on enemy ammunition dumps. “If you've got an idea that will revolutionize the world and doesn't violate too many of the laws of physics, we're listening” says DARPA's Larry Dubois, who manages the Defense Sciences Office, one of seven major divisions at the agency (www.darpa.mil).

Once sold on an idea, the agency likes to get in a scientist's face. “I wanted to show them exactly what we were up against,” says DARPA's Regina Dugan, explaining why she organized Settles's field trip. Dugan and other program managers also expect their researchers to attend team meetings, file monthly reports, and work cooperatively with other contractors. “It's a different culture; you just don't see this with NSF,” says Settles. The agency also manages its money differently. Most government science managers hand out grants that are open-ended and almost never rescinded, but DARPA writes contracts that call for deliverables and allow less promising work to be canceled easily. “It's our duty to kill off projects that aren't working,” says program manager Alan Rudolph.

Although some researchers grumble that DARPA's approach is heavy-handed, it has produced some spectacular results. The Internet, night-vision goggles, and radar-evading stealth aircraft all grew out of DARPA-funded science (see sidebars). Indeed, the agency's track record has so dazzled some policy-makers that they want to use its freewheeling approach as a model. A White House panel, for instance, recommended in April that NSF invigorate its computing research program by adopting DARPA's “strong manager” philosophy, and a few members of Congress recently proposed reorganizing the Department of Energy's troubled nuclear weapons research program into a DARPA-like independent agency. It's too soon to know whether those proposals will fly, but NIH officials have already begun testing whether DARPA-esque management methods, such as assembling interdisciplinary research teams and pushing them to share information, can produce breakthroughs in cancer-detection technologies.

Even insiders, however, say that DARPA's approach has its weaknesses, and that it may not be appropriate for other agencies. Congress, for instance, must decide “how many swings [an agency should be] allowed to take before making a hit like the Internet,” says Rudolph. And even when its projects do succeed, DARPA has had trouble moving findings into the military or the marketplace. In materials science, for example, the agency has “developed these interesting materials,” but often “they sit on the shelf,” says Steven Wax, another program manager.

Perhaps the biggest challenge facing DARPA and other agencies thinking of following in its footsteps is the difficulty of recruiting managers cut from the right cloth. “The DARPA model works best when the person handing out the awards is an intellectual peer of those receiving them,” says sociologist Ed Hackett of Arizona State University in Tucson, a former NSF program manager who has studied the use of peer review and other funding styles in federal agencies. “Recruiting those people [from academia or industry] is very difficult. Part of the reason [the DARPA approach] works is that it is done sparingly.”

## An antidote to groupthink

From the very beginning, DARPA was designed to be different. President Dwight Eisenhower created the agency in 1958 after an investigation following the Soviet Union's surprise launch of Sputnik blamed delays in the U.S. military satellite program on bureaucratic infighting and an unwillingness to take risks. Determined to prevent future lapses, Eisenhower ordered Pentagon planners to create an agency that, in the words of a DARPA-published history, would be “anathema” to the military R&D establishment and would recognize that “great leaps forward cannot be made by committee planning.” Instead, DARPA would rely on a corps of activist researchers to look beyond near-term military needs and fund areas with great potential to revolutionize war-fighting. Today, “the emphasis remains on searching for new ideas,” says DARPA director Frank Fernandez, who joined the agency last year after more than 20 years in the defense industry.

That search has taken DARPA in many directions. The agency has been a major source of funding for computer and software developers, and also invests heavily in materials science, microelectronics, and robotics. It has also made a mark in aeronautics, helping both the Navy and Air Force develop missiles and new aircraft. And 3 years ago, it launched a new biology program, hoping to attract top-notch scientists with ideas for defending against bioterrorism that may be too far-out for traditional NIH funding (Science, 7 February 1997, p. 744). The broadening of DARPA's traditional focus on the physical sciences and computer technology has already fostered unconventional strategies for rapid detection of infectious agents and even gene sequencing (Science, 11 June, p. 1754).

DARPA's spending targets are a source of some tension between the agency and its military customers, however. Whereas Pentagon planners often push for less risky research that will produce near-term payoffs, DARPA officials have jealously guarded their freedom to chase provocative ideas. The two cultures “don't think all that similarly,” says Fernandez. To bridge the gap, he recently told Congress, the military must “learn how to experiment,” while DARPA researchers must “learn the art of warfare.” In the meantime, Fernandez meets regularly with Pentagon brass to discuss the agency's priorities, which are also reviewed by several outside panels.

On the front lines of DARPA's work, however, are the agency's 125 program managers, who are recruited for their technical savvy and desire to leave their mark on a field. They are known within scientific circles by a panoply of nicknames that range from “idea scavengers” and “miracle hunters” to “eccentric” and “idiosyncratic.” Although such traits might be undesirable at most federal agencies, they match DARPA's recruiting rhetoric, which boasts that “the best program managers have always been freewheeling zealots.”

They are also zealots with flush portfolios. A DARPA program manager will typically spend up to $40 million or so on contracts to industry, academic, and government labs for one or more projects. Although managers face a variety of bureaucratic “reality checks” in the spending of funds, including regular reviews by DARPA brass, some become influential figures in their subfields, capable of nudging established research communities in a particular direction or creating collaborations where none existed before. But their influence usually doesn't last long: Managers stay for an average of 4 years, and each year they must fight for their piece of the budget. “It's an environment that rewards hustle and bureaucratic skills as much as real understanding of the technologies,” says historian Alex Roland of Duke University. Adding to the pressure is the fact that DARPA tries to complete up to 20% of its projects each year. “We don't do renewals,” warns Dubois, although some programs are reformulated to get a new lease on life. Still, “the opportunity to have a major, lasting impact in a short time is tremendous,” he says. When Dugan, a mechanical engineer, arrived 3 years ago from a Pentagon think tank in Washington, she knew that her new employer shared her growing interest in finding better ways to locate abandoned landmines, which pose an increasing threat to U.S. peacekeeping forces in the Balkans and elsewhere. Mines can be difficult to identify with current metal- detecting technologies. And a high rate of false signals means that mine clearers spend hours digging up metal odds and ends not related to mines. But there “was no promise I was going to be able to sell a program,” she recalls. After months of research, Dugan set her sights on what her bosses agreed was a promising approach—detectors that, like a trained dog, could sniff the chemical vapors produced by a buried mine's explosive charge rather than homing in on its metal parts. Such an electronic dog's nose would not only reduce the number of false starts but also help detect newer mines that have fewer metal parts. Before she could assemble and manage the interdisciplinary team needed to build a dog's nose, however, Dugan had to plunge into unknown intellectual waters ranging from biology to electronics. “In the beginning, there is a tremendous amount of information you have to absorb,” says the 38-year-old Dugan, who earned a doctorate at the California Institute of Technology in Pasadena before coming to Washington to work on a variety of defense-related issues, including the chemical signatures produced by nuclear missiles and the problems associated with detecting unexploded armaments. “I'm a fluid dynamics person, but I had to learn about olfaction in a hurry,” she says. The crash course paid off when DARPA agreed in 1997 to invest$25 million in Dugan's program over 3 years. Soon after, with the help of an advisory board, she drafted a request for proposals and selected the 13 academic and industry contractors for the research team. Settles, for instance, was contracted to document how dogs sniff without interrupting the flow of scent across their nasal membranes. (The answer: Their noses are designed to inhale fresh air from above and then exhale down and to the sides.) At the same time, neuroscientists were asked to apply to new detectors their insights into how animals can use just a few kinds of cells to differentiate among a wide range of odors (as the eye uses just a few cells to sense millions of colors). The team also included concepts not based on dog models, such as a California company's attempt to adapt an existing airport bomb-sensing system for field use.

Picking the team, however, was only part of what Dugan describes as “juggling 100 glass balls.” Another challenge was to unite everyone behind the goal of meeting DARPA's tight deadlines for producing prototypes. “It was pretty routine to see some resistance to [DARPA's] level of involvement” at the beginning, she says, particularly among university-based scientists who had never experienced the agency's hands-on management style. The process was aided by a series of semiannual team meetings, including the mine demonstration, some hands-on training with existing mine detectors, and a visit from a man who lost his legs in a mine explosion. “There was a lot of social engineering going on” during the events, Dugan says.

Dugan isn't ready to release details of the dog's nose project, now entering its home stretch, except to say that the team recently demonstrated “proof of principle” for one chemical-sensing system. And the most promising device may not mimic a dog's nose at all. Rather, the scaled-down airport detector produced by Quantum Magnetics of San Diego spots buried mines by beaming low-power radio waves into the soil and locking onto a unique signal produced by the explosive charge. But neither approach will move ahead toward a military use unless Dugan convinces one of the armed services to pick up the cost of continuing development. “We're in the throes of important negotiations,” she says.

Even if the Pentagon doesn't bite, however, Dugan's work may still pay off among companies involved in the project, which retain rights to their devices. Indeed, once their term is finished, DARPA program managers often find themselves back in private industry or academia working on the same problems.

As with Dugan's interest in mines, Rudolph's training as a zoologist led directly to the agency's $10 million, 3-year CBS project. Begun last year in a bid to harness the abilities of insects and other animals for military purposes, such as monitoring enemy positions, the project's nearly two dozen initiatives include exploring the practicality of training beetles, moths, and bees to home in on the chemical signature produced by landmines or chemical weapons plumes. Another idea probes the aerodynamics of flies with an eye toward developing microaircraft, while a third is studying the feasibility of creating electronic interfaces to bug brains. That could open the door to equipping the insects with interactive sensors, perhaps even “remote control” devices that direct the insects to crawl or fly in a particular direction. Rudolph, one of the few zoologists ever to work at DARPA, jokes that one of his aims in coming to the agency after nearly a decade at the Naval Research Laboratory was to create “smart bugs, controlled bugs, and robo bugs.” But on a more serious note, Rudolph also hopes to forge links between scientists and engineers working on animal locomotion or perception. “DARPA does seem to be more open to ideas that cut across disciplines,” says Michael Dickinson, a fly aerodynamics researcher at the University of California, Berkeley. Like several academic researchers new to DARPA's ways, Dickinson says he was “a little nervous at first” about his new backer. But he has come to appreciate its approach, he says, in particular its project meetings that give project engineers and life scientists a chance to “learn to speak the same language.” ## DARPA Lite Such dialogue is also one goal of the National Cancer Institute's (NCI's) new Unconventional Innovations Program, which has been dubbed “DARPA Lite” by some observers. Pushed by NCI chief Richard Klausner, it plans to spend$48 million over the next 5 years to turn high-risk studies into new technologies for early cancer detection.

In designing the pilot program, which will make its first awards this fall, “we tried to extract what we thought were the most valuable practices at DARPA, NASA,” and other agencies, says Carol Dahl, director of NCI's office of technology. Although NCI's traditional peer review panels will play a major role in selecting projects, she says agency managers will be “much more involved in program management than usual” and will prod researchers to share information. Among the program's high priorities is developing devices that can detect subtle molecular signals produced by growing cancers—such as the presence of telltale chemicals in the blood—and then transmit the information to external monitoring devices. Inventing such noninvasive sensors, a program announcement notes, “will require the input and collaboration of investigators … not traditionally engaged in cancer research.”

Although some NIH officials and outside scientists resist such directed research, NIH director Harold Varmus says he would like to see more of it. “I'm always asking my institute directors for more DARPA-like projects,” he says. Such a philosophy would mean taking the lead in developing fields such as bioengineering, he believes, rather than waiting for scientists to propose ideas.

Other agencies are watching the NIH experiment, but there are few signs of similar ventures popping up anytime soon. One limiting factor is finding enough DARPA-type program managers. Other agencies “would face some difficulty scaling up [the DARPA approach],” says Arizona State's Hackett. “You couldn't possibly afford 4 years away from the [lab] bench; you'd be dead,” says John Kauer, a DARPA-funded neuroscientist at Tufts University outside Boston, expressing a common sentiment in the community about a practice that, for example, is common at NSF but rare at NIH. And industry scientists often balk at the lower government salaries, a problem Congress tried to address last year by giving DARPA special authority to offer better salaries and benefits to up to 20 new hires. So far, DARPA officials have used the arrangement to reel in about a half-dozen prospects, and Fernandez says that new employment arrangements have made it easier to reassure academics and military personnel that there can be “life after DARPA.”

Despite the disadvantages, however, even Kauer says “DARPA would be a very interesting place to be.” Rudolph confesses that his stint, which runs for 2 more years, has been “an incredibly exciting time. While the demands are enormous and it can be draining personally, I would do it again.” The idea has even occurred to University of Illinois, Urbana-Champaign, electrical engineer Chang Liu, one of Rudolph's CBS researchers. Although the young academic says he wouldn't want to make a career move until he earns tenure, he's intrigued by the chance “to get the pulse of a particular field and orchestrate some innovative research. Mostly,” he adds, “it would be fun to play god.”

12. MILITARY RESEARCH

# DARPA's Highs and Lows

1. David Malakoff

The Defense Advanced Research Projects Agency (DARPA) may be part of the military, but its chief says the agency doesn't measure its performance in terms of clear-cut wins and losses. “Very rarely does anything at DARPA fail in the sense that we didn't learn something,” says director Frank Fernandez. “Our failures for the most part are that we fall short of our goals.” But observers say DARPA's work over 4 decades has included several notable hits and misses:

View this table:
13. MILITARY RESEARCH

# The Real Father of the Internet

1. David Malakoff

Hundreds of Defense Advanced Research Projects Agency (DARPA) managers have tried to put their stamp on an emerging field. But one towers above the rest: Joseph C. R. Licklider, the psychologist and computer scientist who in the 1960s launched what became the Internet. “Lick,” as he was known, epitomized the mix of playful imagination and down-to-earth management skills that current DARPA funders strive to emulate. “He helped set the standard for the proactive DARPA manager,” says historian Alex Roland of Duke University in Durham, North Carolina, who is working on a history of the agency's computing research program.

