News this Week

Science  13 Mar 1998:
Vol. 279, Issue 5357, pp. 1631

    New Role for Estrogen in Cancer?

    1. Robert F. Service

    Although estrogen was supposed to act mainly as a growth factor in promoting cancers, new work suggests that products it forms in the body may also cause the initiating mutations

    The link between the female hormone estrogen and cancer is hard to miss. Both epidemiological and cell biology studies have indicated that it contributes to the development of three of the top five cancers of women—those of the breast, uterus, and ovaries—which together account for an estimated 240,000 new cancer cases a year in the United States alone. Now there is new evidence that estrogen's involvement in these cancers may be much deeper than was thought.

    Natural causes?

    Natural estrogens are metabolized into a variety of compounds, including some (in red) that may trigger cancer.

    Researchers have long known that the hormone is a powerful stimulator of cell proliferation. The common belief has been that cell growth promotes cancer development by increasing the chances that a cell bearing a potentially cancer-causing mutation will multiply. As for the initial genetic damage itself, researchers have tended to attribute it either to spontaneous mistakes in DNA replication as cells divide or to damage triggered by external sources: environmental chemicals, such as those in cigarette smoke, or x-rays and other forms of radiation. But the new evidence points to another culprit originally fingered as a suspect years ago: the metabolic byproducts formed by estrogen in the body.

    Cell culture studies show, for example, that estrogen metabolites can bind to DNA and trigger damage. The same compounds also produce cancer in lab animals. And recent epidemiological work suggests that women who have reduced amounts of the enzymes that help sop up those reactive estrogen byproducts are at higher risk for developing breast cancer. Taken together, these studies provide “pretty good data that, by themselves, hormones can be complete carcinogens,” says David Longfellow, a chemist with the National Cancer Institute in Bethesda, Maryland, who is organizing a conference on the topic, set to take place next week. If the data hold up, they could force researchers to take a closer look at the role of hormones in the body, and ultimately offer new avenues for cancer prevention by removing potentially damaging compounds before they build up.

    But there is likely to be plenty of debate at Longfellow's conference. The view that hormones initiate as well as promote cancer remains far from universal. “This has been hugely controversial,” says Craig Jordon, a pharmacologist at Northwestern University Medical School in Evanston, Illinois. “It's an interesting possibility,” adds Jan-Ake Gustafsson, an estrogen metabolism expert at the Karolinska Institute in Huddinge, Sweden. “But the hard data are still lacking.”

    Disputes rage within the ranks of true believers as well. Estrogen produces several metabolites, and competing research teams—each backing its own horse in the race—argue for different metabolites as being the most important gene disrupter. “It's a mess,” says Patricia Thompson, a molecular biologist at the National Center for Toxicological Research (NCTR) in Jefferson, Arkansas. “But it's a very hot area, and some people feel very strongly about what they think is going on.”

    Starting from scratch

    As early as 20 years ago, researchers recognized that estrogen's growth-stimulating effects could make it a cancer promoter. Rats, for example, more often develop mammary tumors if they are subjected to both a chemical carcinogen and estrogen than to either alone. That picture also fits with epidemiology studies showing that women who are exposed to estrogens for longer—either through early onset of menstruation or late menopause—have an increased risk for developing breast cancer.

    Alarming numbers.

    Estrogens are linked to three of the most common cancers in U.S. women.


    Proponents of the estrogen metabolite hypothesis don't doubt that estrogen promotes cancer growth. Those effects “are clearly very important,” says James Yager, a toxicologist at the Johns Hopkins University School of Hygiene and Public Health in Baltimore. But when it comes to understanding just what causes the initial DNA damage, “the conventional model is just hand waving,” says Leon Bradlow, an endocrinologist and estrogen metabolite hypothesis supporter at the Strang Cancer Research Laboratory in New York City. Beginning in the mid-1970s, however, research showed that estrogen metabolites affect the levels of enzymes designed to clear out active compounds that might initiate cancer. That suggested another possibility: the metabolites themselves might be involved. In this possible mechanism, says Bradlow, “we have a more cogent case.”

    The suspects

    Just what that mechanism is depends on who you listen to, however. The list of possible suspects is long. For example, the principal natural estrogen, estradiol, is transformed in cells into three main components: 2-hydroxyestrone (2-HE), 4-hydroxyestrone (4-HE), and 16-α-hydroxyestrone (16-α). These can react directly with DNA or be metabolized further into other reactive compounds.

