# News this Week

Science  30 Mar 2012:
Vol. 335, Issue 6076, pp. 1550
1. # Around the World

1 - Mariana Trench, Western Pacific Ocean
World's First Solo Dive To Challenger Deep
2 - Strasbourg, France
3 - Moscow
No Second Life for Russian Satellite
4 - Washington, D.C.
Last Days for Cancer Gene Patents?

## Mariana Trench, Western Pacific Ocean

### World's First Solo Dive To Challenger Deep

On the morning of 26 March, movie director James Cameron completed the world's first solo dive to the deepest part of the world's oceans—an 11-kilometer-deep valley within the Mariana Trench known as Challenger Deep, 322 kilometers southwest of Guam. Cameron's submersible Deepsea Challenger was outfitted with biological and sediment sampling equipment; temperature, salinity, and pressure sensors; and still and video cameras to document the historic dive.

It took Cameron 2.5 hours to reach the bottom, but only about 70 minutes to come back up. Leaking hydraulic fluid cut the planned 6-hour dive in half, and Cameron was only able to retrieve a partial sediment core from this lonely seascape. He reported an absence of the large jellyfish and anemones he saw during a test dive in the New Britain Trench off Papua New Guinea. Challenger Deep's fine-silt bottom, he noted, was devoid of any animal tracks. Cameron's descent into Challenger Deep marks the second such manned attempt after late Swiss engineer Jacques Piccard and retired U.S. Navy Captain Don Walsh touched down in their bathyscaphe Trieste in 1960.

## Strasbourg, France

### Leading French Institute Director Ousted

A long-running dispute at one of France's leading biology institutes has culminated in the removal of the director. Developmental biologist Olivier Pourquié, who has led the Institute of Genetics and Molecular and Cellular Biology (IGBMC) in Strasbourg since 2009, was informed last month that his term as director would end on 1 June. The decision to remove Pourquié—taken by the presidents of the institute's funders, CNRS, INSERM, and the University of Strasbourg—follows months of internal rancor, including anonymous e-mails, libel accusations, and attempts to calm the waters with an external mediator.

Pourquié says he tried to bring transparency and a modern governance structure to the institute, but was blocked by a subset of researchers who felt threatened by the changes.

“Pourquié is a scientific star. This was why he was chosen,” says Iain Mattaj, director general of the European Molecular Biology Laboratory in Heidelberg, Germany, who is head of IGBMC's scientific advisory board. But “he had no leadership experience and little experience of the French system.” The institute lacked rules for governing itself, he adds, and “putting [those rules] in place is essential prior to appointing a successor.” http://scim.ag/Pourquie

## Moscow

### No Second Life for Russian Satellite

Russian engineers intentionally crashed the telecommunications satellite Express-AM4 on 25 March, despite 11th-hour pleas from a company hoping to salvage the spacecraft for Antarctic communications.

The satellite launched 18 August, but mechanical failures left it in a too-low and useless orbit. The satellite was still functional, but the Russian Satellite Communications Company's chief financial officer, Dennis Pivnyuk, said earlier this month at the Satellite 2012 conference in Washington, D.C., that the Russian government had considered—and rejected—numerous salvage plans.

Polar Broadband Systems Ltd. had proposed one of those plans, crafting a new orbit for the satellite that they said would provide 16 hours of coverage to the National Science Foundation's (NSF's) South Pole Station. Although they said NSF expressed interest, the company was ultimately unable to obtain the spacecraft from the Russian State Commission.

Engineers at Astrium, which designed the spacecraft, fired its engines to begin a controlled descent at 6:33 a.m. EDT on Sunday. The debris landed a few hours later in the northern Pacific Ocean.

## Washington, D.C.

### Last Days for Cancer Gene Patents?

In the past 10 days, the U.S. Supreme Court has delivered a subtle, two-part blow that cuts the ground from under some hotly contested biotech patents, including those on breast and ovarian cancer genes BRCA1 and BRCA2. On 20 March, the court voted unanimously to reject patents for a test to monitor the anti-inflammatory drug azathioprine, held by Prometheus Laboratories of San Diego, California. The test, according to the opinion penned by Justice Stephen Breyer, was based on standard human responses that are like laws of nature—and laws of nature cannot be patented. The second blow came on 26 March when the court issued a terse announcement saying it would “vacate,” or dismiss, a decision by a lower court that upheld the validity of the BRCA1 and BRCA2 patents. The Supreme Court sent the case back down to the Court of Appeals for the Federal Circuit in Washington, D.C., “for further consideration” in light of the Prometheus decision. Biotech patent expert Christopher Holman of the University of Missouri, Kansas City, School of Law says the justices may want the lower court to knock out those gene patents. http://scim.ag/genepats