Licklider came to DARPA—then called ARPA—in 1962 after stints as a lecturer at Harvard University and at the nearby engineering company Bolt, Beranek and Newman. Fascinated by the social and technological implications of the new phenomenon of networked computer systems, he decided that DARPA needed to get on board. He accepted the task of directing the agency's information processing program only after being assured that he would be allowed to pursue his vision of interactive computing, spelled out in a now- famous 1960 paper on “man-machine symbiosis” (memex.org/licklider.html). “I just wanted to make it clear that I wasn't going to be running battle-planning missions,” Licklider told an Internet historian (www.columbia.edu/∼rh120) shortly before his death in 1990 at the age of 75. “I was going to be dealing with the engineering substratum that [would] make it possible to do that stuff.”

Once on the job, Licklider quickly reshaped the program to fit his vision. He canceled contracts with some companies and moved the money to selected university labs, which he believed were more innovative and more capable of building a community of interested scientists. The companies “were studying how to make improvements in the ways things were done already,” he recalled. “I was interested in a new way of doing things.” In the kind of whimsical wordplay that still marks DARPA program descriptions, Licklider said that he was trying to develop an “Intergalactic Network.” The phrase was later shortened to “Internet.”

Like today's DARPA managers, however, Licklider had to push a fractious group of researchers in a common direction. “I am hoping there will be … enough evident advantage in cooperative programming and operation to lead us to solve problems,” he wrote in a 1962 memo that urged his team to work together. Their efforts, and DARPA's investment in the hardware and software that allowed distant machines to link up, eventually produced e-mail and the Internet. By 1964, when Licklider left the agency (he would return for 2 more years in 1973), his views had become the compass for the agency's work in the field well into the 1970s. And his funding style—described by one academic as “Johnny Appleseed on a mission”—helped build top-notch computer science departments at many universities. “The significant advances in computer technology, especially in the systems part of computer science, were simply extrapolations of Licklider's vision,” says Robert Taylor, one of Licklider's successors at the agency.

Replicating Licklider's success, however, hasn't been easy. In the 1980s, for example, DARPA tripped over efforts to boost “artificial intelligence” (AI) systems that would help pilots fly complex jet fighters or admirals manage chaotic naval engagements. “The initiative failed to realize the grand vision of AI's pioneers,” says Roland, noting that vision isn't always enough. The technology underlying AI has yet to mature, say experts, and there is disagreement about whether it ever will. In contrast, the networking systems that Licklider helped foster have become ubiquitous and changed the way people live and work, enhancing his reputation as the DARPA program manager nonpareil. As Roland notes, “sometimes it also comes down to luck.”

14. 26TH INTERNATIONAL ETHOLOGICAL CONFERENCE

# Behaviorists Listen In as Animals Call and Croak

1. Pallava Bagla

BANGALORE, INDIA—Animal communication was a hot topic when 400 researchers from 39 nations gathered here at the 26th International Ethological Conference from 2 to 9 August. Some of the most vocal discussions explored how animals speak their minds, from calling in grasshoppers to croaking in frogs.

## Female Frogs Join the Chorus

Although it may sound harsh to human ears, the croaking of frogs on a rainy spring night is actually the sound of romance. Male frogs are trying to capture the attention and admiration of females with long, loud calls—displays of stamina that are thought to showcase a male's fitness. Now an Indian researcher has found that the females, often thought to be silent partners in this serenade, take part in the concert as well.

Although a few studies have noted female calls, most research on frog communication has concentrated on the louder males, so that many researchers believed that female frogs had no voice at all, says frog communication expert Stanley Rand of the Smithsonian Tropical Research Institute in Gamboa, Panama. But Debjani Roy, a herpetologist working at the Institute of Self Organizing Systems and Biophysics at North Eastern Hill University in Shillong, India, has found that for three species of Indian frogs, the evening concert is a duet, in which females respond to the males with feeble calls that indicate their readiness to mate. “Roy's discovery throws a very different light on mate acquisition in these species,” says Rand. “Females play a more active role and are no longer passive partners in a system dominated by males.”

Roy was doing a taxonomic survey of the frogs of northeastern India, analyzing frog calls as one way to distinguish the various species. Out of curiosity she and her colleagues analyzed both female and male behavior on their evening sojourns to ponds, where they often worked from 4:00 p.m. to 4:30 a.m., dodging rain, blood-sucking insects, and poisonous snakes. Roy first spotted the frogs visually using torches and kerosene lamps, which gave clues to their sex because females are much larger than males. Then she put microphones near the frogs to record their croaks, and watched and listened until they stopped calling sometime the next morning. Her long nights paid off: In three species—a tree frog, Polypedates leucomystax, a cricket frog, Limnonectes limnocharis, and the so-called skipper frog, Euphylyctis cyanophlyctis—she found that females croaked too, and the female call was “the catalyst” for the next step in courtship.

“The females respond to the male advertisement calls by producing feeble, low tonal calls to initiate courtship,” she explains. In the case of the tree frog, for example, the female emerged and sat near the male that had started calling first, which was always among the largest males. When the male stopped calling, the female gave feeble croaks, apparently indicating that she had made her choice and was ready to mate. The male then began calling again, even more loudly. Finally, they mated.

Roy and her colleagues confirmed this behavior in the lab: When female tree frogs heard tapes of various males played through different speakers, they called after hearing the largest, heaviest males, and they moved toward that speaker. “Females do not mate randomly, but they are intelligent enough to weigh their benefits by mating with males whose calls contain more acoustical energy,” explains Roy. Females are also sparing with their low-pitched, slow-paced calls; in a breeding group of females, Roy found that only those gravid with eggs gave the reciprocal call and were willing to mate. The rarity of female calls may be one reason why they weren't noted before. Although Roy recorded male calls on each of 148 nights of observation, she heard females' croaks on only 15 days.

Roy's discovery of a female call is “significant,” says Peter M. Narins, a behavioral ecologist at the University of California, Los Angeles, who himself has studied vocalization in P. leucomystax but had never recorded a female's reciprocal call. Rand adds that Roy's study suggests that frog communication and courtship is much more complicated than had been thought. Says Rand: “It now seems that males and females, at least in these few species, get into a conversation before they decide to mate.”

## Why Frogs and Insects Sing the Same Song

City dwellers seeking rural peace and quiet are often rudely surprised by the deafening croaks of a bullfrog chorus, not to mention the relentless drone of katydids and grasshoppers. Although wildly different in pitch, duration, and rhythm, all of these choruses have the same purpose: Males are passionately trying to attract mates. Now researchers studying the rhythm of male calls have noted another surprising similarity: Insects and frogs follow the same “rules” of calling. Females tune in to the male that leads the chorus, and males of each species have developed almost identical strategies for being the leader, behavioral ecologist Michael Greenfield reported at the meeting.

“What I find most remarkable is the similarity in mechanisms in both vertebrates and invertebrates,” says Greenfield, who is at the University of Kansas, Lawrence. Such parallels in organisms separated by hundreds of millions of years of evolution, he says, “suggest remarkable levels of convergence in the mechanisms of collective behavior.”

In both frogs and insects, males stay still and lure females with rhythmic calls. Females are most likely to respond to loud calls, which presumably indicate that the male is nearby and perhaps that he is vigorous. Researchers have also shown that when female frogs or grasshoppers hear calls of roughly similar loudness, they listen and orient only to the male that calls first, a so-called precedence effect. This may happen because it's easier for a female to locate the source of a signal if she tunes out any subsequent, competing calls, says Greenfield. From the male perspective, of course, this means that every male in the crowd wants to be first.

Now Greenfield and his colleague Stanley Rand of the Smithsonian Tropical Research Institute (STRI) in Gamboa, Panama, have shown that males in these very different species follow almost exactly the same path to that goal. When a male hears a neighbor's song, he postpones his next call until the neighbor stops, in what neuroethologists call an “inhibitory resetting” mechanism. The male restarts his own rhythmic calling slightly less than one period later, thus boosting his call into the lead. But in a crowd of croaking frogs or whining katydids, a male following this rule too closely would never sing. So theory suggests that males should pay attention to only a few of their noisy neighbors—those most likely to compete for nearby females—and ignore the rest.

The researchers tested the calling patterns of two species of acridid grasshoppers and one species of frog, lugging sophisticated audio equipment and four loudspeakers into the field to simulate combinations of competing callers. The experiments confirmed what theory predicted—males of all three species paid attention to only some of their neighbors. And all used the same so-called sliding threshold rule: Pay attention to the loudest neighbor and all those whose calls are within 6 to 8 decibels of that neighbor. “The mechanisms controlling female response and male signal timing parallel one another” in frogs and insects, says Greenfield.

The researchers discovered other subtleties in the calling rules of the Panamanian tungará frog when they studied it in STRI's Gamboa lab. These frogs modified the sliding threshold principle with a new rule: When several neighbors are all relatively loud, pay attention to the three loudest. But when the nearest neighbor is much louder than any others, pay attention to two neighbors, even if the weaker one is below the sliding threshold. Paying attention to no more than three neighbors may help prevent frogs from inhibiting their calls too frequently in crowded or noisy environments, says Greenfield.

Because the ancestors of frogs and insects split apart more than 540 million years ago, the profound similarity in their calls and the mechanisms used to generate them are probably not due to shared history. Instead, suggests Greenfield, “in the physical world, only a limited number of options are available to solve the problem of signaling in a noisy environment. In this case, evolution in both frogs and insects may have found the same solution.”

That's a plausible notion, agrees Peter M. Narins, a behavioral ecologist and frog communication expert at the University of California, Los Angeles. “It is the living environment that is providing the selection pressure … [so] similar solutions have emerged”—giving frogs and insects the same song.

15. PLANT BIOTECHNOLOGY

# Australian Center Develops Tools for Developing World

1. Elizabeth Finkel*
1. Elizabeth Finkel writes from Melbourne, Australia.

Richard Jefferson runs an institute in Canberra that helps plant scientists and farmers apply the latest technologies to overcome local problems

MELBOURNE—Richard Jefferson advocates grassroots genetic engineering for agriculture. He says that the approach taken by most universities and companies—producing single gene fixes tailored to model systems—is virtually irrelevant to the complex, diverse systems of the developing world. Instead, the 43-year-old molecular biologist argues passionately, what farmers need are the knowledge and tools to develop and disseminate their own strains tailored to local conditions and practices. His Canberra-based institute, CAMBIA—the Spanish word for change and an acronym for the Center for the Application of Molecular Biology to International Agriculture—is dedicated to that goal. And a lot of people are rooting for him.

CAMBIA, founded in 1992 with rhizobial molecular geneticist Kate Wilson, is intended to be a genetic workshop for the developing world. It's also a clearinghouse for intellectual property issues. With support from the Rockefeller Foundation and smaller grants, Jefferson has already helped Chinese scientists develop a new strain of long-lived rice, and he and his colleagues are testing a technique for rapidly generating and screening genetic variants that will thrive in local conditions. Last month these efforts received a major boost when CAMBIA was chosen as the biotechnology arm of the Institute for International Tropical Agriculture (IITA), the Nigeria-based research center that is part of the Consultative Group on International Agricultural Research (CGIAR).

Many plant scientists think this recognition is long overdue. “We all want to plow back our inventions to help developing countries, but … CAMBIA's focus of developing new tools is unique, and I'm very pleased it's being recognized,” says Roger Beachy, director of the Donald Danforth Plant Science Center in St. Louis, Missouri. “While we might train a student how to diagnose a specific disease in cassava, Richard is trying to develop brand-new technologies.” At the same time, they note that some of those technologies have yet to be tested. “They're either completely wrong or three jumps ahead,” say Peter Raven, director of the Missouri Botanical Garden in St. Louis, about one of CAMBIA's projects.

Born in the United States, Jefferson has an impressive record of crafting innovative tools for improving crops. In 1985, as a graduate student at the University of Colorado, Jefferson developed a technique that monitors the activity of transgenes by tagging them with the bacterial enzyme β-glucuronidase (GUS). Because GUS activity is easily detected by a color-producing enzyme assay, researchers use the assay to follow the activity of the transgene. The original paper is the most widely cited in the plant science literature. A few years later, at the Plant Breeding Institute in Cambridge, U.K., he used GUS as a stethoscope to monitor transgenes in field trials of the world's first genetically engineered food crop, a potato. Jefferson quotes Henry Ford, another man who changed the face of an industry, in describing the depth of his commitment to CAMBIA, which has relied heavily on self-funding from GUS royalties. “Obstacles are those things you see when you take your eyes off the goal,” he quips.

One of CAMBIA's current projects involves a new screening method for identifying transgenic plants through positive selection. The technique gives plants an exclusive energy source—in this case, cellobiuronic acid, a substrate that GUS breaks down to release glucose—and avoids the stress caused by the current approach of exposing plants to an antibiotic that only the transgenics are able to withstand. By using a novel secreted form of GUS (Bacillus OZ GUS, or BOGUS) that can metabolize cellobiuronic acid outside the cell, the technique offers the prospect of producing healthier, more genetically stable plants and eliminates concern about producing plants with antibiotic resistance.

A second project focuses on apomixis, the parthenogenetic reproduction of a plant from its seed. “Apomixis, if it's done right, could do more to revolutionize and democratize agriculture than any technology since the dawn of plant breeding,” says Jefferson. The technique would allow the small breeder to lock in the vigor of new genetic combinations, currently possible only by buying expensive hybrid seed or through labor intensive breeding protocols. It would also benefit crops like cassava or potato, by purging the pathogen load they aquire during their vegetative propagation. CAMBIA is collaborating with its partner institute, IITA, and two other CGIAR centers to generate this trait using CAMBIA's new technologies.

Scientists in the developing world say that Jefferson's tools are just what they need. Qifa Zhang, a professor at Huazhong Agricultural University in Wuhan, China, has just developed a new transgenic rice strain carrying an antisenescence gene, which prolongs the grain-filling period of the plant. Results from the first test crop show it has increased productivity by 40%. “We and a lot of breeders have been helped by CAMBIA,” says Zhang, a member of Rockefeller's rice biotechnology network, whose student spent several months at CAMBIA constructing vectors and establishing the transformed rice lines.