    Most researchers agree that 2-HE isn't dangerous, because it fails to show up as a mutagen in cell culture studies or as a carcinogen in animals. Instead, proponents of the metabolite hypothesis fall into two main camps, one blaming 4-HE and the other 16-α.

    For members of the first camp—led by Joachim Liehr of the University of Texas Medical Branch in Galveston and Ercole Cavalieri of the University of Nebraska Medical Center's Eppley Institute for Research in Cancer and Allied Diseases—one touchstone is a 1986 experiment Liehr and his colleagues performed with male Syrian golden hamsters. These animals are unusual in that their kidneys express the receptors through which estrogens turn on cell growth.

    When the researchers exposed the hamsters to a synthetic estrogen that is an even more potent activator of the receptor than the natural hormone, relatively few of the animals developed kidney cancers. But nearly all hamsters given 4-HE, which still binds to the estrogen receptor but is a less active growth stimulator, developed tumors within 6 months. “That tells me receptor-mediated binding cannot be the sole process responsible for tumor induction,” says Liehr.

    Later studies have shown that 4-HE is produced at the scene of the crime—the tissues where estrogen-linked cancers develop. Most estrogen metabolism takes place in the liver, where the product is primarily the safe 2-HE. But in 1996, researchers led by Johns Hopkins University environmental health scientist Tom Sutter discovered that an enzyme known as cytochrome P-4501B1 converts 17- CANCER New Role for Estrogen in Cancer — Service 279 (5357) β-estradiol to 4-HE. And last month, they reported in Carcinogenesis that this enzyme is more abundant in the breast than in tissues not prone to estrogen-linked cancers. Sutter adds that the enzyme also seems to be present in other cancer target tissues as well, including the uterus and ovaries. Indeed, human breast tumor tissue has an even higher level of 4-HE than normal tissue, perhaps because the enzyme is overactive there. But the data on whether the same is true for other human tumor tissues are still out.

    What actually damages DNA and leads to tumor development, says Cavalieri, may be the products of 4-HE. Numerous enzymes can change 4-HE into compounds called 3,4-semiquinones and 3,4-quinones. These compounds, Cavalieri and his colleagues report in the 30 September 1997 issue of the Proceedings of the National Academy of Sciences, bind to DNA, creating adducts, both in the test tube and in living animals, but then quickly fall off, taking with them two of DNA's bases, adenine and guanine. These gaps in the DNA, Cavalieri and his colleagues believe, have a strong potential to create gene mutations. “We believe it's [these] adducts that cause cancer,” says Cavalieri. Indeed, when he and his colleagues injected 4-HE quinones into newborn mice, the result was liver cancer.

    Two recent epidemiology studies seem to support the link between 4-HE and its byproducts and cancer. These studies focused on an enzyme known as catechol-O-methyltransferase, or COMT, that seeks out 2-HE and 4-HE and tags on a methyl group, making them more water soluble and therefore easier for the body to excrete. People who make less of the enzyme than others would in theory clear 4-HE and its toxic brethren more slowly. So the two studies—one led by Yager, the other by Thompson and her NCTR colleague Christine Ambrosone—looked to see if low levels of the enzyme would put women at greater risk of beast cancer.

    Both studies, each of which included between 230 and 300 women, did find that particular groups of women with low COMT levels had a higher incidence of breast cancer. The picture isn't altogether tidy, though. Yager's study, published in the 15 December 1997 issue of Cancer Research, found that only postmenopausal women with low COMT have a higher risk for breast cancer. In contrast, Thompson and Ambrosone's study, which is not yet published, saw a higher risk only in premenopausal women who smoke. Still, says Richard Weinshilboum, a pharmacogeneticist and metabolism expert at the Mayo Medical School in Rochester, Minnesota, the results “begin to make you think there is some fire under this smoke.”

    Take two

    Bradlow, however, sees 4-HE and its brethren as bit players. “In certain animal systems it is important,” says Bradlow. “But it's a very minor product in humans. We don't think it plays a major role” in cancer development. The problem, he says, is that in humans only about 5% of estrogens in the tissues end up as 4-HE. More than twice as much end up as 16-α, which he thinks is the most dangerous estrogen metabolite.

    Bradlow offers multiple lines of evidence to support this idea. He and his Strang colleague Nitan Telang found that cultured breast cells subjected to 16-α have an increased rate of DNA repair activity, an indication that the cells have a higher than normal level of mutation. In addition, in work that has not yet been published, Bradlow and his colleagues have seen that in tissue stains, 16-α invariably shows up in and around breast tumors.