2. # Newsmakers

Endre Szemerédi of Rutgers University in New Brunswick, New Jersey, and the Alfréd Rényi Institute of Mathematics in Hungary, has received the 2012 Abel Prize for mathematics, “for his fundamental contributions to discrete mathematics and theoretical computer science,” according to a prize committee announcement made 21 March. The 6 million kronor (1 million) prize has been given out annually by the Norwegian Academy of Science and Letters since 2003. Szemerédi's work, mainly in combinatorics (the study of arrangements of finite systems) and number theory, has shown that, as systems made up of discrete components—such as hyperlinked pages on the World Wide Web—get large, there is structure even when the system is utterly random. His most famous result is an eponymous theorem establishing the presence of arbitrarily long arithmetic progressions, the epitome of orderliness, within any sequence of integers that doesn't become infinitely sparse. The paper it appeared in “is a real tour de force,” says Terence Tao, a mathematician at the University of California, Los Angeles, who was on the committee that chose this year's winner. http://scim.ag/Szemeredi ## Obama Nominates Dartmouth President to Head World Bank President Barack Obama surprised many people on 23 March by nominating Dartmouth College president and global health expert Jim Yong Kim to head the World Bank. “It's time for a development professional to lead the world's largest development agency,” said Obama in a speech at the White House's Rose Garden. Kim is a physician and anthropologist who co-founded Partners in Health with Paul Farmer and later headed the HIV/AIDS program at the World Health Organization (WHO). His work with Partners in Health included launching a pioneering program for treating multidrug-resistant tuberculosis in Peru. At WHO, he led a program called 3 by 5 that pushed for the goal of having 3 million HIV-infected people on antiretroviral drugs by 2005. The project did not hit its target, but more than 6 million people now have access to those lifesaving drugs. The World Bank is expected to elect a new president when it holds a meeting on 21 April. The United States traditionally has selected presidents for the bank, but there has been an increasing call for transparency, and Kim's nomination does not guarantee him the job. http://scim.ag/jimyongkim 3. # Random Sample ## Frozen in Time Earth scientists aren't going to let a unique collection of glacier photographs melt away. This month, glaciologist Matt Nolan of the University of Alaska, Fairbanks, won a grant from the National Science Foundation to start digitizing and preserving about 100,000 large-format aerial photographs of glaciers in Alaska, Canada, and Washington state taken between 1960 and 1995. “It is quite simply the best and most comprehensive record of these glaciers over that time span,” says Nolan. “It was the satellite data of its time.” Most of the negatives were made by Austin Post, a photographer legendary among ice scientists for his sharp eye and insight. The now-retired Post never got a college degree, but in 2004 was awarded an honorary doctorate for his work, mostly done for the U.S. Geological Survey. Ultimately, Nolan hopes to revisit the glaciers documented by Post—such as this glacier near Summit Lake, Alaska, shown above in 1970—and shoot updated portraits to see how the rivers of ice have changed. ## The Science of Judo Every form of exercise uses a different combination of the body's metabolic systems for energy. It's relatively easy to parse out those combos for cyclical sports like running and cycling using exercise machines in a laboratory. But that's harder to do with more unpredictable sports such as martial arts. So a team of Brazilian researchers have taken the lab into the dojo to study the energy requirements of the Japanese art of judo. Aerobic metabolism does most of the work during long periods of exercise, such as running a marathon. For shorter, more intense exertion, or when the oxygen runs out, muscles can break down sugar anaerobically. And for very short bursts of energy, muscles can use energy-storing compounds, called phosphagens, in muscular tissues. To assess how the body gets energy during a more unpredictable sport, Emerson Franchini, a physiologist at the University of São Paulo in Brazil, and his team outfitted judo practitioners with a portable physiology lab: a mask attached to a device worn on the torso that analyzes gases in the martial artist's breath and measures the pulse. The athletes tried different kinds of throws, and also sparred at high intensity. Metabolically, judo turned out to be a mix of aerobic sports like running and anaerobic sports like weightlifting, the researchers report online this month in the Journal of Visualized Experiments. “Normally, judo is thought to be highly lactic” because it requires intense bursts of energy, Franchini says. But the data revealed that phosphagen metabolism was crucial for throwing people, and aerobic metabolism was also higher than expected. Franchini says the findings should help judo teams train. By knowing their energy expenditures, for example, martial artists can better customize their diet. http://scim.ag/judomet ## By the Numbers 1.875 million joules Record-breaking amount of ultraviolet laser light delivered to the target chamber of the National Ignition Facility at Lawrence Livermore National Laboratory on 15 March. £ 100 million New investment pledged 21 March by U.K. Chancellor George Osborne to boost science and engineering research in U.K. universities. ## ScienceLive Join us on 5 April at 3 p.m. EDT for a live chat with experts on the Comprehensive Nuclear-Test-Ban Treaty. http://scim.ag/science-live 4. Psychology Research # Psychology's Bold Initiative 1. Siri Carpenter* In an unusual attempt at scientific self-examination, psychology researchers are scrutinizing the reproducibility of work in their field. Pick up the January 2008 issue of Psychological Science, turn to page 49, and you'll find a study showing that people are more likely to cheat on a simple laboratory task if they have just read an essay arguing that free will is an illusion. It was a striking study that drew widespread attention both from psychologists and from many media outlets. But should you believe the