Zhang is also excited about using the next generation of CAMBIA tools. One is TransGenomics, a strategy to reawaken what Jefferson and his colleague Andrzej Kilian believe is an untapped “Jurassic Park” of diversity lurking in the plant genome. TransGenomics is based on recent findings that suggest that new plant species evolve primarily by deploying relatively unchanged proteins in new functions rather than by making major changes to those proteins. Recent work by plant geneticist John Doebley and others at the University of Minnesota, St. Paul (Nature, 18 March, p. 236), for example, traces the difference between maize and teosinte, its wild grass ancestor with branches rather than a single spike, to changes in the regulation of the gene controlling this trait.

TransGenomics takes a page out of evolution's own handbook in attempting to reorchestrate the activity of genes. Current techniques used in functional genomics programs disrupt gene function, affect all tissues, and fix the final mutations. Typically, these screens have not produced viable plants with useful new traits. TransGenomics is designed to kick-start the evolutionary engine by adding rather than losing functions, confining the changes to specific tissue, and permitting modifications. The screen relies on peppering the plant genome with different combinations of a gene trigger, gal4, and its target, the UAS sequence, and watching as the combination of trigger and target fires up the activity of a new gene. Kilian hopes within 2 years that breeders around the world will be using CAMBIA's transgenomic rice seeds and creating their own crops with its tool kit.

Many in the research community don't quite know what to make of the new technique. “I don't know if it will work or not,” says Eric Kueneman, special adviser for agriculture at the U.N. Food and Agriculture Organization. “But it takes the heat off using transgenic plants [because new traits are generated from the plant's own repertoire rather than from foreign genes].” IITA director Lukas Brader hopes the technique will aid African crops that have stubbornly resisted improvement. “Together we should be able to do something really good,” he predicts.

However, future plant breeders and agroscientists will need more than science to succeed. They also must compete in a world where intellectual property (IP) issues can block the application of socially useful research. “Just about everybody in public institutes has been incredibly naïve about IP rights,” says Gary Toenniessen, director of Rockefeller's rice biotechnology program. “It's been a shock to us to realize that you cannot use the results of research you funded because almost everybody's product is tied up in IP [disagreements].”

That's why CAMBIA is developing a Web-based resource tool for IP issues, with support from Rockefeller, CGIAR, and the International Food Policy Research Institute, a CGIAR center based in Washington, D.C. “The resource should be of value to everyone from graduate students to agribusiness giants like DuPont, who can only gain from dealing with more sophisticated partners,” explains Carol Nottenburg, an immunologist turned patent attorney who Jefferson recruited from private practice in Seattle to head the project.

Interactive software will allow users to address each step of the process, from ascertaining existing IP claims to formulating strategies to navigate around them—with CAMBIA's own patented technologies as bargaining chips. Zhang welcomes this resource as he contemplates how to develop his new rice strains. “We have a lot of difficulty in dealing with documents, terms, rights. Basically we don't know how to negotiate,” he confesses.

Although CAMBIA's freedom from institutional restrictions shortens its response time, some wonder how the largely self-funded operation will muster the resources it needs to meet its ambitious goals. “It's somewhat distressing that, [despite Jefferson's] vision, it's been relatively difficult,” says Beachy. “It's still a small operation.”

But plenty of people hope CAMBIA beats the odds. “I told Richard years ago it was an absolutely brilliant idea that would never work,” says Ed Rybicki, a professor of microbiology at the University of Cape Town in South Africa. “But I'm impressed [with how far he's come].”

16. POSTDOC ACTIVISM

# Getting to the Front of the Bus

1. Dan Ferber*
1. Dan Ferber is a writer in Urbana, Illinois.

Postdoctoral associations are sprouting throughout North America, giving voice to a population that most universities have ignored

Eliene Augenbraun had heard the stories. Since coming to Johns Hopkins Medical Center in 1992 as a postdoctoral fellow in cell biology, she'd heard about postdocs too scared of their hard-driving advisers to take a vacation or even a day off, of foreign postdocs paid $8000 a year because they feared that asking for a raise might cause a vindictive professor to decide to withdraw the sponsorship needed to retain their work visa. She'd heard about—and even knew firsthand—postdocs who had been assaulted after parking blocks away in a rough Baltimore neighborhood because they were denied university-subsidized parking at secure garages. But because they were dependent on their advisers to recommend them for the scarce faculty jobs they coveted, “people complained quietly and in hushed tones,” Augenbraun says. She even has a story of her own (see sidebar on p. 1516). Before long, Levi Watkins had heard the stories, too. As a professor of cardiac surgery and associate dean for postdoctoral programs at Johns Hopkins University Medical School, he began to meet regularly with Augenbraun and other postdocs from the new Johns Hopkins Postdoctoral Association. Although “cordial but noncommittal” at first, Augenbraun says, the 55-year-old Watkins soon felt a bond with these young postdocs. He had never forgotten his early days in segregated Montgomery, Alabama, where his family had listened to sermons by the Reverend Martin Luther King Jr. and where he had skipped school to join King's freedom marches. Or his college days at Tennessee State University, where he led marches protesting Nashville's racial conditions. Or his experience as the first black medical student ever at Vanderbilt University Medical Center. When the postdocs told him that their low pay, poor benefits, and invisibility within the institution made them feel as if they were sitting in the back of the academic bus, he says, “that touched me. I knew what they were talking about, and the analogy was a good one.” Despite their feelings of isolation, the Hopkins postdocs were not alone. Since 1992, disaffected postdocs have banded together at more than 15 research institutions and national laboratories across North America. At most of these institutions, a small core of committed activists has pushed hard for institutional changes, including improved salaries, benefits, career training, and protection against exploitation. At some universities, the postdocs have found institutional allies like Watkins. At a handful, deans have taken it upon themselves to push for better treatment of postdocs. But at most other universities, change has come slowly, if at all. That slow pace has also led postdocs to call for help from the government agencies that fund the universities where they work, although the National Institutes of Health (NIH), the largest supporter of postdocs, appears loath to dictate policies to the private sector. The causes of postdoc activism aren't hard to find: a tight job market, longer tenure in low-paying and temporary positions, and an uncertain status on campus. A survey that tracked 1980s Ph.D.s in biochemistry, for instance, found that 86% did postdocs and 40% did two or more. And National Science Foundation figures show that the average tenure for a postdoc in the 1990s is 45% longer than in the 1960s (see p. 1517). In one of the most comprehensive surveys of postdoc attitudes, covering 1322 Canadian postdocs from all disciplines, respondents said that they generally felt valued by their research group but not by their institution, and were unhappy with their salaries and career prospects. The survey also found that job satisfaction, including a sense of receiving proper credit for their work, declined the longer they remained postdocs. Universities have not been blind to these developments. Last year the Association of American Universities (AAU), whose 62 members employ most of the roughly 38,000 postdocs in the United States, proposed a series of steps that universities should take to address the issue. “At some universities, PIs [principal investigators] have been able to bring in postdocs, pay them what they will on whatever basis they choose, and keep them as long as they want—with no benefits and no grievance procedure,” says Steven Sample, president of the University of Southern California in Los Angeles and chair of the AAU committee that wrote the report. The report recommends that universities assign responsibility for postdoctoral fellows to one administrator, who would ensure health care coverage for postdocs and set a limit on the duration of a postdoc, along with issuing standard policies for postdoc stipends, benefits, workers' compensation, grievance procedures, and misconduct. This summer AAU is surveying its members about the status of their efforts and plans to review the results at a meeting this fall. Many faculty members won't be happy with these sorts of changes, predicts Rama Kasturi, a former postdoc and temporary assistant dean at the University of Cincinnati Medical School. Kasturi ran into faculty resistance while establishing a postdoctoral scholars program there in 1998: “Some department chairs do not wish to yield one inch of control to anyone on any subject,” she says. But Sample says university administrators need to stick to their guns: “When that memorandum comes from the provost, it might give heartburn to some people. But they'll get over it.” The Johns Hopkins postdocs have had more success than most in bringing about changes. By 1994 the university had formally recognized the association and approved minimum salary guidelines, which recommended—but did not mandate—that postdocs receive at least what NIH postdoctoral fellows and trainees get. The postdocs also gained their first representative on the committee that set the school's postdoc policy. In June 1999 the university adopted sweeping changes in postdoctoral policy, including closer oversight of departmental policies and minimum salaries for postdocs that are pegged to the NIH pay scale, although salary increases will be phased in over 3 years. Anyone who stays longer than 6 years must be hired as an employee with full benefits. All postdocs must have health insurance; if their grant or fellowship won't pay for it, the department must. Each department is required to form a committee that will meet annually to formally evaluate each postdoc's work, and there are plans for a new career-counseling center. Although the postdocs are still pushing for dental insurance, lower parking fees, and the same health plan as employees receive, Watkins says he feels that “we're well on our way.” And the postdocs say he deserves some of the credit for their progress. “He basically trained us in civil rights activism, and that made us so much more effective,” Augenbraun says. Some universities aren't waiting to be pushed into treating their postdocs better. In 1997, the University of Pennsylvania School of Medicine established an office of postdoctoral programs, which activists cite as a model. Mentors are required to send each new postdoc a detailed letter of appointment that specifies duties, salary, benefits, and term of appointment, says Trevor Penning, who directs the office. Stipend levels must match or exceed NIH fellowships, and postdocs are also entitled to 6 weeks of paid parental leave. Penning's office also runs monthly open roundtables for postdocs and sponsors workshops on such career skills as preparing for a job search and good interview techniques. Last year the University of California, Berkeley, which has established similar policies, hired a former Berkeley grad student to help organize a postdoctoral association. And the school is prepared to go to bat for what is needed. “You really need a dedicated high-level administrator [to back them up],” says chemist and graduate school dean Joseph Cerny, who oversees the office. Postdocs at other schools are still looking for that level of commitment from administrators. A group founded in 1995 at the University of California, San Francisco (UCSF), has prodded the school into establishing its first career center for postdocs. The university has also formalized its grievance process for postdocs, brought postdoc representatives onto committees that set postdoctoral policy, established an annual orientation for postdocs, and negotiated a high-quality group health insurance plan. But co-founder Patricia Bresnahan, now a postdoc at the Gladstone Institute of Virology and Immunology in San Francisco, says that a recommended minimum salary is a guideline, not a requirement, that incoming postdocs are not informed about the grievance policy, and that some postdocs pay for the new health insurance policy out of their own pocket. Still, she acknowledges that there has been progress. Adds postdoc Sharon Stranford, past president of the group, “at UCSF we have it better than most places.” One of those places, say postdocs there, may be the University of Utah, Salt Lake City. Joni Seeling, who helped found a postdoctoral association there in 1997, points out that Utah postdocs have no minimum salary scale, no time limit on appointments, no university-sponsored training in career skills, no formal evaluations, and no official grievance procedures. Some postdocs also have to pay for their own benefits, she adds. “What you realize is that you're here at the whim of the PI,” Seeling says. “If they decide they don't like you anymore, they can fire you with no notice.” John McCullough, associate vice president for academic affairs at Utah, agrees that most of the postdocs' complaints are legitimate, although he says that university policy does require written contracts for postdocs. “If they don't [have them], I'd like to hear about it,” he says. The university is “reconsidering some parts” of that policy, so that “postdocs have the rights they deserve … and everyone knows the rules of the game.” And although McCullough says that “we would be very happy to listen to a group of postdocs,” Seeling says that postdocs want formal representation on the committee that sets policies that affect them. “We're looking for rights,” she says. “We want to be acknowledged as part of the university.” Many postdoc groups would like help from NIH and other federal agencies in their push for recognition. In particular, they believe that supporting more postdocs on fellowships and traineeships, and fewer on research grants, would improve conditions by reducing the number of principal investigators tempted to exploit the labors of their postdocs for their own scientific advancement. “You have a PI whose career depends on the labor force in the lab, and yet he is also designated to look out for the postdoc's interests [as a mentor]. That can represent a conflict of interest,” says Bresnahan. NIH should also require PIs to follow clear and binding training mandates as a condition for receiving NIH funds to pay for postdocs, argues Cincinnati's Kasturi. “NIH has absolutely abdicated its responsibility on this subject,” she says. “Basically, they just hand out this money, and they ask nothing in return except that the research be published.” Not so, says Wendy Baldwin, NIH's director for extramural research. NIH expects universities with training grants to offer postdocs educational and career experiences as well as the opportunity to conduct research, she says, but it's very hard to separate the two: “It's hard to learn about research without doing research.” She also recognizes the temptation for a PI to ignore the training component: “If I need a technician, but I can get a postdoc for the same price, someone who is highly motivated and eager to learn, then it's hard to say no.” Although the current system has flaws, Baldwin says, she would not want NIH to lay down rules about how universities should treat postdocs. “I'm not convinced that we should be setting internal hiring policies [at universities],” she says. “I think we have to be very careful about how heavy-handed the government is about what you can do and what you can't do. … There are labor force issues that are not NIH issues.” One issue that clearly falls within NIH's control is its treatment of the 2230 postdocs—half visiting fellows from other countries—on its Bethesda, Maryland, campus. NIH began to look more closely at their situation 5 years ago, says Michael Gottesman, head of NIH's$1.5 billion intramural research program, “when we were trying to figure out how to run NIH on a steady-state budget.” Although recent healthy increases have eased that worry, he says, NIH director Harold Varmus and others still saw a need to help postdocs “gain their independence.” Toward that end, NIH created $1000 awards for postdocs to present their work at meetings and an internal committee to hold workshops on career-related skills. Several institutes also offer postdocs a chance to compete for funding that they can take with them to their first “real” job. And this spring, NIH issued its first guide on training and mentoring, which acknowledges that postdocs and graduate students “could benefit from a more explicit set of expectations.” But given the reluctance of federal officials to intervene in the affairs of their constituents, the task of reshaping U.S. postdoctoral training is likely to rest with postdocs, faculty, and research administrators themselves. At Johns Hopkins, Watkins says he is beginning to get calls from deans at other universities who have heard about how his institution has improved its culture for postdoctoral education. “It has been very important to me personally and to the university,” he says, “that the postdocs are full-fledged members of our medical family—that they are, in fact, on the front of the bus.” 17. POSTDOC ACTIVISM # What Postdocs Want--and How 1. Dan Ferber* 1. Dan Ferber is a writer in Urbana, Illinois. Postdoc activists are no longer agitating for changes just within their own universities—they are pushing for reforms nationwide. In October 1998, postdocs from nine universities, research institutes, and national laboratories met in Palm Springs, California, at the annual conference of the Group on Graduate Research, Education, and Training, a subgroup of the Association of American Medical Colleges that includes faculty and deans from major North American medical schools. The meeting, informally dubbed the “postdoc summit,” produced “a laundry list of what it takes to treat postdocs well,” says Victoria McGovern, a program officer at the Burroughs Wellcome Fund in Research Triangle Park, North Carolina, which helped fund the meeting. Their recommendations include: Written contract. Agreed upon in writing by postdoc, supervisor, and host institution, the contract would ensure that principal investigators and universities comply with federal laws on family leave, harassment, and discrimination based on race, gender, age, or disability; that postdoc performance be evaluated annually and in writing; and that universities provide formal grievance procedures for postdocs. Uniform job title and benefits. Regardless of source of funding or academic department, all postdocs at a given institution should have the same job title, and that title should not be shared by non-Ph.D. technicians or graduate students. Postdoctoral associations. Universities should support either a postdoctoral association or central postdoctoral office, which would conduct a survey to gauge the needs of postdocs, offer an orientation and a manual for new postdocs, and provide courses on science survival skills such as writing grants and research papers and landing a job. ## Postdoc representation on institutional policy-making committees. National postdoctoral organization. Funded by contributions from universities that train postdocs or federal agencies, the organization would offer a postdoctoral voice in national science policy debates, support fair labor practices and salaries and benefits commensurate with their Ph.D., keep postdocs abreast of job trends, and help them identify funding sources for fellowships and research. The postdocs have enlarged their network since then, communicating online through the Association of Science Professionals, says Patricia Bresnahan of the Gladstone Institute of Virology and Immunology in San Francisco, who organized last year's meeting. They will meet again in October to explore creation of a national association. “We got postdocs on the agenda,” Bresnahan says proudly. 18. POSTDOC ACTIVISM # Irreconcilable Differences 1. Dan Ferber* 1. Dan Ferber is a writer in Urbana, Illinois. Tension between a trainee and an adviser is not limited to science. But postdocs with grievances may be especially powerless because of their ambiguous status—neither student nor staff—and the tremendous power that faculty members hold over them. The problem is exacerbated by academic cronyism, postdocs say: Few faculty members are willing to openly criticize their peers, and institutional grievance procedures offer postdocs little protection from retribution. At the same time, such conflicts often serve as a rallying point for efforts to organize postdoctoral associations on campus by highlighting the flaws in existing policies. Here are brief descriptions of two conflicts that helped push two postdocs out of science. ELIENE AUGENBRAUN was a leader in organizing the Johns Hopkins Postdoctoral Association. But she faced a deteriorating relationship with her adviser, cell biologist Ann Hubbard. Feeling powerless to mend the relationship, Augenbraun attempted to move to the lab of one of Hubbard's collaborators and to transfer her National Institutes of Health (NIH) fellowship. Hubbard blocked the move, Augenbraun says, and tried to foil her attempt to take a leave of absence from the fellowship by demanding that she instead abandon the award and promise never to ask for NIH funding again. “She was being a bully, and no one would stand up to her,” Augenbraun says. Hubbard denies trying to block Augenbraun from retaining her fellowship. But she does recall her reaction to Augenbraun's proposed move. “I said to [my collaborator] that it would create a difficult situation … but that it was his decision to make.” Augenbraun didn't change labs. She also didn't file a formal grievance, before a grievance committee consisting mostly of faculty members that would recommend action to the dean, although some officials urged her to do so. One reason was that only one member of the medical school faculty whom she asked would agree to testify on her behalf. She also felt that any favorable decision would have been a Pyrrhic victory, as she would still be working in an area in which Hubbard was a major player. Instead, Augenbraun abandoned plans to become an academic researcher and in 1995 obtained a science policy fellowship to work at the U.S. Agency for International Development. She now runs a company called ScienCentral that produces science news for television and the World Wide Web. Cancer immunobiologist CHERYL LOVE-SCHIMENTI, a former postdoc at the University of California, San Francisco (UCSF), who worked at the San Francisco Veterans Administration hospital, says that conflicts with her adviser, Daniel Bikle, and the university's inability to resolve them fueled her disillusionment with research. In particular, she says, Bikle was angry at her for a series of injuries and health problems she suffered that kept her from work, a situation that generated a hostile attitude in the lab. “He used to tell me that he didn't know how on Earth I had earned a Ph.D. I'd go into every group meeting nervous and with a sick stomach,” she says. Bikle says he “doesn't remember ever specifically saying [those words].” He says her discomfort stems from a different source—a failure to do her job. Love-Schimenti's performance declined after a successful first year in 1994, he says, causing her to “feel humiliated” when presenting her results to the lab group. “She wasn't performing,” he says. Love-Schimenti says a series of triumphs in early 1996, including a 3-year,$146,000 grant from the Department of Defense, an invitation to speak at a breast cancer meeting, and publication of one of her papers helped the relationship for a short time but that another injury caused her to miss work again and revived the antagonism. As the relationship worsened, Love-Schimenti took her concerns to a higher level, and the issue was aired at an informal meeting in May 1997 involving her, Bikle, and two UCSF officials. But although she says that Bikle was told to stop his “harassment,” Bikle says that he was never asked to change his behavior. No written records exist of the meeting. By then Love-Schimenti was trying hard to find another job, but she says Bikle also foiled her attempts to land a research position at a new cancer center. Bikle denies taking any actions to hinder her.