    He also cites human epidemiological studies. One, which he and his colleagues are about to publish in the British Journal of Cancer, followed 85 women. It showed that those who had lower 2-HE to 16-α ratios in their urine were more likely to go on to develop breast cancer than were women in whom that ratio was weighted more strongly in favor of 2-HE.

    Even as the two camps continue to amass data to seal the case against their suspects, other possible culprits are continuing to emerge as well. Metabolites from both 4-HE and 2-HE can go on to create other reactive products, such as superoxide, hydroxyl radicals, and partially oxidized and reactive lipids, all of which themselves could be involved in damaging DNA and turning cells cancerous. Meanwhile, the same link between hormones and cancer is also being investigated as a cause of prostate cancer, although for now the data here remain more shallow. Of course, even though separate teams are currently backing their own favorite molecules, there's a good chance that everybody is in part correct, each feeling a different portion of the elephant. “It's not necessarily this or that, but this and that,” says Yager.

    The skeptics

    But many estrogen researchers aren't ready to back any metabolite hypothesis and remain troubled by what they see as inconsistencies. For one, estrogens and their metabolites don't register as highly mutagenic on standard tests such as the Ames test. “You test estrogen in a straightforward assay and it simply comes out negative,” says Gustafsson. Liehr counters that this is expected, because the Ames test and others like it are designed to measure potent rather than weak carcinogens. “If any endogenous compound was a strong carcinogen, we'd all be dead,” says Liehr.

    The critics also cite the fact that estrogen is present in the body in tiny quantities—too tiny, in fact, to produce enough byproducts to worry about, says Jonathan Li, a pharmacologist and cancer researcher at the University of Kansas Medical Center in Kansas City. Liehr agrees that the amounts of estrogen circulating in the blood are low, but points out that's not all there is to worry about. Estrogen is also synthesized directly in cells in target tissues by an enzyme called aromatase. It converts the androgen testosterone, which women make in small quantities, into estradiol.

    Indeed, Richard Stanten and his colleagues at the University of Virginia Health Sciences Center in Charlottesville have found that the amount of estrogen synthesized by aromatase in breast tissue far exceeds the amount of the hormone circulating in blood. “Since the tissue itself makes estrogen, there is enough present to make high levels of estrogen metabolism to make genotoxic activity plausible,” says Stanten.

    Plausible but not certain. Most estrogen metabolite researchers believe that certainty will come in time as the studies continue to roll in. Studies of mice that have been engineered to either over- or underexpress particular metabolite-controlling enzymes could be particularly enlightening, says Eppley Institute biochemist Eleanor Rogan. But even then, sorting out all the signals won't be easy because of the complexity of the system, she says.

    But the effort will be worthwhile, because if further evidence does nail down the idea that estrogen metabolites are mutagenic, it may be possible to intervene to reduce the risk of cancer, says Longfellow. If it turns out that women who over- or underexpress crucial enzymes have an increased cancer risk, for example, researchers could try to design drugs to bolster or block the levels of these compounds. “After all, these are things that can be modulated,” Longfellow says. But for now the primary challenge remains confirming the role of estrogen metabolites in the first place. “The evidence is building,” says Yager. “But the burden of proof still lies in developing more direct evidence.”

    • * “Estrogens as Endogenous Carcinogens in the Breast and Prostate,” Westfields International Conference Center, Chantilly, Virginia, 16–17 March.

    Additional Reading

    1. 738.
    2. 739.
    3. 740.
    4. 741.

    Did the First Complex Cell Eat Hydrogen?

    1. Gretchen Vogel

    Successful unions can start off in the strangest ways. Take eukaryotic cells, which compose all “higher” organisms and generally contain energy-producing organelles called mitochondria. Mitochondria were once free-living bacteria, and most researchers believe that early in evolution ancestral eukaryotic cells simply ate their future partners. But two researchers are now arguing for a less haphazard start to this ancient partnership. The first eukaryotes, they say, had an appetite for the waste products of the original mitochondria. The union of these organisms was simply a matter of survival.

    So happy together.

    Exchanges of molecules including hydrogen may have bound microbes together in the first complex cell. In a modern analog, bacteria snuggle close to hydrogen-producing organelles (dark structures, ~2 micrometers long) inside a protist (left).

    In last week's Nature, William Martin of Braunschweig Technical University in Germany and Miklós Müller of Rockefeller University in New York City draw on genetic data, biochemistry, and the lifestyles of some simple organisms today to argue that the first eukaryote evolved from a methanogen, a microbe that consumes hydrogen and carbon dioxide and produces methane. Its partner—the future mitochondrion—was a bacterium that made hydrogen and carbon dioxide as waste products.