result? There's no reason to think that the study, conducted by psychologists Kathleen Vohs of the University of Minnesota Carlson School of Management in Minneapolis, and Jonathan Schooler, who is now at the University of California, Santa Barbara (UCSB), is incorrect. Yet according to many psychologists, their field has a credibility problem at the moment, and it affects thousands of studies like this one. Part of the angst stems from recent high-profile cases of scientific misconduct, most dramatically the extensive fraud perpetrated by Dutch social psychologist Diederik Stapel (Science, 4 November 2011, p. 579), that have cast a harsh light on psychological science. Yet there is no evidence that psychology is more prone to fraud than any other field of science. The greater concern arises from several recent studies that have broadly critiqued psychological research practices, highlighting lax data collection, analysis, and reporting, and decrying a scientific culture that too heavily favors new and counterintuitive ideas over the confirmation of existing results. Some psychology researchers argue that this has led to too many findings that are striking for their novelty and published in respected journals—but are nonetheless false. As a step toward testing that disturbing idea, one project begun this year offers an online site (PsychFileDrawer.org) where psychologists can quickly and easily post, in brief form, the results of replications of experiments—whether they succeed or fail. University of California, San Diego, psychologist Hal Pashler, one of the project's developers, says the goal is to counteract the “file drawer problem” that plagues all of science, including psychology; researchers usually just file away straightforward replication studies because most journals decline to publish such work. In an even more daring effort, a group of more than 50 academic psychologists, which calls itself the Open Science Collaboration (OSC), has begun an unprecedented, large-scale project to systematically replicate psychological experiments recently published in leading journals. “We're wringing our hands worrying about whether reproducibility is a problem or not,” says psychologist Brian Nosek of the University of Virginia in Charlottesville, who is coordinating the effort. “If there is a problem, we're going to find out, and then we'll figure out how to fix it.” Robert Kail, a Purdue University developmental psychologist and editor of Psychological Science—one of the three journals whose papers the OSC is attempting to replicate—is optimistic that a high percentage of published findings will be replicated. Nonetheless, he views the field's recent attention to the issue of false positives as healthy. “There has been a lot of speculation about the extent to which it's a problem,” he says. “But nobody has actually set it up as an empirical project. It's a great thing for somebody to actually do that.” Schooler, who is not directly involved with the project but whose free will study will be replicated, considers the OSC replication study a “bold initiative.” Yet he's concerned that if the project confirms few studies, it could unfairly indict psychology. “I think one would want to see a similar effort done in another area before one concluded that low replication rates are unique to psychology,” he says. “It would really be a shame if a field that was engaging in a careful attempt at evaluating itself were somehow punished for that. It would discourage other fields from doing the same.” Indeed, the prospect of exposing psychology's foibles has upset some scientists. “I had a senior person in the field ask me not to do it, because psychology is under threat and this could make us look bad,” Nosek says. “I was stunned.” The point of the project, he says, is not to single out individual studies or disparage psychology. “We're doing this because we love science. The goal is to align the values that science embodies—transparency, sharing, self-critique, reproducibility—with its practices.” ## Am I doing something wrong? Reproducibility is supposedly a basic tenet of science, but a number of fields have raised concerns that modern publishing pressures inhibit replication of experiments. In a well-known 2005 PLoS Medicine essay, epidemiologist John Ioannidis, now at Stanford University in Palo Alto, California, argued that in biomedicine, many if not most published research claims are false. He outlined a number of factors—including small sample sizes, small effect sizes, and “flexibility” in the research process—that contribute to a high rate of false positives. One reason those false positives aren't caught is because of a lack of emphasis on replication studies, which is “standard across fields of science,” says Columbia University statistician Victoria Stodden, who studies reproducibility and openness in computational science. Nosek first became interested in the problem of replication in psychology several years ago, after he started having persistent problems confirming others' results—something his lab routinely does before extending research to address new questions. Believing himself to be a careful methodologist, he wondered whether something was wrong with the studies he was attempting to reproduce. For example, many social psychological studies have asked participants to unscramble sentences containing words that connote a particular concept that the researchers aim to “prime” in the participants' minds—say, impulsivity, or anger, or happiness; the idea is to test the effects of those mental constructs on a subsequent task. The scrambled-sentence task has been used in many published studies, but Nosek's group has rarely been able to get it to work, and he suspects that the effect may be much more limited than the published literature would suggest. The idea of systematically testing the reproducibility of psychological science percolated in Nosek's mind until late last year, when revelations of Stapel's scientific misconduct brought the issue to a boil. Stapel, whose studies were widely cited and had drawn frequent media attention, admitted last fall to fabricating data on more than 30 studies dating back to the mid-1990s. One of those high-profile studies, published in Science and now retracted, concluded that chaotic physical surroundings promote