Love-Schimenti eventually found a research job elsewhere at UCSF, but has since left research and is training for a career in graphic arts and Web page design. Bikle says he regrets that her postdoc experience turned sour and that other issues “just got in the way” of a potentially successful academic career.

19. SCIENCE CAREERS

# Will the Job Market Ever Get Better?

1. Karen Schmidt*
1. Karen Schmidt is a writer in Washington, D.C.

Although predictions of scientific supply and demand are notoriously unreliable, there are trends that contain both good and bad news for postdocs

Tony Mendez wanted to teach and do research at a major university. But only a few such positions were advertised when he finished his Ph.D. in nuclear physics at Florida State University in 1993, and the market hadn't improved much when he completed a postdoc at the University of North Carolina in 1996. After failing to make the short list anywhere, he set his sights on small liberal arts colleges. Wrong again. “A lot of people, like me, had the perception that they can try for the small colleges,” says Mendez. “But these places were being flooded, too.” Today, Mendez works for a company in Tennessee that builds cyclotrons. He misses the classroom, but with a family to support he's glad that the pay is better in industry.

Renee Williard could see the storm clouds gathering in 1995 as she finished her Ph.D. in pharmaceutical chemistry at the University of California, San Francisco (UCSF). Realizing that the academic job market was so tight that her chances of getting a university position were almost nil, she chose a postdoc involving research on health policy. “I felt like I was jumping off a cliff into the unknown,” she says. It turned out that grants and permanent jobs in that field were equally scarce. Today Williard holds a part-time job coordinating pharmacy benefits for San Francisco's department of public health and also works as a consultant and free-lance medical writer.

Anecdotes abound of disenchanted postdocs whose dreams of becoming a research professor at an elite university have been ground to dust by today's tight job market. And even those who get their wish don't have an easy time of it. Although unemployment may be low—1.5% in 1997 for those 1 to 3 years post-Ph.D., according to the latest figures from the National Science Foundation (NSF), below even the 1.9% for a similar cohort in 1995—the job search can be long and harrowing, says Charlotte Kuh, executive director of the Office of Science and Engineering Personnel at the National Research Council (NRC). “We educate graduate students, they go into postdocs, stay there longer and longer, and then at the end, it's not at all clear that there's a real career for them,” she says.

That sense of disillusionment, if not despair, has stirred many scientific organizations to take a closer look at the job market for their members. But the slew of reports in recent years tends to be heavy on hearsay and opinion, mixed with data that are often incomplete or out of date. Particularly embarrassing was a short, internal 1987 report by NSF, whose warning of a massive shortfall of Ph.D. scientists in the 1990s was disseminated widely by scientific leaders. As tens of thousands of postdocs can bitterly attest, that “prediction” failed to materialize.

This article will not attempt to forecast the job market for scientists, nor handicap the most promising alternatives to academe for those with Ph.D.-level skills. But ceding prognostication to the disciples of Nostradamus still leaves plenty of room to explore recent career trends affecting postdocs that may inform their decisions. Here are some that seem especially striking or significant.

Postdocs have become a major force in the U.S. scientific community. From just under 14,000 in 1979, the number of postdocs has nearly tripled, to 38,050, in 1997, according to NSF data. Between 1995 and 1997 alone, postdoc ranks swelled by 2665. A flattening in the number of Ph.D.s awarded, dating from the mid-'90s, could begin to affect the size of the postdoc workforce, however, although the overall numbers are likely to remain large.

View this table:

As federal dollars are concentrated at a relatively small number of universities, so too are postdocs: NSF statistics show that 50 universities employ 67% of the postdocs working in the United States, and the 20 universities with the largest number of postdocs are all ranked in the top 25 universities for federal funding, topped by Harvard, with 2505.

More fields expect graduates to do a postdoc, especially some of the life sciences, and the amount of time spent as a postdoc is increasing. NSF figures show that, for all science and engineering fields, the percentage of Ph.D.s who go on to do postdocs has risen from 25% for the pre-1965 cohort to 41% for the 1992 to '94 group, the most recent surveyed, and that the median time served has stretched from 20 to 29 months, a 45% increase. The greatest rise in popularity is in engineering, which jumped from a mere 8% of the pre-1965 cohort to 28% in the most recent group, although the average stint is only about 15 months. At the other end of the spectrum, 71% of those who earned a U.S. Ph.D. in the biological sciences between 1992 and '94 did a postdoc, up from 40% for the pre-1965 cohort, a percentage that has been stable since the mid-'80s. And they spent an average of 46 months in that category.

The numbers may be even higher as you move up the pecking order: A study of Harvard Ph.D.s. from 1988 to '93 found 59% to 68% in the natural sciences took postdocs after graduation. In one subfield, biochemistry, Maresi Nerad and Joseph Cerny at UC Berkeley found that 86% of a cohort of 654 who earned Ph.D.s from 61 research universities in 1983 to '85 did postdocs.

View this table:

Postdocs feel that the competition for academic positions, always intense, has grown worse. A survey of 1996–97 graduates by the American Society for Microbiology found that more than twice as many microbiology Ph.D.s considered the job market “bad or hopeless” as considered it “good or excellent.” A 1996 survey of 1322 Canadian postdocs found that “confidence in finding a job in their chosen field in Canada is strikingly low,” although the 16% of Canadian postdocs working in the United States were considerably more optimistic than their counterparts who remained in Canada. The survey also found that only 13% would “unreservedly recommend” that young people take the same career path, which for two-thirds of the postdocs is headed toward a tenured faculty position. A sobering 25% said they would not recommend that a student follow in their scientific footsteps.

At the same time, NSF's Mark Regets points out that it has always been tough to get these jobs and that academia has been a tight market for a long time. Indeed, the Berkeley study shows that only 35% of biochemistry grads hold tenure or tenure-track jobs in academia 10 years after their Ph.D., a number equal to the share working in a combination of the business, government, and nonprofit world. NSF figures show that only 20% of life scientists with Ph.D.s in 1992 to '94 held such positions 3 years later, a figure less than half that of the 1968 to '70 cohort.

Even for those who do obtain tenure-track jobs, the long trajectory has meant that many young scientists marry and start families before they have a clear picture of their professional future. Figures from the Harvard survey, by the Graduate School of Arts and Sciences, show that to be the case even for those starting out at the top. Whereas roughly 60% of those in the natural sciences who graduated before 1990 had secured nonpostdoc academic jobs by 1996–97, only 44% of the 1992–93 class had enjoyed such success.

There are weak signs that the strong U.S. economy is creating more job opportunities outside academia, causing new graduates to reduce or avoid their reliance on postdoctoral positions. Recent NSF figures show that 1997 Ph.D.s took fewer postdoctoral positions than their 1995 cohorts, says Regets. For instance, in the biological sciences, the percentage holding postdocs 1 year after graduation dropped from 70% to 58%; in physics the numbers shifted from 57% to 38%. Although Regets is quick to point out that the number of people taking postdocs does not directly reflect the health of the job market, “when we see this much of a change in physics and biology [the two fields with highest percentages going on to do postdocs], it's easy to speculate that the improved job market is a factor.”

Indeed, the latest survey results from the American Chemical Society found that fewer Ph.D. chemists who graduated in 1997–98 took postdocs than did the previous class—45% compared with 51%—and more found permanent jobs—44%, up from 35%. The American Institute of Physics has found a similar shift away from postdocs and toward more permanent employment—primarily in industry—in a follow-up survey of recent degree recipients.

Universities are expected to boost faculty hiring in preparation for the children of the baby boomers, who will be filling college classrooms for the next 10 to 20 years. This demographic Tidal Wave 2, as it's called, is predicted to result in a 26% rise in incoming freshmen over the next 12 years. It will be strongest in Western states—fueled by immigration as well as birth rates—and weakest in the north central region, where population is stagnant or declining.

At the University of California, a new campus will open in Merced in 2005, and university planners project 3000 new faculty positions, across all fields, throughout the university system, in addition to the need to replace retiring professors. Although the characteristics of those posts have not yet been determined, Sandra Smith, assistant vice president for planning and analysis in the office of the UC president, says that “it's quite likely that we will be hiring tenure-track faculty.”