    Scientists pondering how the first complex cell came together say the new idea could solve some nagging problems with the prevailing theory. “It's eminently sensible,” says evolutionary biologist Russell Doolittle of the University of California, San Diego. But he and others aren't ready to embrace the new scenario. “It's elegantly argued,” says Michael Gray of Dalhousie University in Halifax, Nova Scotia, but “there are an awful lot of things the hypothesis doesn't account for.”

    In the standard picture of eukaryote evolution, the mitochondrion was a lucky accident. First, the ancestral cell—probably an archaebacterium, recent genetic analyses suggest—acquired the ability to engulf and digest complex molecules. It began preying on its microbial companions. At some point, however, this predatory cell didn't fully digest its prey, and an even more successful cell resulted when an intended meal took up permanent residence and became the mitochondrion.

    For years, scientists had thought they had examples of the direct descendants of those primitive eukaryotes: certain protists that lack mitochondria. But recent analysis of the genes in those organisms suggests that they, too, once carried mitochondria but lost them later (Science, 12 September 1997, p. 1604). These findings hint that eukaryotes might somehow have acquired their mitochondria before they had evolved the ability to engulf and digest other cells.

    How it might have happened came to Martin one evening when he looked at a picture of a protist called Plagiopyla. These one-celled eukaryotes have hydrogen-producing organelles called hydrogenosomes, which are thought to be related to mitochondria. And in their cytoplasm, clustered among those organelles, live hydrogen-consuming methanogens.

    Looking at those hungry methanogens, Martin recalls, “the cell sort of evolved before my eyes.” He discussed the idea with Müller, and “all of a sudden everything fell into place,” Martin says. They concluded that what they saw inside the protist—the partnership of the organelles and the methanogens—mirrored the union that had led to the first eukaryotic cell.

    Müller and Martin think that the association between the ancestral methanogen and a hydrogen-producing bacterium started casually, in an oxygen-free, hydrogen-rich environment. The microbial pair later found itself far from that original environment, where the methanogen could not survive without its partner. Then, Martin and Müller suggest, a transfer of genes cemented the partnership, allowing the host to enclose its guest completely. The new genes enabled the methanogen to import small molecules, make sugars, and break them down into food for the enclosed cell. These genes probably came from the guest bacterium, which could also use oxygen to produce energy—as mitochondria do today.

    The hypothesis is “the most cogent explanation for why a eubacterium and an archaebacterial cell should get together in the first place,” says Gray. If it is right, current ideas about the relationship between eukaryotes and archaebacteria might shift. In the current picture, eukaryotes originated near the base of the tree. They branched off from the archaebacteria long before those organisms diverged into the main groups present today, such as the methanogens. Martin and Müller's hypothesis would shift the first eukaryotes well up the tree, tying them more closely to the archaebacteria.

    But Gray and others still have reservations about the scenario. “It's possible, but it's not as plausible as the standard idea” that the original host of mitochondria was a bacteria eater, says evolutionary biologist Tom Cavalier-Smith of the University of British Columbia in Vancouver. “It makes more sense if the host came from a bacterium that had experience digesting food and had transporter enzymes already,” so that it could import small molecules and feed its guest.

    Martin and Müller say that an analysis of the complete sequences of eukaryotic and archaebacterial genomes should show who is right. Their theory predicts that on the whole, the genes that eukaryotes derived from archaebacteria will look most like those of methanogens. It also suggests that direct descendants of the earliest eukaryotes may still lurk in dark, anaerobic environments. The best places to search for a living example of the ancestor of us all, Müller says, “are, of course, foul-smelling, muddy, or inside of a digestive tract.”


    Surveyor Shows the Flat Face of Mars

    1. Richard A. Kerr

    Every planet harbors a mystery that is key to understanding its fundamental nature. Earth's concerned its division into either low-lying ocean basin or high-standing continent. Once researchers realized that plate tectonics created dense ocean crust that sinks to form deep basins and light continental crust that floats high, the mystery was solved and the essential forces shaping Earth's surface were understood. Now, the first results from the altimeter aboard the Mars Global Surveyor (MGS) spacecraft are helping to solve an equally fundamental puzzle about that planet.

    Matching a martian profile.

    The flat martian lowlands resemble the topography of the South Atlantic ocean floor off Africa.