stereotyping and discrimination. News of Stapel's fraud led many psychologists to question whether the field possesses sufficient checks and balances to prevent such willful deception. It also stirred some researchers' nascent worries that—rare acts of fraud aside—commonly accepted practices in psychology research might produce an unacceptably high number of false positive results. “It just became obvious that it was time to do some science to figure it out,” Nosek says. In November 2011, Nosek approached a few departmental colleagues to propose a collaborative effort to study the field's reproducibility. The effort soon expanded to include researchers from universities in the United States, Canada, and Europe. This winter, OSC members began identifying studies to include in their replication sample. They chose three high-impact psychology journals—Psychological Science, the Journal of Personality and Social Psychology, and the Journal of Experimental Psychology: Learning, Memory, and Cognition—and began logging key details of the first 30 articles published in each journal in 2008. They reasoned that articles published during this time frame are recent enough that most original authors can find and share their materials, yet old enough for the OSC to analyze questions such as whether a study's reproducibility correlates with how often it has been cited subsequently. Many psychology articles include more than one study to support their conclusions, but the OSC decided in advance to select only the final study from each article; if that study was not feasible to replicate, they worked backward until an eligible study was identified. After identifying eligible experiments, individual OSC members began choosing the ones each would repeat and contacting the original authors to obtain materials and fill in methodological details. So far, the response from original authors has been overwhelmingly positive—only two have declined to provide materials needed for replication. More than 30 replication studies are now under way, and the group aims to conduct at least 50. (New researchers can join the collaboration anytime.) The results OSC members will seek to reproduce vary widely, from a study of how photos of unsmiling black men automatically activate attention in some people's minds, to a study that examined how the timing in which perceptual stimuli are presented affects short-term memory. The replication studies are being funded by the OSC researchers who select them; however, in keeping with the samples used in the original studies, most will be cheap because they'll use undergraduates who participate for course credit—although using study populations that are mostly limited to young, white college students from Western industrialized cultures has its own drawbacks (Science, 25 June 2010, p. 1627). ## Repeat after me While reproducibility is often held up as the “gold standard” in science, Nosek argues that “direct replications,” in which researchers follow an original experiment's procedure as closely as possible, are rare. “There is no incentive for replication—it's all about the new idea,” he says. Not all psychologists agree. Social psychologist Norbert Schwarz of the University of Michigan, Ann Arbor, says that although the OSC project is valuable, concern about the field's robustness may be overblown. Direct replications may be rare, but Schwarz points out that conceptual replications—in which researchers tweak a study's materials or procedure to test a hypothesis in a different way—make up the bulk of psychology's literature and “are highly valued because they look at the stability of a phenomenon across different content domains.” Nosek agrees that conceptual replications are important and common, and temper worries that journals are littered with false findings. But conceptual replication assumes that the replication addresses the same phenomenon as the original demonstration, which may not always be the case, he says. Direct replications are important, he and others say, because psychology, like other disciplines, has practices that may lead to too many false positives. In the November 2011 issue of Psychological Science, University of Pennsylvania psychologist Joseph Simmons and colleagues showed, through computer simulations and actual experiments, that “flexibility” in research decisions such as how many research subjects to include in a study, how many outcomes to measure, or whether to break down analyses according to participants' gender can more than double the chances of getting a false positive. When several such “researcher degrees of freedom” are in play, as is commonly the case, a study is more likely than not to mistakenly yield a statistically significant effect. Nosek believes the prevalence of research practices that lead to unintended bias is rooted in the fact that negative results are virtually unpublishable—a state of affairs that has grown more extreme in recent years, especially in psychology. “It is important to my career that I do studies that are publishable,” he says. One way to do that is to capitalize on the kinds of practices that Simmons and colleagues identified. Another is to run many studies with small numbers of subjects and publish the ones that “work.” A third and especially insidious problem, Nosek says, is that researchers can easily fool themselves into believing that chance positive results are actually what they had hypothesized all along—then publish such findings as though they were confirmatory tests of existing hypotheses. There is currently nothing to prevent such “motivated reasoning,” as psychologists call it, from contaminating the scientific literature. ## Looking ahead Nosek isn't just depending on OSC's look at past studies to improve psychology research practices. He's also looking forward and has developed, with his graduate student Jeffrey Spies, an online database (openscienceframework.org) where psychological scientists can easily organize and—if they choose—register study materials, hypotheses, planned analyses, and data. Nosek hopes that in addition to aiding laboratory workflow and providing an outlet for results that might not otherwise see the light of day, the registry will increase researchers' sense of accountability both to their community and to themselves. Looking forward to using