Whatever the number, the NRC's Kuh and others believe strongly that students must be better informed about career prospects in these uncertain times. And postdocs aren't the only ones struggling to keep up with these changes. Many faculty members seem to take a very narrow view of where their students are headed. When Nerad and Cerny asked young scientists about their mentors' career expectations for them, 55% said their adviser encouraged them to pursue academic jobs, and only three (less than 1%) reported advice aimed mainly at obtaining posts in industry, government, or the nonprofit sector. In addition, very few universities offer career counseling or job placement services for postdocs. “With the world changing and many more postdocs going into industry, the need for this kind of office has grown tremendously,” says Catherine Connor, who directs the University of Illinois Biotech Placement Center, one of the few to serve postdocs.

For the most part, postdocs in the 1990s have had to blaze their own career path, a route crowded with competitors and strewn with economic boulders. And that seems like good advice for the upcoming generation as well. As Williard, who as a graduate student at UCSF organized a Women in Science group to explore nonacademic career options, puts it, “I had to re-create myself.”

20. ECONOMICS

# Cheap Labor Is Key to U.S. Research Productivity

1. Jeffrey Mervis

The low pay for postdocs has been a boon to U.S. research universities. But why are their salaries so low? And is it fair?

In 1876, Harvard-trained zoologist William Brooks was awarded all of $500—worth$7630 today—for a year's advanced study at the brand-new Johns Hopkins University. Brooks was one of 20 scholars, and the only American-trained Ph.D., chosen for what today would be called a postdoctoral fellowship. The slots were created to attract a cadre of what Hopkins's first president, Daniel Gilman, described as “men of mark, who show that they are likely to advance the sciences they profess.” And the novel strategy worked: Brooks and three of his colleagues joined the Hopkins faculty and never left.

Today, Brooks would have many more choices of where to do his postdoc and much more competition for slots in the best labs. He would probably also find it quite a bit harder to land a tenured job at a top-notch university once his postdoc ended. But one thing hasn't changed much—the low pay. Embryologist Donald Brown of the Carnegie Institution of Washington sums it up this way: “What's the most economical way to fund high-quality research? There's no question that you get the biggest bang for your buck by using postdocs.”

Postdocs' expertise and commitment are crucial to the research enterprise, as most senior scientists freely admit. So economic theory suggests that market forces—supply and demand—should set their pay levels. But the reality, at least in the United States, is that the decision is more likely to be made by a government bureaucrat based on how much an agency is willing to spend on these unsung heroes. That remuneration, adjusted only infrequently, then becomes a standard for the rest of the community—by turns a ceiling for universities trying to pinch pennies, a benchmark for those schools who want to be in the academic mainstream, and a springboard for agencies and organizations hoping to attract an elite clientele by offering considerably more. Pay scales also vary by disciplines and by support mechanism—whether the postdoc receives a competitive fellowship, an institution-based traineeship, or is funded on an investigator's grant. But whatever the number, the odds are good that it will be a lot lower than what graduates in fields outside science—especially those with a business or law degree—earn.

View this table:

When it comes to low postdoc pay in the sciences, economists may be the exception that proves the rule. A 1998 report by the Commission on Professionals in Science and Technology, based on a survey of the members of 14 professional societies, found that economics postdocs typically earn 70% to 80% more than their counterparts in the life sciences, and that only computer scientists topped economists on median salaries in academia and business (see graph). In a playful article this summer in the Journal of Economic Perspectives entitled “It's Better Being an Economist (But Don't Tell Anyone),” Harvard economist Richard Freeman suggests that natural scientists themselves are partly to blame for their low wages, including levels for postdocs. The reason, he explains, is that so many young scientists are more committed to their work than to financial rewards. They refuse to bail out even when the job market tightens, creating an excess supply that holds down wages. And postdocs are on the bottom of that heap. “It's a little embarrassing,” admits Freeman, who is working with the American Society for Cell Biology to gather data on career paths of students from some of the country's top biology labs, “because physicists and mathematicians are so much smarter than we are.”

In the United States, the de facto salary standard for academic postdocs in the life and health sciences has become what the National Institutes of Health (NIH) pays recipients of its National Research Service Award (NRSA) traineeships. For a first-year postdoc, an NRSA comes with a stipend of $26,252, up from$20,292 only 2 years ago and the first sizable hike in many years. (NIH doesn't call postdocs employees because they are supposed to be in training. As a result, their pay is not considered a salary and, thus, not subject to cost-of-living increases. Nor does it include benefits, although most institutions offer at least limited health coverage.)

That amount, which one faculty member calls “barely livable” and which NIH officials acknowledge was behind the times, actually applies to only 7000 NRSA postdocs at outside institutions. The 2230 intramural scientists working on the NIH campus enjoy a higher pay scale that factors in the area's higher cost of living. However, the NRSA levels have been adopted by many U.S. universities and research labs as the first rung on the salary schedule for all postdocs, most of whom are funded through grants to faculty members. “NIH has done a good job standardizing salaries and keeping them low,” says physicist Peter Fiske, who in 1996 moved from a postdoc to a staff scientist position at Lawrence Livermore National Laboratory in California and has been active in the lab's postdoc association.

It's not a role that NIH officials embrace, however. “Sometimes things are used in ways that are not appropriate,” says Wendy Baldwin, NIH's head of extramural research. She emphasizes that institutions are free to supplement the NRSA stipends and that they, not NIH, must decide how much to pay postdocs supported on research grants. In addition to the NRSA postdocs, NIH funds about 6500 postdocs a year on extramural research grants, more than any other government agency. [The National Science Foundation (NSF) supports about 4650 postdocs annually on research projects, plus another 150 to 200 as postdoctoral fellows, at stipends that vary but are generally higher than NRSA awards.]

Unintended or not, NIH's policies toward postdoc pay affect the entire community. The recent across-the-board 25% boost in NRSA stipends, which rise to $32,700 after 2 years and top off at$41,268 for postdocs with seven or more years of experience, creates pay inequities between those hired at the new levels, which went into effect 1 July, and existing staff. University officials and lab chiefs say that a dual pay scale is bad for morale and difficult to justify, but that they don't have the money to make things right. And some think NIH should make up the difference. “We have a policy that says all postdocs should get the NRSA recommended levels, but there's no mechanism to bring up the rest,” says Trevor Penning, director of postdoc programs at the University of Pennsylvania, who sees it as part of NIH's duty to meet the full cost of sponsored research but who admits that the idea has gotten a chilly reception from NIH officials. “I don't know how much it would cost, but NIH has to step up to the plate and address the issue.”

Of course, the more NIH or any other agency pays postdocs, the less money it has for the rest of its research portfolio. Unless their budget rises sharply—as NIH's did last year, allowing director Harold Varmus to boost NRSA's $430 million budget by$81 million—agency officials must either shrink other portions of a grant or make fewer awards. Neither scenario is appealing to researchers, who say grants are already too small and success rates too low.

Absent a sudden windfall, however, universities and faculty members must scramble to make up the difference. Johns Hopkins University is planning to match the new levels for all its postdocs over 3 years, for example, although Jeremy Berg, chair of the department of biophysics, says that “[researchers] are encouraged to meet those levels as soon as possible” for postdocs on their grants.

For investigators with large labs, that's not a big problem. “We use the NRSA pay scale as a starting point,” says Cliff Tabin, a professor of genetics at Harvard Medical School in Boston with a current roster of 10 postdocs. “If someone is making less, I supplement them to that level. If somebody gets a fellowship that pays more, that's great. And since I don't want anyone to take a pay cut after 3 years [when the fellowship ends], I try to keep them at that level and raise up everybody else.”

But smaller labs may have a more difficult time. “The Yale standard is the new NRSA levels, which is barely livable,” says Marina Picciotto, an assistant professor of psychiatry at Yale Medical School, who has three postdocs and a mixture of NIH and private funding for her work in molecular neuroscience. “And they recommend that you supplement it. But some of my grants are expiring, so it could be tough right now to move anybody up.”