    On Mars, the mystery is a great crustal dichotomy: Much of the planet's northern hemisphere is a low-lying plain roughly centered on the north pole, while the rest of that hemisphere and all of the southern hemisphere are ancient highlands. Explanations have ranged from Earth-like plate tectonics to the cosmic catastrophe of a huge impact. The MGS altimeter results, reported on of this issue by a team led by geophysicist David Smith of NASA's Goddard Space Flight Center (GSFC) in Greenbelt, Maryland, hint that Earth-like tectonic forces and perhaps even an ancient ocean have shaped Mars's northern lowlands.

    The MGS results show that the northern lowlands are remarkably flat across thousands of kilometers and smooth on a scale of hundreds of meters. It's “the flattest surface in the solar system for which we have data,” says geophysicist Maria Zuber of the Massachusetts Institute of Technology, a teammate of Smith's. “The only thing that comes close is the heavily sedimented floors of Earth's oceans; it's actually flatter than that.” The similarity to Earth's oceans suggests that Mars's great basin formed the same way—by plate tectonics. But altimetry alone can't solve the riddle of martian topography or rule out other origins for its giant basin, she warns.

    The tantalizing new data come from the Mars Orbiter Laser Altimeter (MOLA) aboard the MGS. MOLA works much the way a ship's acoustic depth finder traces out the sea floor, but instead of using sound waves, it bounces an 8-nanosecond laser pulse of infrared light off the martian surface at 300-meter intervals. By measuring the light's round-trip time, MOLA gauges with 10-meter accuracy the height of the land, averaged over the width of the laser's 150-meter-wide beam. Changes in the shape of the pulse after reflection provide a measure of the smoothness of the surface the beam scanned.

    After 18 north-south tracks, “the remarkable thing is that the northern hemisphere is flat over thousands of kilometers,” says Zuber. On the scale of the 2000-kilometer-long tracks within the lowlands, the surface is level or slopes up toward the south at about 0.05°, she says. In most places between 50°N and the polar ice cap at 80°N, topography rises and falls by only 50 meters over hundreds of kilometers. This means, Zuber says, that the northern lowlands are flatter than the lava floods of the lunar maria, flatter than the vast volcanic plains of Venus, flatter than deserts on Earth. The smoothest part of the central Sahara, for example, is twice as rough as the martian lowlands. The most comparable topographic profiles Zuber could find are from terrestrial sea floors, for example the one running from the middle of the South Atlantic Ocean onto the edge of South America (see diagram for a similar oceanic profile). On the 100-meter scale, the smoothness of the martian lowlands is also comparable to that beneath terrestrial oceans. “You can see where this is going,” says Zuber.

    Indeed, this tempting match between Mars's lowlands and Earth's ocean basins fits a 1994 proposal by geophysicist Norman Sleep of Stanford University. He suggested that the lowlands are an “ocean” basin created by a martian version of plate tectonics that long ago ground to a halt. The lowlands would be underlain by dense crust produced by sea-floor spreading, and plate motions would have raised and roughened the boundary between the lowlands and highlands, another feature seen by MOLA.

    Others have proposed that whatever the basin's origins, there might once have been water filling it (Science, 12 February 1993, p. 910), a notion consistent with the extreme smoothness. Such smoothness is typically produced by some kind of sedimentation, such as the steady rain of tiny particles that smooth out the roughness of ocean crust.

    But other explanations of the basin's origins remain. For example, some researchers have suggested that one or several large asteroids or comets blasted Mars billions of years ago, leaving a thinned crust that sank as it cooled, an idea favored by MOLA team member Herbert Frey of the GSFC. No one knows just what kind of topography such a monumental impact would leave, so the altimeter data can't yet support or refute that idea. And Frey also points out that massive lava flows might account for the smoothness of the lowlands, especially with a patina of windblown sediments on top.

    Finding out what made northern Mars so flat will take more data. For now, says planetary geologist Michael Carr of the U.S. Geological Survey in Menlo Park, California, “we just don't know” what created the great martian dichotomy. But an Earth-like ocean basin is the hometown favorite.


    Ancient Island Tools Suggest Homo erectus Was a Seafarer

    1. Ann Gibbons

    In 1968, a Dutch missionary living on the Indonesian island of Flores found stone tools alongside the bones of an extinct type of elephant called a Stegodon, known to have lived at least 750,000 years ago. If the tools were as old as the Stegodon, this was a spectacular discovery, for Flores lies beyond a deep-water strait that separates most Asian and Australian faunas. The tools meant that the only human species then living in Southeast Asia, Homo erectus, must have been able to cross this biological barrier, called Wallace's line.