the registry for his own research, Nosek says he sees it as “a way of guarding the truth against my own motivated reasoning.” As for the OSC's replication project, data collection on the currently chosen studies should be completed by the end of this year, and Nosek hopes the results will be published not long after. One challenge for the OSC researchers will be in setting criteria for what constitutes a successful confirmation, given that a replication can take varying forms, from a result that has the same statistical significance as the original finding to a pattern of results that is merely in the same direction as the original. That makes it difficult to say what percentage of failed replications should be considered problematic. And it's one reason that Nosek believes the most interesting results of the study may lie not in the raw rate of reproducibility but in what factors predict a study's reproducibility. For example, if the studies that replicate best are also the most widely cited, that would suggest that despite biases in publishing practices, scientists can separate the wheat from the chaff. If studies that were published along with several others in a multistudy paper—a formulation that often involves multiple confirmations of the same basic effect, at least conceptually—replicate more often than those that were single-study papers, that might hint that “bite size” reports are problematic, as some critics have suggested. One limitation in interpreting the OSC's results stems from the fact that the group is not tackling a representative sample of all psychology research. Studies that used statistical analyses that cannot yield clear evidence of replication have been excluded, as have studies that would be infeasible to replicate because they require specialized materials, instrumentation, or participant populations. Furthermore, most studies included in the sample are drawn from cognitive and social psychology; other subfields, such as developmental and clinical psychology, neuroscience, and animal behavior, are not included. Stanford University social psychologist Nalini Ambady says several junior colleagues have told her they're worried about this disproportionate focus because if a high percentage of OSC studies fail to be replicated, many people may conclude that it is social psychology alone that is problematic. She sympathizes with that argument. “I think if you want to do it, then you should do a fair, representative sampling,” Ambady says. “Don't just focus on the social psychological studies that are the low-hanging fruit because they are generally cheapest and easiest to conduct.” The study's heavy focus on social and cognitive psychology is a reflection of the researchers who have gotten involved so far, Nosek responds: “If there are people in other subfields who want to join the project, we will be delighted to broaden it.” The OSC won't produce a “definitive study,” he stresses. “It will be a first estimate of the reproducibility of findings that are in important journals in psychology.” Columbia's Stodden agrees that psychology's efforts to address the issue shouldn't be cause for criticism. Psychologists' scrutiny “is very admirable,” she says. “I think other fields could benefit from that kind of self-reflection.” Cognitive psychologist Rebecca Saxe of the Massachusetts Institute of Technology in Cambridge, who is participating in the collaboration, is also optimistic. The project, she says, “has the potential to be spun as negative and nihilist. But to me, it's the opposite. Science is about discovering true things, and when you do find something that's true enough that others are able to replicate it, that's just thrilling.” • * Siri Carpenter, a freelance writer based in Madison, Wisconsin, worked 12 years ago in a lab with Brian Nosek, the organizer of the Open Science Collaboration. 5. Microscopy # New Lens Offers Scientist A Brighter Outlook 1. Angela Saini* Embittered by a past fight with a commercial microscope maker, a British scientist hopes to revolutionize imaging with a new microscope he will build and sell himself. Against a wall behind his lab bench, Brad Amos keeps a framed pencil sketch of a gnat larva. He drew the picture as a schoolboy in 1961, viewing the 2-millimeter-long critter through a small brass microscope made by the German optics company Carl Zeiss. “That was my first microscope,” says Amos, an emeritus research group leader at the U.K. Medical Research Council's Laboratory of Molecular Biology in Cambridge. It was far from his last. Amos went on to become one of the designers of the laser scanning confocal microscope, famous for producing crisp images of cells by cutting out the blurry flare that came from using standard fluorescence microscopes, a staple of modern cell biology. The laser confocal hit the market in 1986, and Amos guesses there must be between 5000 and 10,000 such devices in use today. He can't be sure, he says, because he lost control over the invention in the 1990s after a bitter patent dispute with Carl Zeiss. (Other microscope manufacturers, including Leica and Olympus, make confocals as well.) Now, at age 66, Amos has his eyes set on another revolutionary microscope, which he has worked on since 2000. It utilizes an invention he calls the “Mesolens,” a lens capable of capturing a similar level of detail as a confocal alone, but on a far larger scale. Where existing microscopes have lenses up to 3 centimeters long, the Mesolens is a massive half a meter, which means it can capture a specimen as big as a mouse embryo in a single shot. The Mesolens pushes other boundaries as well. One of the novelties of a confocal microscope is that it can view up to 0.22 micrometers below the surface of a specimen, provided it is translucent enough. When used inside a tailor-made confocal microscope, the Mesolens plumbs depths of up to 3 millimeters. The final effect is a 3D picture that allows one to zoom in as far as the level of the cell. (The name is derived from the Greek meso, or middle, because it's somewhere between a conventional microscope lens and a close-up photographic lens.) The microscope could save biologists the effort of patching together hundreds of small high-resolution images if they want a larger one. “It would probably take 1 day to image a mouse embryo with a