Laboratory chiefs have greater discretion to set salaries for postdocs paid on their grants. Kamiya says he has paid particularly productive postdocs as much as $4000 a month in their final year. They also receive money to attend conferences. “I can be generous because the funding for this program is very generous,” he says. Despite their growing popularity, postdocs are still struggling for public recognition. Although credit card issuers long ago realized that postdocs are worthy credit risks, there is a stigma attached to temporary positions in a land where employment for life is still an ideal. Ohsugi says people outside the scientific community are surprised that someone with a Ph.D. “has to settle for a temporary position. To me, it's not such a serious problem,” she adds. “But it really worries my parents.” What worries Ohsugi and her compatriots is the gloomy outlook for post-postdoctoral employment. The problem is the lack of new permanent positions, a consequence of a government decision to shrink public payrolls. One unofficial calculation shows that the 10,000 postdocs will be chasing just 1000 to 2000 full-time academic job openings over the next few years. That estimate has caused Atsushi Iwamae, a chemistry postdoc at the University of Tokyo, to flee his 5-year postdoc position after just 2 years to become a research associate at Kyoto University. The permanent position won't give him as much freedom to pick his own research targets, he acknowledges, but Iwamae didn't want to be caught without a job at the end of his postdoc. “There are very many candidates for each open position,” he says. While acknowledging the tough job market, Ikegami and others say that the increased competition should raise the quality of research at national institutions. “We can choose those who are really the most talented,” says Ikegami. A more lasting solution, proposed by Ikegami's committee, would sidestep the limits on permanent positions for scientists by creating “super” postdoctoral positions for younger researchers who have completed one postdoc and are ready for more independence (Science, 9 April, p. 233). More experienced researchers capable of leading a team would be eligible for independent researcher positions. Both types of positions would be for fixed terms and would be filled through an open, competitive selection process. The trade-off for impermanence, says Ikegami, would be more money and more freedom for a researcher to pursue his or her own interests. The proposal is being reviewed as part of a larger package of prospective science initiatives. 22. FOREIGN EXCHANGES # Going Abroad Needn't Mean Going Into Exile 1. Jeffrey Mervis Sometimes it pays to be uninformed. Among ambitious young U.S. scientists, conventional wisdom says that doing a postdoc abroad is a bad career move because it takes one out of the scientific mainstream and makes it tougher to enter an already tight job market. But William Skarnes, now an assistant professor of molecular and cellular biology at the University of California, Berkeley, hadn't heard that received wisdom when he weighed his options after receiving a Ph.D. in 1990. Instead, he accepted a postdoctoral fellowship from the international Human Frontiers Science Program (HFSP) to work in the lab of a particular scientist in the United Kingdom. “I wanted to learn about early mouse embryology from Rosa Beddington [then at the Medical Research Center laboratory in Edinburgh, Scotland],” says Skarnes, who turned down other offers to go there. “I don't think that I would have learned as much about mouse development anywhere else. … I didn't realize until I came to Berkeley that some people consider it a liability to do a postdoc abroad.” Indeed, Skarnes and others who have done one say that a postdoc abroad can sharpen one's skills, broaden one's perspective on how other countries do science, and augment one's network of contacts—without draining one's bank account. “So when I heard about the good stipend and bench money that goes with the [HFSP] fellowship, I thought it would be a great opportunity,” he adds. The 2-year HFSP fellowships, created in 1989 and funded by the United States, Japan, and the European Union, are available for Ph.D.s from participating countries who wish to study abroad. But demand from U.S. scientists is not high. Last year they made up only 4% of the applicants, while 54%—and 62% of the 160 winners—chose to come to the United States. And programs aimed specifically at U.S. postdocs have a hard time overcoming the conventional wisdom. “We have a lot of problems attracting [U.S.] applications,” admits Rolly Simpson of the Burroughs Wellcome Fund, who says a 9-year-old program to study in the U.K. attracts barely two dozen applicants annually for its 10 slots, despite generous stipends and support for travel, supplies, and other research needs. To sweeten the pot, last year officials stretched its 3-year Hitchens-Elion postdoc fellowship into a 5-year award that includes 2 years' salary and start-up funds as a new faculty member at any institution. But the response this year was no better. “People just don't want to go to the U.K.,” he says. That's too bad, say postdocs who have left home. Many U.S.-based HFSP fellows who have studied in Europe say that they enjoyed the more collegial and personable atmosphere than they had found in the United States. “You interact more with people. All the doors are open and everybody shares. They even borrow things right off your lab bench, which takes getting used to,” says Rebecca Hartley, a molecular biologist at the University of Iowa, Iowa City, who worked with H. Beverly Osborne at the University of Rennes in France. Marina Picciotta, an assistant professor of psychology at Yale Medical School, says that her postdoc under Jean-Pierre Changeux of the College de France, whose lab is at the Pasteur Institute in Paris, “gave me a broader focus on neuroscience—and since the community is smaller, I got to meet some of the best scientists in Europe.” For some fellows, their stint overseas began with an unscientific but nevertheless compelling desire to travel. “I thought that a postdoc would be a perfect time to go abroad,” says Picciotta. “I was warned that it would set me back, but I was willing to do a second postdoc in the States, if necessary.” For others, the HFSP fellowship allowed them to link their scientific progress to interest in a particular culture. Marc Lamphier of Eisai Pharmaceutical Co. in Andover, Massachusetts, a subsidiary of Eisai Co. Ltd. of Japan, majored in Japanese studies as an undergraduate and had worked as a translator and interpreter before getting his Ph.D. in molecular biology at Harvard in 1991. He parlayed a postdoctoral fellowship at Osaka University into a post with a Japanese research agency before returning to the United States in 1997. “Eisai is moving to set up labs in the West, and they need people who can bridge the gap,” says Lamphier. “And industry seemed like a good place to apply my international experience.” Overseas postdocs may even offer advantages for academics facing tenure review, says immunologist Janis Burkhardt, who did a postdoc at the European Molecular Biology Laboratory (EMBL) from 1992 to 1996 before joining the University of Chicago faculty. “At EMBL I had colleagues from all over the world,” she says. “So when I come up for tenure, it will be a lot easier for me to show evidence of an international reputation.” In December, HFSP will hold a 10th anniversary celebration featuring successful postdocs touting the benefits of doing science in another country. Meanwhile, Burroughs Wellcome Fund officials hope that helping postdocs establish their careers will appeal to an audience that “is reluctant to leave the U.S. system,” says program manager Martin Ionescu-Pioggia. “This new program not only gives them a chance to study in the U.K., but it helps them in getting a job, too.” 23. FOREIGN EXCHANGES # Europeans Who Do Postdocs Abroad Face Reentry Problems 1. Michael Balter A stint abroad is crucial for many European Ph.D.s, who must overcome government resistance to temporary positions in order to become academic researchers HEIDELBERG, GERMANY, AND PARIS, FRANCE—Florence Horn is fast becoming a scientist of the world—but not necessarily by choice. Born and raised in Normandy, France, Horn received her Ph.D. in bioinformatics from the University of the Mediterranean in Marseilles, and this fall she'll complete a 3-year postdoctoral fellowship at the European Molecular Biology Laboratory (EMBL) in Heidelberg. Then she's headed off to the University of California, San Francisco, for a second postdoc—if she can find salary support. Horn would like to return to France. But at 32 she's 1 year over the age limit to compete for an entry-level position at the French research organization where her expertise might be most welcomed—the basic science agency CNRS. And the French government funds no domestic postdocs for people in her field, in keeping with a 3-decade-old policy that says it would be unfair to offer people temporary posts with no promise of permanent employment. Officials acknowledge that postdocs represent a rich vein of scientific talent that could bolster French science. But except for a few specially targeted programs, France's research policy-makers are reluctant to put public money into the pockets of French postdocs. Many young European scientists share Horn's dilemma. Science met recently with nearly a dozen EMBL postdocs who, although they come from a wide variety of countries and backgrounds, have all been encouraged to do their postdocs outside their home countries. Now they are struggling to reintegrate themselves into their native scientific communities. Some have been luckier than Horn: When Guillermo Montoya, a structural biologist from Spain, leaves EMBL at the end of next year, he will become a group leader at the Biophysics Institute now under construction in Bilbao. But Austrian structural biologist Susanna Lüdemann's experience may be more typical. With few opportunities in her native country, she has resigned herself to being a permanent expatriate. “If I wanted to pursue a scientific career, I had to leave Austria,” she laments. German postdocs who seek an academic career at home must also overcome the formidable Habilitation, a lengthy process of qualification for university posts. Although the Habilitation's days may be numbered (see sidebar), the road to a permanent position remains long and rocky. The problems faced by many European postdocs reflect differences in scientific culture between continental Europe on the one hand, and the United States and the United Kingdom on the other. Whereas U.S. and U.K. graduate students are seldom pushed to do postdocs outside their countries, continental Europeans see a foreign stint as a feather in a postdoc's professional cap. “A postdoc abroad is generally considered necessary for your CV,” says neuroscientist Markus Missler of the University of Göttingen in Germany. A U.S. posting in particular is so highly valued that some German researchers jokingly add “iAg”—in Amerika gewesen (been in America)—to their abbreviated titles. Yet despite the luster that foreign travel can add to a résumé, the tight job market for researchers can make reentry difficult for even the most promising young scientists. And although continental postdocs might envy their British colleagues for being able to stay home, a domestic postdoc in Britain has its own downside: a huge corps of Ph.D.s who find it hard to get off the postdoc treadmill. A major reason many countries shove their newly minted Ph.D.s out the door is the hierarchical European university systems. A change of scenery is deemed essential for the nurturing of scientific talent. “We strongly encourage them to leave the place where they have been trained as graduate students,” says Ernst-Ludwig Winnacker, president of the Deutsche Forschungsgemeinschaft (DFG), Germany's basic research granting agency. “The idea is for people to have mobility.” Indeed, “mobility” has become a mantra for European research policy-makers. “We have decided not to fund French postdocs in French research organizations,” says geophysicist Vincent Courtillot, director of research in France's science ministry. “It prevents mobility, and we don't think that is healthy.” As a result, nearly 70% of French postdocs do their post-Ph.D. training outside France. Another reason for the flight of young talent is that most European countries are not strong in all scientific fields. Thus, postdocs wanting to apprentice in the best labs often must dig out their passports and hop on a plane. Government funding policies reinforce these political and scientific realities. For example, of 475 grants to young scientists awarded by the DFG in 1998, 84% went to researchers working in institutions outside Germany. An even stronger incentive to leave home is built into the European Union's (EU's) postdoc programs. The largest of these, called the Marie Curie Fellowships, provides postdoc positions to nearly 3000 young researchers—but only in labs outside their native countries. In some countries, the dearth of domestic postdocs until recently left doctorates with no choice but to leave home. “When I was a Ph.D. student in the 1980s, my only possibility was to go abroad,” says chemist Dario Narducci of the University of Milan in Italy, who did his postdoc at an IBM research center in New York. But during the last few years, the Italian government has begun funding 3-year postdoc positions in the universities and public research centers, which have been snapped up by young researchers. Despite these new opportunities, however, many already-established Italian scientists believe that a tour abroad is still the best path to scientific excellence. “If they don't go out of the country, they will not grow intellectually,” says immunologist Mario Clerici, also of the University of Milan, who spent a lengthy postdoc at the U.S. National Institutes of Health. The concept of a postdoc abroad as a rite of passage for young researchers has never caught on in the United Kingdom, which relatively speaking might seem like a postdoc paradise. The Engineering and Physical Sciences Research Council, for example, one of six government-funded research councils, alone supports about 4500 postdoctoral posts, almost all in the U.K., while the Medical Research Council foots the bill for just over 2000. Medical charities also play a large role: The mammoth Wellcome Trust funds about 2100 postdocs, for example, compared with 244 for the largest biomedical charity in France, the Association for Cancer Research. But a large number of U.K. postdocs fall into a career limbo created by the combination of rising Ph.D. production and scant growth in the number of permanent positions. For example, a recent survey of physics postdocs by the London-based Institute of Physics found that nearly 30% had been postdocs for more than 6 years. There are an estimated 35,000 of these so-called “contract researchers” in Britain, a number that has “doubled and then doubled again since the 1980s,” says David Bleiman, assistant general secretary of the Association of University Teachers. “It has become a monster that is out of control.” France stands at the other extreme. While the EU, Germany, and many other individual European countries provide some government funds to postdocs, at least to those willing to go abroad, France has held to a policy that dates from the 1970s, when the Socialist government decided that it was unfair to offer such temporary posts to young people. Many French scientists believe this policy, breached only for small programs in targeted areas, is hurting the nation's research effort. “Going from the Ph.D. thesis to a position is difficult in France,” says Simon Wain-Hobson, a virologist at the Pasteur Institute in Paris. “But what can you expect when there is no clear and coherent postdoc system? It is costing us dearly.” For example, Wain-Hobson says, “good French students get good offers from abroad, or from industry, where they can make money and live comfortably.” Indeed, the postdoc problem in France is on the front burner in the ongoing debate over reforming French research (Science, 18 June, p. 1898). Courtillot, who is responsible for the research ministry's postdoc programs, agrees that French policy has created a “postdoctoral gap” that can make life difficult both for young scientists seeking postdocs and the labs who need them. In an attempt to address the problem, the research ministry has recently created 250 government-funded postdoc positions each year in industry and at high-tech public research agencies such as the Atomic Energy Commission and the national space agency CNES. Moreover, Courtillot told Science that the ministry began this year funding 100 foreign postdocs coming to France as the start of a program that it hopes will grow to 500 slots a year by 2001. The government is hoping that this program will lead to a reciprocal arrangement with other countries. “If a postdoc comes to my lab, I am not going to ask Stanford University or the National Science Foundation to pay his or her salary. I have students in the U.S. and Australia, and I haven't paid for them,” says Courtillot. But he adds that, outside of the 250 specially targeted postdocs, the government is still unwilling to provide state funding for a broad range of French postdocs to work either in France or in foreign labs. Despite the persistence of such policies, the tide may be turning for expatriate scientists like Florence Horn. A recent report by the French Academy of Sciences targets bioinformatics as a high-priority area for government funding. “I don't want to come back just to come back,” says Horn. “But when I do, I will have a lot of skills to offer.” 24. GERMANY # Germany Tries to Break Its Habilitation Habit 1. Michael Balter HEIDELBERG, GERMANY—For young researchers seeking an academic career, the trek from graduate school to a permanent position is long and arduous under the best of circumstances. But in Germany, tradition has laced this tortuous road with an additional minefield: the infamous Habilitation, a post-Ph.D. degree for aspiring professors. In the sciences, the Habilitation consists of a long—10 years or more is not unusual—postdoc spent in the laboratory of a senior professor. It ends at the discretion of the candidate's academic advisers, and with recipients in some fields pushing 40. What's more, young German researchers doing postdocs abroad (see main text) cannot begin the process until they have returned. View this table: But the Habilitation's days may be numbered, at least in the natural sciences. The Deutsche Forschungsgemeinschaft (DFG), Germany's basic research granting agency, has begun a fellowship program that may provide an alternative—and quicker—route to a tenured position for postdocs returning from abroad. And last year, the German parliament gave universities greater leeway to design the path to tenured positions. The universities have been receptive: In July, Germany's Conference of Rectors and University Presidents (HRK) agreed that the time from entry into a Ph.D. program to qualification for a tenure-track position should not exceed 10 years. “The doctorate should be reached by age 27 or 28, and then another 6 years at most for young people to get the tenure qualification,” says HRK president Klaus Landfried. In particular, the HRK is urging those in the natural sciences to switch to a program of “qualification professorships” that would cut the umbilical cord between postdoc and adviser much earlier than under the present system. The recommendations come in the wake of a report by a blue-ribbon panel commissioned by the government and issued in June. A key recommendation of the panel, led by materials scientist Richard Brook, chief executive of the U.K.'s Engineering and Physical Sciences Research Council, was to replace the Habilitation with something akin to the assistant professor system in the United States (Science, 4 June, p. 1595). “Young researchers should be encouraged to do independent work more quickly than they do now in the German system,” Brook told Science. “The Habilitation means you can be 40 years old before you find out whether you have an academic career ahead of you.” The bells tolling the possible death of the Habilitation are music to the ears of German postdocs. “The current system is a closed shop; nothing is openly discussed or public, and there is no security or official status for the applicant,” says Thomas Dandekar, a German structural biologist at the European Molecular Biology Laboratory (EMBL) in Heidelberg, who received his Habilitation degree in 1994. “There should be Germany-wide rules” for how successful candidates are chosen, he adds. And Germans who do postdocs abroad should be doubly happy, given their late start at obtaining the Habilitation. “Everyone tells you it is prestigious to go abroad,” says EMBL postdoc Thomas Preiss, who is currently working on his Habilitation degree. “But when you come home, no one is waiting for you.” The DFG's new fellowship program is an attempt to redress this situation—at least for the most talented young scientists. The 5-year fellowships, begun earlier this year, allow researchers to go abroad for a postdoc and return to a more independent position that leads to the chance for tenure. The new scheme is named after mathematician Emmy Noether, the first German woman to receive a Habilitation degree. Flooded with applications, the DFG has so far awarded more than 50 grants, with another 20 or 30 expected by the end of the year. “Our hope is that the funds will allow people who have gone anywhere in the world to come back as independent researchers,” says DFG president Ernst-Ludwig Winnacker. That hope is shared by the first grant recipients. “No other fellowship would put so much trust in researchers at such an early age,” says Emmy Noether winner Florian Hollfelder, a bio-organic chemist from Berlin currently doing a postdoc at Harvard Medical School in Boston. “The program really puts Germany back on the map for young scientists.” 25. DENMARK # Denmark Proposes Postdoc Tonic for Universities 1. Dennis Normile Countries pondering how to blend postdocs into an established system might want to watch an experiment about to begin in Denmark, which plans to create up to 400 postdoctoral positions at the nation's universities and research centers over the next 4 years. Denmark roughly doubled its university system in the 1960s and 1970s, says Ove Poulsen, deputy permanent secretary for research at the Ministry of Research and Information Technology. But the 4000 tenured faculty members have “grown old with the system,” he says, and creating “a postdoc culture” is seen as a necessary tonic. The medicine will cost the national government$15 million—1% of its overall investment in research—which, matched by universities, will create up to 100 postdoc positions a year in each of the next 4 years. To get the money, university administrators must explain how the new positions will help their institutions create world-class centers of excellence in new or existing areas of research.

Jens Oddershede, dean of science and engineering at the University of Southern Denmark, Odense, and a member of the government's advisory National Science Council, says the initiative will help university administrators start grooming replacements for the large numbers of faculty members approaching retirement. Oddershede says the Danish postdocs will be more like junior faculty members than their U.S. counterparts, with teaching duties and greater control over their own research activities. And thanks to an expansion of graduate programs earlier this decade, there are a lot of well-qualified younger scientists who could fill the positions.

Poulsen says Denmark will try to avoid one problem facing countries with entrenched postdoc systems—young researchers stuck in endless “postdoc cycles”—with guidelines requiring universities to specify the number of postdocs likely to be given tenured positions. Such a target will help postdocs assess their chances of success and force universities to plan ahead. But “we're still wrestling with the details,” he admits.

# Eight Attributes of Highly Successful Postdocs

1. Constance Holden

Successful young faculty members offer postdocs some frank tips on how to pick the right lab—and how to flourish in it

Oliver Hobert had a clear vision of his ideal postdoc: Find a prominent principal investigator at a top U.S. lab in a nice geographic area with a well-developed scientific community. An impossible dream? Not for Hobert, who received his Ph.D. in neurobiology in 1995 from the Max Planck Institute for Biochemistry in Martinsried, Germany.