    But when the missionary, Theodor Verhoeven, reported his findings in the journal Anthropos, his claim was roundly dismissed. Although trained in classical archaeology, Verhoeven was an amateur, so researchers discounted his field work. And the accepted idea was that deep waters blocked human exploration until about 50,000 years ago. Although H. erectus was known from just 600 kilometers away on Java, most researchers were convinced that this early human lacked the social and linguistic skills needed to cross Wallace's line by piloting a raft over deep, fast-moving waters. Even after Dutch and Indonesian paleontologists backed Verhoeven's findings with new excavations and paleomagnetic dating in 1994, the claim was still considered dubious.


    Stone tools found between layers of volcanic rock on the island of Flores show humans were there about 800,000 years ago.


    In this week's issue of Nature, however, an international team presents new dates for stone tools from Flores, based on a different and more reliable technique called fission-track dating, that confirm H. erectus's presence there 800,000 years ago. The authors propose that the early humans who left behind these simple flakes and cobbles were “capable of repeated water crossings using watercraft” and may even have had language, needed to cooperate to build rafts. The “cognitive capabilities of H. erectus may be due for reappraisal,” says archaeologist Mike Morwood of the University of New England in Armidale, Australia, lead author of the paper.

    Most researchers accept the new dates for the artifacts, but they are sharply divided over what the findings reveal about the toolmaker. A few questions linger about whether the artifacts are really tools—and no H. erectus bones have been found on Flores to dispel these questions. Some researchers add that H. erectus might have accidentally drifted over to Flores on a raft or even walked on some previously unknown land bridge, says Colin Groves of Australian National University (ANU) in Canberra: “The Flores data do not seem convincing that H. erectus made boats.” Nonetheless, he agrees with others that the tools “are quite remarkable evidence of the distributional extent and environmental flexibility of our perhaps underestimated cousin, H. erectus.

    H. erectus in Asia has long been eclipsed by its relatives in Africa, where the species is thought to have arisen more than 1.8 million years ago. In the first known exodus of human beings from Africa, H. erectus then spread around the globe, settling in China and Java perhaps as early as 1.8 million years ago (Science, 25 February 1994, p. 1087). But although these early humans spread thousands of kilometers over land and across shallow straits, they seemed to have been incapable of deep-water crossings. In technical, social, and organizational skills—not to mention language—H. erectus was thought to lag far behind later humans.

    H. erectus's limitations seemed especially severe in Asia. Starting 1.5 million years ago, the Africans made better tools—two-sided stone hand axes—while the Asian members of the species either left almost no tools, as in Java, or only simple cobblestone choppers and flakes. “This group of Eastern hominids has always been regarded as impoverished in technological or cultural capabilities, as compared to their contemporaries in Africa,” says Philip Rightmire, a paleoanthropologist at the State University of New York, Binghamton.

    For the past 4 years, however, Dutch and Indonesian paleontologists have been coming up with support for Verhoeven's 1968 claim—and for a more flattering picture of Asian H. erectus. A Dutch and Indonesian group led by Paul Sondaar of the Natural History Museum in Rotterdam, the Netherlands, applied paleomagnetic dating, which is based on well-known reversals in Earth's magnetic field recorded in volcanic rock, to a rock layer just below 14 stone artifacts they had found in volcanic ash beds at a site called Mata Menge. The dates, about 750,000 years old, nicely matched Verhoeven's. But the results, published in 1994 and 1997 in French and Australian journals, were considered suspect. That was partly because of the lack of human bones and the uncertainties of this type of dating at the site, and also because the initial publication was in conference proceedings and was missed by many researchers, says Iain Davidson, an archaeologist at the University of New England in Armidale, Australia.

    Now new dating of ash layers from Mata Menge confirms these findings. The ash contains minerals such as zircon that are ideal for fission-track dating. Over time, atoms of uranium-238 in a zircon grain in volcanic rock undergo spontaneous fission, producing fragments that streak across the crystal lattice like a meteor in the sky and leave tracks about 10 micrometers long. By chemically etching the crystals, geochronologists can see and count the tracks; the more tracks they see, the more time has passed since the rock crystallized.

    Using this method on 50 individual grains of zircon from ash layers just above and below the tool-bearing sandstone layer, Paul O'Sullivan and Asaf Raza at La Trobe University in Bundoora, Australia, where the technique was pioneered, came up with ages of 800,000 to 880,000 years for almost all of the grains. Although it's easy to undercount the tracks, fission-track dating is considered reliable in the right hands, such as those of O'Sullivan's team. “The research group is tops, as good as they come,” says Andrew Carter, a geochronologist at the London Fission Track Research Group. “I can't find any faults with it at all. They've gone out of their way to undertake more grain analysis than is conventional.”