standard confocal microscope using a stitching and tiling method. With the Mesolens, it takes only an hour,” says Amos's colleague Gail McConnell, a biophotonics professor at the University of Strathclyde in the United Kingdom. Amos has only a prototype for now, assembled in Strathclyde with the help of McConnell and John Dempster, an imaging software designer. But the device has already created ripples of excitement among scientists. Inquiries have been coming in ever since Amos revealed the first confocal pictures from the Mesolens at the Royal Society in London in early February. “I want to be his first customer,” says Scott Fraser, a professor of biology and an imaging expert at the California Institute of Technology in Pasadena, who is confident that the Mesolens will break new ground in microscopy. The microscope has obvious biomedical applications, Amos says, such as imaging cancerous tissues, but he has also been contacted by a geologist interested in using the Mesolens to study pores in rocks. ## Lucky encounter Few people know how tough the journey has been for Amos. In his early career as a biologist, he discovered the calcium binding protein, spasmin. But a professional crisis in 1980, following the death of a patron of his work, left him depressed and purposeless. A lucky encounter with John White, a developmental biologist with a background in physics and electronics who had already started work on the confocal microscope, prompted Amos's first foray into microscopy. “I saw John's prototype and I immediately saw how I could contribute to that,” he says, adding that he always “liked making things with my hands.” But then came another setback. After the confocal microscope went into production, he was hit with years of litigation by Carl Zeiss, which claimed that the patent on his design was invalid because aspects of it had been developed earlier. By the end of the 1990s, Amos had lost the case—and his invention. “It was a mess,” he says. But he did not lose heart. Instead, Amos began exploring the possibility of a microscope with an enormous lens. The problem was, the bigger it was, “the higher the angle through which the glass has to bend the light, and the greater the likelihood of aberrations,” he says. An astronomer at the Royal Observatory at Greenwich in London, who heard him lecture on the topic, warned that this was a challenge no off-the-shelf lens would be able to meet. The only option was to build it himself. Amos ended up working with Esmond Reid, a British freelance optical designer with expertise in giant telescopes. Together they tinkered with different lenses, trying to find one that would maximize the size of the image without sacrificing too much resolution. “I thought it might be possible, but I was no more than 50% confident,” Amos says. “The business of lens design is arcane. There is so much secret knowledge that isn't available to researchers like me but held by commercial manufacturers.” It took the pair a year to hit an acceptable balance. Another technical barrier, which emerged later, was making sure the laser beam that scanned specimens didn't produce a jittery picture. Modern lasers are reliably stable, but the mirrors inside the microscope that reflect these beams are less so. “One of the big things we were worried about was being able to program the mirrors to angle themselves precisely so that the laser would go back to the correct position every time,” McConnell says. The solution lay in large, high-quality mirrors usually used by the defense industry to scan landscapes. In January this year, McConnell finally took the first laser scanning confocal images of a mouse embryo using the Mesolens (see photo, above). “I was agog. It was a very elegant example that the system was actually going to work. It gave me a lot of confidence,” she says. The confidence extends beyond Amos's team. “When I visited his workshop and he showed me the initial prototype and the images, I was blown away,” says Rafael Yuste, a neuroscientist at Columbia University. He believes the Mesolens has potential for imaging “the activity of every neuron in the brain of an animal.” Fraser adds that it will have important applications in biotech as well, such as imaging large arrays of nucleic acids or proteins. ## Going it alone The obvious next step is to develop the Mesolens into a product that can be manufactured and sold. The conventional route would be to patent the invention before taking it to an optics company for factory production. But Amos intends to do neither. His battle with Carl Zeiss has soured his view of industry. “It was a rude awakening into the nature of companies, and I'm afraid I've never changed my opinion. It's a nasty business,” he says. In any case, no optics companies have approached him with an interest in producing the Mesolens. Instead, Amos plans to develop and manufacture the device himself. Even without patents, he is confident that no company has the research and development expertise to make a copy. He has started a company with Reid and has already made substantial investments from his own savings, even carrying out some of the technical work from a workshop in the garden behind his home. “I have two lathes, two milling machines. I can do precision metalwork from there,” he says. “I will grow the business by the bootstraps.” Not everyone thinks that's a good idea. “We need to rely on commercial enterprises to get the technology to the field,” Yuste says, “by leveraging their production and distribution capabilities to lower the cost of manufacturing.” Richard Ankerhold, vice president of product sectors in the BioSciences division of Carl Zeiss Microscopy, warns that small companies can face logistical problems when it comes to global support. “If there's a service call or if he has to install it outside the U.K. where he is based, it could get very difficult,” Ankerhold says. Amos argues that he can keep prices low by running his Mesolens business as close to a nonprofit as possible and relying on word of mouth for marketing. “I don't think it will cost very much more than an ordinary confocal microscope,” he says, which would be about450,000. “My hope is that we'll be manufacturing to order next year.” After all his turbulent years in the microscope industry, Amos seems to have finally found his focus.