He snared a 2-year fellowship from the international Human Frontiers Science Program, based in Strasbourg, France, and won a spot in the Caenorhabditis elegans lab of biologist Gary Ruvkun at Harvard Medical School in Boston. Working with a well-known investigator in a lab small enough to ensure him individual attention but big enough to allow him some independence, Hobert honed his knowledge of the model organism, the first multicellular organism whose genome was fully sequenced, and in 3 years got his name on six published papers—four as first author.

Indeed, his postdoc experience was so productive that in January 1998 David Hirsh, chair of the biochemistry and molecular biophysics department at Columbia University, called him up and asked if he was interested in a job. Now, at 32, Hobert is an assistant professor of neurobiology at Columbia and the recipient of a prestigious international research grant from the same Human Frontiers organization that funded his postdoc.

Such a career trajectory is not in the cards for everyone. But Hobert's success to date rests on more than his intellect: Along the way he also made some canny career decisions that set him apart from his peers. Indeed, when Science called up a score of high-achieving young scientists to learn about their postdoc experiences, at least half of them began by explaining that they probably had not done things the way most people do. That was a tip-off to what's special about this group—they trust their own instincts and march to the beat of their own drums.

And that's only part of their formula for success. They've also had the foresight to pick promising fields before they got overcrowded and the ability to see where their work fits into a larger picture. And although they are independent thinkers, they also understand the importance of traditional markers of success: pedigree, publications, and fundability. Finally, they are the type of person others want to be around.

We've transformed these common themes into what might be called eight attributes of highly successful postdocs.

## ATTRIBUTE #1: Get ahead of the curve

Every hot young scientist surveyed managed to meld what he or she is really interested in doing with an area that has a lot of growth potential. “Assess your research field,” says Scot Martin, 29, an assistant professor of environmental sciences and engineering at the University of North Carolina, Chapel Hill, and a recipient this year of the Presidential Early Career Award in Science and Engineering. “Think about what's opening up and what will be exciting areas that will interest academic departments once you finish your postdoc.”

That approach has paid off handsomely for computer scientist Melanie Mitchell, 40, of The Santa Fe Institute in New Mexico. After doing a thesis on artificial intelligence and cognitive science, Mitchell switched to evolutionary computation—applying ideas from biological evolution to computer programming—for her postdoc at the University of Michigan in the early 1990s. “I was following what I wanted to do the most,” she says. At the same time, “I thought the field had a lot of possibilities, and there were not that many people working in it.” There are now, and Mitchell—a finalist this year for the McDonnell Foundation's $1 million Centennial award to young investigators—has become the scientific equivalent of an investor who bought into a hot stock before it shot up. But picking a field just because it looks fashionable is not a good idea, warns Alexandre Barvinok, 36, an associate professor of mathematics at the University of Michigan, Ann Arbor. “When you see a bandwagon approaching, it's already too late to jump on,” he says. “Better to do what you think is right [for you].” ## ATTRIBUTE #2: Follow your heart Hot young scientists can't emphasize enough the importance of sticking with what turns you on. Barvinok recites the advice of one of his mentors, Louis Billera of Cornell University: “If you do things the way you want to and others are unhappy, it's their problem. If you do it the way they want you to and they're still unhappy, it's your problem.” Carolyn Bertozzi, a 32-year-old assistant professor of chemistry at the University of California, Berkeley, agrees that you're more appealing to employers if you're following your bliss. “I definitely find myself most attracted to those who are in my mind a singularity. … I like to see someone drawn to their chosen field out of pure interest, excitement, and passion,” she says. Bertozzi went against the advice of her professors when she decided to do a postdoc in immunology after her Berkeley Ph.D. in chemistry. “I was counseled very strongly against it. They warned me that everyone would forget me if I left chemistry,” she says. But she wanted to apply chemistry to a “very interesting problem” involving cell adhesion that Steven Rosen was working on at the University of California, San Francisco. The detour wound up helping, not hurting, her career. “When I applied for academic positions, I'm sure I gave a job talk that people hadn't heard from anyone else,” says Bertozzi, who joined the Berkeley faculty in 1996. Her latest achievement is a MacArthur “genius” fellowship, given in June after she figured out how to modify sugar molecules on cell surfaces to reveal characteristics of carbohydrates. The technique can be applied to learning more about cell communication and protein folding, for example. ## ATTRIBUTE #3: Remember the big picture As important as knowing one's own field is knowing how it fits into the work of others. Nalini Ambady, an associate professor of psychology at Harvard University, wasn't thrilled about the prospect of a postdoc after getting her Ph.D. in social psychology from Harvard in 1991—“People [in psychology] usually go straight to academic positions,” she says. But there weren't many jobs around, so when her adviser Bob Rosenthal offered her a position, she took it. That decision allowed her to continue a line of research—studying how people make social judgments about each other after brief observations—that bears on such areas as teaching effectiveness and doctor-patient relationships. “I was able to plan for the next 3 to 4 years and start thinking programmatically about the work and where it fits in,” says Ambady. “If I'd gone straight into an academic position, I would have been overwhelmed.” The experience has clearly paid off: She was hired by Harvard in 1993, and this year she received a presidential early career award for “fundamental contributions to understanding the accuracy of social judgments based on thin slices of information.” ## ATTRIBUTE #4: Acquire a pedigree Although following your own instincts is vital, it's also important to look good to potential employers. That's what doing a postdoc with a prominent person can achieve, as well as generating an invaluable network of contacts. “The notion of the pedigree still holds some water,” says Bertozzi. Neuroscientist Randy Buckner, 29, an assistant professor of psychology at Washington University in St. Louis and another Centennial finalist, agrees: “Coming from a great lab is a major predictor of future success.” Chemist Cassandra Fraser, 36, an assistant professor at the University of Virginia, Charlottesville, who did her postdoc at the California Institute of Technology (Caltech) with well-known chemist Robert Grubbs, says she has observed that a strong candidate from a top lab will often have an edge over even a reportedly brilliant applicant from a so-so lab. Not everyone agrees. Ruvkun says some people coming out of great labs may look good based on the work of others, on the principle that a high tide raises all boats. “So someone [good] coming out of a lab not so well known is really impressive.” As important as a pedigree is a track record. A strong list of publications helps the potential employer figure out, as Buckner puts it, “Is this person a ‘closer’?” For every paper published, Buckner believes, “there are two others that people have dropped” just short of getting them ready for publication. ## ATTRIBUTE #5: Do your homework So how do you land that great postdoc position? Start by acting like a grown-up. “There's a transition you have to make, from thinking of yourself as a person in training to thinking of yourself as an independent scientist with opinions to offer,” says neuroscientist Jennifer Groh, a 33-year-old assistant professor at Dartmouth College in Hanover, New Hampshire. Applicants need to be thoroughly prepared to explain what they can contribute to the lab. Ruvkun says he looks for people who “exude mastery”—something Hobert did “right from the start.” But, he says, “when people apply I'm always surprised at how few of them have actually read everything from my lab.” Hobert agrees: “Now I'm interviewing [postdoc applicants] myself, and it's important if this person really read the papers I've written. … I would think only half of them do.” Enthusiasm is important, too, he adds: “I just had an applicant here whom I really liked. But I did not make him an offer for the simple reason that he didn't send me an e-mail afterward telling me he was interested.” ## ATTRIBUTE #6: Bring money Another very attractive quality is the ability to stand on your own feet financially. Hobert offers this bit of advice: “If you are accepted in a lab, and the principal investigator tells you you don't need to apply for funding, apply for funding anyway.” It looks good on your CV, helps you organize your thoughts, and demonstrates that you can generate “excitement about your research plan.” Conversely, adds North Carolina's Martin, if you are applying for a postdoc position, “a bad way to open the conversation is by asking, ‘Do you have money?’” Martin used his fund-raising prowess to land in the lab of a Nobel Prize-winner. After doing his Ph.D. at Caltech on water (aquatic photochemistry), he got interested in air and decided he wanted to work with Mario Molina at the Massachusetts Institute of Technology (MIT). Instead of asking for a job, however, Martin asked Molina to sponsor his fellowship application to the National Oceanic and Atmospheric Administration. From there it was a short step into the MIT lab, which Martin joined 2 weeks before Molina won the 1995 Nobel Prize for ozone chemistry. The type of fellowship can also be very important, says Andres Garcia, 30, a biomedical engineer who snared a tenure-track position at the Georgia Institute of Technology in Atlanta only 1 year into a Ford Foundation minority postdoc in David Boettiger's microbiology lab at the University of Pennsylvania. “Bioengineering is a very hot field, and the prestige of the Ford Fellowship made me look good,” says the Puerto Rican- born Garcia. ## ATTRIBUTE #7: Forget the want ads “A lot of people view the postdoc position as just a buffer against a bad job market. They look at the ads and see who will pick them up for 2 years while they wait to get their shot at the market,” says Bruce McCandliss, who is finishing up a postdoc in cognitive neuroscience at the Center for the Neural Basis of Cognition at the University of Pittsburgh. That's the wrong approach, says McCandliss. A postdoc, he says, “offers the chance to form a unique collaborative relationship that should not be considered lightly.” McCandliss, 33, wanted to extend his doctoral work on tracking learning-related changes in brain waves into a project that uses functional magnetic resonance imaging (fMRI) to track changes in brain activation as children learn to read. “The domain was brand-new,” he says, and very few investigators were working on this question. He decided that Pittsburgh was the best place in the country for him to both learn fMRI and collaborate with experts on reading. Although there was no direct funding available for such a postdoc position, McCandliss brainstormed with his Ph.D. adviser Michael Posner at the University of Oregon and several researchers in Pittsburgh. He ultimately won a grant from the McDonnell Foundation for his salary, and the Pittsburgh team later obtained a large grant from the National Science Foundation. “I thought of the postdoc as a way to set up the most exciting collaboration I could imagine, rather than a process of hunting for an advertised job,” says McCandliss. “People who wait and answer general ads for postdocs can wind up working on a preexisting project and have less of a hand in designing the whole experience.” Another high achiever who strayed off the beaten track is 32-year-old computer scientist Michael Littman, an assistant professor at Duke University, who pursued what he now calls a “predoc.” He spent 4 years as a researcher at Bellcore after graduating from Yale University in 1988. “My family was convinced that I'd never go back to school,” he says. But the experience allowed him to discover his real interests, laying a solid foundation that “helped me go through my Ph.D. faster.” It also served the function of a regular postdoc: His current lines of research—artificial intelligence and cross-language information retrieval—“both have their roots in my time at Bellcore.” ## ATTRIBUTE #8: Be a team player “Your accomplishments can get you into the top 20 out of 200,” says Martin. “But after that any one of those top 20 can do the job well.” At that point, it's the nuances that count—and that's where the interpersonal part comes in. Groh is a good example, according to Mike Shadlen of the University of Washington, who did a postdoc with her. Groh, who does experimental work with humans and monkeys on how visual and auditory signals combine and generate behavioral responses, is “extremely generous and kind-spirited,” says Shadlen. She's “very open with her ideas—enthusiastic, willing to engage, willing to be wrong. She's the kind of person you want to hire in your lab.” Groh, who also has an impressive pedigree—she worked with David Sparks at the University of Pennsylvania and did a postdoc with William Newsome at Stanford University—makes it clear that interpersonal skills have to include savvy as well as nice. When she interviewed for a postdoc, she asked people about the atmosphere in the lab and whether there were any hidden conflicts. And she expects people applying to her lab to do the same. Her combination of attributes won her nine job offers while a postdoc, including a$1 million start-up offer—four times the norm—from The Rockefeller University, which she turned down because she and her husband were looking for jobs in the same place.

Following all these rules doesn't guarantee you the job of your dreams, of course. But ignore them at your peril, say those former postdocs who have found them helpful in achieving success at a young age.

27. DIVERSITY

# Minority Postdocs Are Rare, Independent Breed

1. Jeffrey Mervis

The tiny number of minority postdocs suggests that the problem starts at the beginning of the pipeline. Still, very little is being done to plug the leaks near the end, where careers are meant to blossom

For the past 3 years, 10 talented U.S. African Americans have won a biomedical postdoctoral fellowship under a program funded by pharmaceutical giant Merck & Co. and administered by the United Negro College Fund (UNCF). This year, the program will expand to 14 awardees, thanks to additional funding from another drug company, Parke-Davis. It doesn't sound like much until UNCF's director of science education, Jerry Bryant, tells the rest of the story: “By my estimation there are only 124 African Americans currently doing postdocs in the biomedical fields who are U.S. citizens or permanent residents. And we're funding a significant proportion of them.”

Minorities are rare in science all along the educational pipeline, but by the postdoctoral level the factors that pluck African Americans, Hispanics, and Native Americans out of science have shrunk the pool to vanishingly low numbers. And because the best way to change the situation is to boost the numbers entering the pipeline—reaching back to high school and even elementary school—minority postdocs haven't gotten much attention. Proponents of these special programs argue that their tiny numbers make each minority postdoc precious and that focusing attention on them is the best way to encourage others to follow in their footsteps. But the vast majority of scientists seem to feel that the scarcity of underrepresented minorities is not an issue that should be addressed at the postdoc level, where performance is all that matters.

Even tracking the numbers of minority postdocs is difficult. Although African Americans, Hispanics, and Native Americans make up 24% of the U.S. population and graduate from high school at rates close to those of whites, they receive only about 14% of the undergraduate science and engineering degrees and only 7% of science and engineering Ph.D.s. By the time they become postdocs, their numbers are so small that the National Science Foundation (NSF), in its regular survey, doesn't even collect data by race; hence Bryant's estimate is based on a rule of thumb that the postdoc population in biomedical fields is roughly twice the number who earn Ph.D.s annually in those fields. Surveys that follow the fate of Ph.D.s likewise come up empty when tracking minorities. A Berkeley study that tracked down 654 biochemists 10 years after their Ph.D., for example, contains replies from only nine Latinos, four African Americans, and no Native Americans.