    But some researchers still wonder exactly what's being dated on Flores—human artifacts or just shattered rock. Smithsonian Institution paleoanthropologist Rick Potts notes that the ratio of 14 artifacts to 45 stone pieces recovered at the site in 1994 is only 31%, and he thinks at least 50% of stones at a site should be tools. “If this were a site in Africa, the fact that most of the rocks are not artifacts would make us doubt it as a lithics site,” he says.

    However, Morwood, an archaeologist invited to work with the Dutch and the Indonesians to check the authenticity of the tools and the stratigraphy at the site, insists that there is “absolutely no doubt about them being artifacts.” Other experts who have seen these artifacts agree that they're the real thing. “I think yes, having seen a few,” says Peter Bellwood, an archaeologist at ANU. Furthermore, some of the flakes are made of chert, a rock not found at the site, suggesting that the tools were made elsewhere.

    Other hints of H. erectus's presence on Flores come from the creatures that lived there, say Morwood and Sondaar. Sometime after 900,000 years ago, Flores's pygmy stegodons, giant tortoises, and giant Komodo dragons all suddenly went extinct. They were replaced by large stegodons, which apparently swam there in herds. Human hunters may have arrived and driven the pygmy stegodon and other animals to extinction, says Sondaar—making this the earliest extinction to be blamed on humans. All this has convinced those who have worked at Mata Menge that H. erectus was there—and that they arrived by raft or other watercraft. Even when the sea level was at its lowest, these humans would have had to cross 19 kilometers of water to get to Flores from the closest island of Sumbawa—after a 25-kilometer crossing over treacherous waters between Bali and Sumbawa. And an even longer crossing would be needed if they came from Sulawesi to the north, says Morwood. “You've got to be talking about watercraft,” he says. That has broad implications for H. erectus in Asia and beyond: “They were intelligent, thinking animals. Once you take into account the use of watercraft and their rapid radiation out of Africa, you have to rethink H. erectus. They must have had language for the collective effort needed to achieve this sea travel.” He speculates that the species reached the southern Indonesian island of Timor, where undated tools have also been found—and from there, perhaps even Australia.

    Few others are willing to go so far. “Australia would have been out of sight, whereas the island-hopping route to Flores was marked by huge volcanoes visible from afar,” says Bellwood. And Groves points out that the tectonics of these volcanic islands is so unstable that there may even have been a land bridge briefly connecting them. “Let's be cautious about what conclusions we draw about the navigational skills of H. erectus,” says Groves.

    Even if H. erectus did float to Flores, it could have been by accident, on a primitive raft, adds Davidson. Monkeys have been seen floating on makeshift rafts of mangrove tree limbs and vegetation in Indonesia, he says. Still, says Rightmire, this “does help to dispel the notion that H. erectus in general, and Eastern H. erectus in particular, were relatively slow to react to challenges posed by the environment,” because they not only navigated deep-water straits but adapted to life on an island, where the environment is thought to have been far different from the forest habitat of the mainland.

    The new findings also fit well with other work showing that Asian H. erectus has been underrated. Controversial new dates from sites in Java suggest that H. erectus persisted there from as early as 1.8 million years ago until as recently as 30,000 years ago, implying that they were able to adapt to varied terrain and climate. Other new studies suggest that H. erectus left behind sophisticated hand axes in southern China (see sidebar on p. 1636). For those who have worked on Flores and long believed in H. erectus's presence there, the new results are vindication. Says Sondaar: “I am happy that the finds of Verhoeven are finally recognized.”


    In China, a Handier Homo erectus

    1. Ann Gibbons

    More than 50 years ago, Harvard University anthropologist Hallam Movius divided the peoples of the Early Stone Age into two cultures: those who could make sophisticated, two-sided stone hand axes, known as Acheulean or mode 2 technology, and those who could not. This invisible technological barrier, which came to be known as Movius's line, separated handy H. erectus in Africa, the Middle East, and Europe from their less adept cousins in Asia, who left only mode 1 technology—simple stone flakes and cobbles used for chopping. Movius therefore wrote off the entire Asian continent as “a marginal region of cultural retardation.” Because the climate and terrain in the forests of Asia remained stable for the past 2 million years, Movius wrote, humans there didn't advance in culture but stayed backward for eons.