• * Angela Saini is a writer in London.

# Partners Prepare to Pick a Site For World's Biggest Telescope

1. Daniel Clery

The Square Kilometre Array will be continent-wide in scale and push technology to the limit. There are two finalists for this €1.5 billion prize.

Astronomers in South Africa or Australia and New Zealand will soon be popping champagne corks in celebration of capturing one of the biggest prizes in science. It's the right to host the Square Kilometre Array (SKA), an instrument that will span a continent and comprise about 3000 radio-telescope dishes.

The winning bidder for this €1.5 billion project, which could be announced as soon as next week at a 3 April meeting in Amsterdam, will receive a huge boost to its high-tech industry and a calling card for attracting top academic talent to its shores. Not surprisingly, the two finalists in a process that began nearly 2 decades ago have been pressing the flesh and talking up their bids to representatives of the five countries that have voting authority.

Astronomers hope that the last lap in the race doesn't descend into a drawn-out political scramble that damages and delays their long-sought telescope. Indeed, those who spoke with Science didn't have strong views about which of the two candidate sites should be chosen. “It's very hard from a scientific perspective to choose one from the other,” says Chris Carilli of the U.S. National Radio Astronomy Observatory in Charlottesville, Virginia, who co-chaired SKA's science advisory committee.

The idea for a telescope array with a total collecting area of a square kilometer originated in a 1993 workshop. The next decade was spent building support in the community, identifying SKA's scientific goals, developing technology, and getting funding agencies on board. In 2006, four potential sites were narrowed down to two (Science, 18 August 2006, p. 910). Since then the candidate sites—Australia–New Zealand and South Africa teamed with eight other African countries—have been honing their bids and spending millions in the process.

Both finalists possess the two most important criteria for the site: a large, flat area that is as free as possible of any radio transmissions that might interfere with reception. “The RFI [radio frequency interference] environment is stunning at both sites,” Carilli says. SKA will have a 5-km-wide core containing about half the total number of antennas—both the familiar dishes plus a new type made of arrays of static dipoles. One-quarter of the antennas will stretch out from there in five curving arms out to about 180 km. The remaining antennas will extend the arms out to more than 3000 km. Such distances are needed to get high resolution. The signals from the distant dishes are combined by computer to form images with the same resolution as a single dish the size of the whole array.

The core site of the Australia–New Zealand bid sits in the shire of Murchison in the midwest region of Western Australia. Murchison is the size of the Netherlands but with only 110 residents. “We passed very stringent legislation to preserve radio quietness” at the Murchison site, bid leader Brian Boyle says.

The team has laid fiber-optic cable to the site, has constructed roads and buildings, and is installing renewable power supplies. The most visible demonstration of its commitment is funding a precursor telescope known as the Australian SKA Pathfinder. Its complement of 36 12-meter dishes is due to be completed this year.