Those tiny numbers, plus the ambiguous status of postdocs at most institutions, also make it very difficult to track the impact of the few programs aimed at underrepresented minority postdocs. The National Institute of Allergy and Infectious Diseases (NIAID) in Bethesda, Maryland, for example, gives its intramural scientists an extra postdoc slot in their labs if they bring on a promising minority. Richard Asofsky, a training officer who oversees the program, says he's recruited about a dozen minority scientists that way in the past 5 years. That's quite an accomplishment, he asserts, given that there were “maybe two” minority postdocs among the more than 200 at the institute when the program began. Exact figures do not exist because the National Institutes of Health (NIH), NIAID's parent body, doesn't keep data on the racial composition of the 2230 postdocs on campus. Such data are only required for employees, and postdocs are considered trainees, explains Richard Wyatt, executive officer for intramural research at NIH. “Quantification is a problem for us,” he admits.

But the problem is not simply one of counting heads. NIAID's active outreach is not part of an NIH-wide strategy. Rather, it's an outgrowth of a commitment by NIAID director Anthony Fauci, backed by the tireless efforts of Milton Hernandez, Asofsky's counterpart in the extramural research program. A cardiovascular physiologist who came to NIH in 1988 from Howard University Medical School in Washington, D.C., Hernandez takes the numbers personally: “I'm a Chicano from Texas, and when I got my Ph.D. in 1971 I knew every Chicano in U.S. medicine because there were less than 20 of us. Even today I would guess there's not more than 80.”

Hernandez promotes the careers of talented minority students by combining his responsibilities as the institute's training officer, its coordinator of minority supplements to investigators with existing grants, and the head of its minority predoctoral fellowship program. He says such triple duty “puts me in a unique position” among administrators at NIH's 25 institutes and centers to track the minority talent pool from grad school on up.

That personal touch is also in evidence at NSF, which offers one program to support minority postdocs in the biological sciences. It was begun a decade ago by George Langford, now a biology professor at Dartmouth College and a member of NSF's governing National Science Board, who saw a need for more minorities in academia and who persuaded his bosses within the biology directorate to back his idea. Despite its longevity, it has never spread to NSF's five other research directorates or to the education directorate, and the current program officer, Carter Kimsey, admits that she tries to keep “a low profile” in an effort to avoid upsetting foes of affirmative action. Last year those critics forced NSF to drop the minority component of its graduate research fellowships; they had already succeeded in getting NIH and NSF to cancel two precollege summer programs aimed at minority students (Science, 2 January 1998, p. 22).

Both current postdoc programs are run so quietly that they are often a secret even to those trying to expand the number of minority scientists—as well as to the target population. Physiological ecologist Robert Jackson is Web master for a site that lists career resources for members of his professional society, including opportunities for minorities (www.botany.duke.edu/jackson/ecophys). But Jackson, an assistant professor of botany at Duke University who is white, was taken aback when a reporter told him this summer about the 10-year-old NSF program. “I never knew it existed, and neither did another NSF program manager whom I asked,” says Jackson. Likewise for Ghislaine Mayer, a Haitian-born cell biologist, who starts work this month at NIAID in Louis Miller's laboratory of parasitic diseases after graduating this summer from the Albert Einstein College of Medicine. “I'm familiar with other minority programs,” she says. “I just didn't know that NIH had one for postdocs. And I didn't know I was part of it.”

In fact, most researchers don't think such minority postdoc programs are needed, and few universities promote them. University administrators are “very hands-on” when it comes to promoting diversity among undergraduates and professional school students, says biologist David Burgess, academic vice president and dean of faculty at Boston College, who is also president of the Society for the Advancement of Chicano and Native American Scientists. “But those same administrators traditionally have been entirely hands-off on graduate admissions and the hiring of postdocs. Which is ironic, of course, because that's the future of academic science.”

That “tradition” may grow from the fact that postdocs are hired by researchers themselves, and that most scientists believe an effort to take race into account would distort the process of hiring the most qualified candidates. “We definitely try to recruit the best minority students,” says Cliff Tabin, a professor of genetics at Harvard Medical School in Boston and co-director of the school's graduate program, “and on a faculty level we think it's important to have role models for minority students.” But when it comes to postdocs, he says, “to me diversity means having someone in my lab who's a biochemist, a virologist, who does surgery on embryos, and who works with Drosophila. Having people from different countries and cultures is nice, too. But I don't pay attention to race.”

Tabin and other faculty members believe strongly that increasing the number of underrepresented minorities “is everyone's business.” But they don't see it applying to postdocs. “That's not something that's addressed when it comes to hiring them,” says Theresa Compton, a virologist at the University of Wisconsin, Madison, medical school. “I think it falls between the cracks.”

Asofsky certainly doesn't dispute the importance of scientific excellence in choosing postdocs. But he'd like to see scientists move beyond “thinking about checkable boxes” when they weigh an applicant's merits, a practice that he and others say favors those students with the best academic pedigrees: “The problem is that most scientists without experience working with Ph.D.s from lesser known schools don't have a clue how to assess their talents.” Rice University mathematician Richard Tapia, who trains a large number of minority and women Ph.D.s, says he's struggled to place many of his students in top labs for that very reason. “They don't look as good on paper as somebody from Berkeley or Harvard,” he says. “But the standard criteria don't reflect what they have had to overcome or their determination to succeed.”

Hernandez may be one of the few science administrators who probe those cracks for hidden scientific talent. And he wishes he had more company. “It doesn't require a lot of time, but you need to make the commitment,” he says. The solution is equally obvious for minorities, he adds. “They need to finish [school], and then they need people like me to help them connect. That's all it takes.”

28. WORKING CONDITIONS

# A Day in the Life of a Topflight Lab

1. Gretchen Vogel

The work day lasts well into the night, but the rewards are considerable for postdocs in Robert Langer's chemical engineering laboratory at MIT

CAMBRIDGE, MASSACHUSETTS—If Robert Langer's chemical engineering laboratory at the Massachusetts Institute of Technology (MIT) were an independent company, it would dwarf many of the biotech start-ups in the Cambridge area. But there's a big difference: None of the workers get stock options, high salaries, or other lucrative financial inducements. Instead, they get paid in a different coin of the realm—the chance to publish in the world's top journals, and an edge in the race to become an academic top dog themselves.

Langer's 20 postdocs, 15 graduate students, three or four visiting professors, and two dozen undergraduates are spread out over most of the third floor of MIT's Whitaker Health Sciences Building, and they spill over into offices in the neighboring building. The labs themselves bustle with people measuring out reagents in fume hoods, calibrating instruments that measure the strength of a polymer film or, off in a sterile corner, injecting rat brains with new materials designed to improve drug delivery.

On a day-to-day basis, the lab does run a bit like a company, with Langer as the president and CEO of a $3 million to$5 million a year enterprise and the postdoctoral fellows and senior graduate students as research directors who oversee other members of the lab. Although he didn't plan it that way, Langer says the analogy isn't too far off. “I come up with the general ideas and raise the money,” he says, which comes from the National Institutes of Health, the National Science Foundation, and an array of biomedical and pharmaceutical companies. The specific projects and the day-to-day benchwork are the province of postdocs and students. The lab works on an endless variety of projects, from drug delivery to artificial organs. Among the lab's recent papers are pharmaceutical-dispensing microchips (Science, 29 January, p. 619) and artificial arteries (Science, 16 April, p. 489).

Postdocs, Langer says, are a crucial part of those projects; he estimates they do perhaps half of the lab's work while accounting for only slightly more than a third of the personnel. Indeed, a review of the last 6 years shows that 51% of the 246 papers from the lab carried the name of a postdoc as first author. Besides benchwork, he says, postdocs train and supervise junior lab members, help write grants and papers, and give talks at meetings.

Each postdoc sets his or her own hours, but several say it isn't unusual to work 12-hour days and 4 or 5 hours over a weekend. On a typical day the early risers are in the lab by 6:30 or 7 a.m., and the lab starts to fill up around 8:30 a.m. There is no discernible lunchtime lull, and the desktop shakers and centrifuges don't fall silent until nearly 8 p.m. A sign on the main office door asks people to lock up if they're the last one out, which often isn't until the wee hours of the morning. “A lab without coffee doesn't run as efficiently,” quips postdoc David Putnam, who says some of his best ideas have come from 2 a.m. gab sessions with other postdocs or students in the lab. The long hours are self-imposed, says postdoc Eric Crumpler, or dictated by a particular project. “There isn't peer pressure to spend time in the lab,” he says.

There is also the relentless pressure of the job market. “Publish or perish trickles down to us as well,” says Putnam. And tenure-track jobs are harder to come by, notes Maria Rupnick, a former postdoc, now a research associate overseeing several projects in Langer's lab. “It used to be if you went to Harvard or MIT, then your ticket was written. But these days you can work at a world-renowned lab and still not get the job of your choice. Doing a postdoc in Bob's lab gives you a leg up, but the pressure is still there.”

Although the exact contribution of postdocs is difficult to measure, in some laboratories it is undoubtedly much higher than Langer's estimate. An unofficial survey at Harvard and its medical school, home for about 2500 postdocs, offers some striking evidence. In pathologist Peter Howley's laboratory, which studies human papillomavirus, all but one of the 30 papers published between 1993 and mid-1999 lists a postdoc as first author. In developmental geneticist Douglas Melton's lab, a postdoc was first author on 75% of the papers published between 1990 and 1999. In molecular biologist Tom Maniatis's lab, 70% of the papers in the last 6 years carry a postdoc at the lead slot. And what's true for elite institutions applies to the journals that publish their work. A survey of the research articles in two recent issues of Science, for example, found that 43% of the first authors were postdocs.

Authorship on important papers isn't the only way of reaping credit for good work. It's not unusual for Langer's postdocs and senior graduate students to give talks at major meetings and to be interviewed by the media. Research associate Prasad Shastri, who has just been promoted from postdoctoral fellow, recalls speaking with reporters from New Scientist, NOVA, and the Discovery Channel when his work on electrically stimulated nerve regeneration attracted attention last year. “[Langer] doesn't think the project is successful just because of him,” Shastri says.

Langer says he has always encouraged lab members to take the initiative. “I'm not someone who likes giving orders,” he says. “I give people a very open environment. … I'm just there to act as a guide.” That freedom, he says, helps his postdocs learn how to think critically: “I want to get [postdocs] to the point where they're asking questions rather than looking for answers.” Langer says he learned that skill from his own postdoc adviser, cancer researcher Judah Folkman of Children's Hospital in Boston. “In Judah's eyes, almost anything is possible. He has ideas about everything.” For his part, Folkman says he and Langer both place a high priority on “increasing the scientific self-confidence of the young scientists” who work in their labs.

Rupnick says Langer has definitely shaped her ability to formulate scientific questions. When she arrived in the lab, she says, she went to him and asked what he'd like her to work on. “What excites you?” was Langer's reply. After tossing around a few ideas, Langer told her to “go think some more.” For two and a half months, she says, Langer would tell her only to do what would make her happy. “The level of frustration was just enormous—and it was perfect,” she says. After finally settling on an idea, and winning Langer's approval, she was ready to pour her heart and soul into it. “He pushes you out of your comfort zone. That's how you develop a scientific ego.”

Such freedom has its downside, however. “People sometimes end up reinventing the wheel” for part of a project, Rupnick says. “And if you tend to ask big questions, 2 or 3 years can go by without a publication.” Several postdocs in Langer's lab say they collaborate on several secondary projects as an insurance policy against the failure of their primary, high-risk project.

Even so, Rupnick says she much prefers Langer's style of mentoring. “I have been in [other] labs where the mentor saw you as an extension of her ideas and as a means of accomplishing significant aims 1, 2, and 3 in grant A,” she says. “In those labs, where the ideas don't flow that easily and are not tested that rigorously, the end product is a junior faculty member who doesn't ask the big questions. Or if they think of the big questions, they don't have the scientific ego to go after them.”

Delayed rewards.

Recent study shows that even Harvard Ph.D.s from the graduate school of arts and sciences take a long time to move from postdocs to tenured faculty jobs.

View this table:

One of Langer's talents is spotting those who are up to the challenge, Shastri says. Langer has plenty of candidates to choose from—he estimates that he gets 1000 inquiries a year for five or six slots. “We are picky,” he admits, and he looks for more than top academic qualifications. “People who thrive here are people who want to be independent,” he says. He relies heavily on recommendations from an applicant's Ph.D. supervisor, looking for people whose advisers say they are the best they've ever seen.

David Putnam says he sought to distinguish himself from the throng of applicants by persuading the department secretary at his graduate school, the University of Utah, to let him chauffeur Langer to and from the airport when he came to speak at the campus. That gave him a captive audience to discuss his ideas. Although the car ride alone didn't persuade him, Langer says that a combination of good grades, a strong Ph.D. project, and an excellent recommendation put Putnam ahead of other candidates.

For those who make the cut, the chance to work in Langer's lab can provide a tremendous boost into the job market. “He's spawned some very brilliant careers,” says former postdoc Marsha Moses, now an assistant professor at Children's Hospital at Harvard Medical School. Indeed, Langer's former postdocs populate top-tier universities across the country, and several head their own biotech start-up companies. Many still collaborate with Langer.

What postdocs don't get from Langer is day-to-day advice on the details. “There are some mentors who will help you very much with what to put in your gel, or how many animals to use. Bob is not good at that,” says Rupnick. Indeed, as one of 60 lab members, a postdoc can go for several weeks without talking with Langer in the lab, although they say he's almost always available if they need him. “He's amazing about returning calls,” Rupnick says—even if it's 11:30 at night and he's calling from a plane on the way to Japan. Lab members also have an open invitation to schedule a meeting in his appointment book.

Langer also extends his mentoring beyond the lab—in particular, to a pickup basketball or softball game, or to his annual party at a beach house on Cape Cod. Both current and former lab members speak warmly about the all-day gathering that features sand, sun, and science. “It's a kind of think-tank vacation,” Rupnick says. “He definitely uses the occasion to say, ‘How are you, how's work going, have you thought of this?’” But the result in no way resembles the forced camaraderie of a dreary company picnic, she says. “Science to him IS a good time. So it's not surprising that they mix.”