The main site for the South African bid is in the Karoo region of the Northern Cape. Its farthest antennas will reside about 4500 km away in Ghana. The Karoo is, like Murchison, remote, flat, and dry. But its elevation—1000 meters above sea level compared with Murchison at less than 300—gives it a slight advantage for making observations at higher frequencies.

The bid team has also put in buildings, new roads to the site, and power and data cables. South Africa's new radio-quietness law “gives the science minister power over any development in a 500-km-diameter area around the site,” team leader Bernie Fanaroff says.

The team has so far built and is commissioning an array of seven dishes: the Karoo Array Telescope (KAT-7). Its dishes are constructed from carbon-fiber composite as a test of the sort of mass-production techniques that will be needed for SKA's many dishes. A larger precursor, the 64-dish MeerKAT, is in the pipeline.

The two finalists submitted their formal bids in September, and representatives of both teams attended a question-and-answer session in December with the Site Selection Advisory Committee (SSAC), a group of 12 independent experts. The committee delivered its recommendation to the SKA organization in February.

Although some 19 countries have been involved in planning SKA, only seven were ready to become members of a not-for-profit company that was set up in November in the United Kingdom to take legal responsibility for the project. The signatories have pledged €69 million of the €90 million needed for the 4-year preconstruction phase. Of these seven, the site candidates—namely, Australia, New Zealand, and South Africa—are excluded from voting on the site, leaving the final say-so to representatives from China, Italy, the Netherlands, and the United Kingdom. On 19 March, Canada became the fifth voting member.

Science has not seen SSAC's recommendation, but press reports suggest that it gave South Africa a slight edge. In any case, both teams are making sure their bids are seen in the best possible light, with press releases trumpeting every milestone, however minor. South Africa announced this month that KAT-7 had succeeded in observing neutral hydrogen in a nearby galaxy (a major goal for SKA), for instance, while Australia said that signals from telescopes in Australia and Korea had been successfully combined across a distance of 8000 kilometers.

On 19 March, the SKA board discussed the site selection committee's recommendation. The voting governments are weighing the two proposals before their next meeting on 3 April, but SKA organization director Michiel van Haarlem is not expecting an immediate decision. “It's going to take as long as it's going to take,” he says.

Researchers are hoping the wait won't be too long. They are eager to get started on the next phase of the project, which is to decide the final structure of the array and produce construction-ready plans by 2016. If construction indeed begins then, observations using 10% of the total array could begin in 2020.

The SKA organization has not finalized plans for the array, in part because they depend on the site chosen but also because astronomers were waiting for technology to catch up with their ambitions. When SKA was first imagined in the 1990s, the focus was on as large a collecting area as possible. But there was no way to affordably process and store all the data. “It needed a leap in technology. The computing resources didn't exist,” says Albert Zijlstra, director of the Jodrell Bank Centre for Astrophysics in the United Kingdom.

The initial goal of SKA was to detect light from neutral interstellar hydrogen that may allow astronomers to peer back into the “dark age” of the universe before it was populated with stars. The light is very faint, so a huge collecting area is needed. Astronomers soon realized that a telescope of that size could do other things, too. Early last decade, Carilli and Steven Rawlings, a U.K. astrophysicist who died earlier this year in unexplained circumstances, headed a panel that identified five key scientific goals: probing the dark age for the first stars and galaxies, searching for prebiotic molecules, investigating the origins of the cosmic magnetic field, testing gravity by observing pulsars around giant black holes, and studying galaxy evolution and dark energy.

These disparate targets require a multi-talented array, both sensitive (a big collecting area) and high resolution (widely spaced antennas). It will also have to span wavelengths ranging from 3 centimeters to 4.3 meters. Long wavelengths are a revived area of study for astronomers rejuvenated by recent advances in computing power. Instead of using dishes, astronomers are starting to use static arrays of simple dipole antennas on the ground.

These “aperæture arrays” pick up signals coming from all directions. Using fast signal processing, computers insert slight delays between the signals from one row of dipoles compared with the next, and a further delay on the one after that, and so on. In this way the processing “steers” the sensitivity of the array in a particular direction so that it only “sees” a certain patch of sky.

The clever trick with this computer-assisted beam-forming is that an aperture array can form multiple beams and look at many parts of the sky at once. This ability makes them incredibly fast at doing surveys, something that dishes struggle with. “The survey speed will be 10,000 times faster than now,” says Michael Kramer of the Max Planck Institute for Radio Astronomy in Bonn, Germany. SKA will be able to survey all the sky visible from the site once a month. That feature will allow astronomers to spot changes in the radio sky as never before.

Beyond achieving SKA's five goals, astronomers dream of discovering something completely unknown. Says Kramer: “We're pushing parameter space so much with SKA, we're bound to discover something we don't know about yet.”