    Now, says Smithsonian Institution paleoanthropologist Rick Potts, “the Movius line is breached.” New international excavations in China reveal that at least a few early Asians were also making two-sided, or bifacial, stone tools as much as 730,000 years ago. Other researchers aren't convinced that these tools indicate mode 2 sophistication. But they agree that Stone Age Asians were probably engaged in more complex activities than Movius gave them credit for, abilities also suggested by signs that H. erectus somehow reached a far-flung island in Southeast Asia(see main text).

    Since Movius's time, bifacial tools have been found at almost a dozen sites in eastern Asia. But the dating of the sites is unreliable, and they appear to be less than 200,000 years old—long after the time of H. erectus and into the era of our own species, H. sapiens, says Indiana University, Bloomington, archaeologist Kathy Schick. And although researchers realized that early Asians might have used perishable tools made of wood or bamboo, the lack of advanced stone tools in Asia was still puzzling.

    Now firmer—and much earlier—dates may be emerging from the 800-square-kilometer Bose Basin in southern China. In the red dirt of this hilly, rural area along the Youjiang River, Huang Weiwen, an archaeologist at the Institute of Vertebrate Paleontology and Paleoanthropology in Beijing, found thousands of stone tools, including bifacial hand axes. Since 1994, Huang and Potts have systematically excavated and dated basin sites in the same stratigraphic layer, using fission-track and paleomagnetic dating, to show that bifacial hand axes from the center of the basin were made between 700,000 and 800,000 years ago.

    These hand axes don't look exactly like the classic Acheulean technology, but their shape suggests that they were manufactured systematically and, therefore, that the toolmakers shared technological traditions, says Potts. He argues that this elevates them above mode 1 technology and that it's time to redefine “mode 2” technologies in Asia: “For so long, we've equated mode 2 with the Acheulean.”

    But no one knows whether this group of axmakers was a band of failed immigrants from the west or whether the toolmaking tradition was born and refined in Asia. To answer that question, archaeologists need to discover and date such tools elsewhere in Asia. “You can't just rely on one basin to tell the whole story,” says Potts. “There's so much to do in China.”


    Polyhedra Can Bend But Not Breathe

    1. Dana Mackenzie
    1. Dana Mackenzie is a mathematics and science writer in Santa Cruz, California.

    Anyone who has made an origami crane knows the delight and wonderment of conjuring a moving creature from the static geometry of lines and triangles. Although the flexibility of the paper is what allows the crane's wings to flap, mathematicians showed 20 years ago that a geometric equivalent could be constructed: a closed, three-dimensional figure made of rigid triangles, which can be squeezed or stretched into a new shape without distorting the faces. The finding upset what had been an article of faith for geometers and engineers—that a structure whose surfaces are made of triangles must be rigid. But a new proof shows that flexible polyhedra still face constraints: They have to keep their volume constant as they move. As Robert Connelly of Cornell University puts it, “You cannot build a mathematical bellows.”

    The discovery of flexible polyhedra, with their infinitely changeable angles, had blown a hole in the long-standing belief that a given set of edge lengths can yield only a finite number of shapes. But the new result, published jointly in the German journal Beiträge zur Algebra und Geometrie by Connelly, Idzhad Sabitov of Moscow State University, and Connelly's student Anke Walz, implies that edge lengths do narrow down one of the most important aspects of shape—the volume—to a finite number of possibilities.

    In 1813, the French mathematician Augustin Louis Cauchy had proved that convex polyhedra—structures with flat faces, straight edges, and most important, no indentations—are always rigid. But that left open the question of whether polyhedra with indentations could flex. Around the turn of the century, a French engineer named Raoul Bricard found they could, if the faces were allowed to pass through each other. However, in the strictest sense Bricard's example, a flexible surface with eight faces, was not a polyhedron. For example, such a shape cannot be made into a physical object.

    However, Bricard turned out to be on the right track. In the 1970s, Connelly managed to build a true flexible surface by elaborately altering Bricard's example, eliminating certain faces and allowing certain edges to detour around others. Later, Klaus Steffen of the University of Düsseldorf discovered a flexible polyhedron with only nine vertices and 14 triangular faces, which is believed to be the simplest one possible (see illustration).

    As soon as these first models were built, mathematicians began playing around with them. “You could not help noticing,” says Herman Gluck of the University of Pennsylvania, “that … though they might compress some portion of the space within, there was always another portion that expanded.” Dennis Sullivan of the City University of New York blew smoke into a model and observed that none came out when the model was moved back and forth—suggesting again that it was not acting as a bellows.

    The key to proving what Connelly called the Bellows Conjecture was a vast generalization of a formula discovered by an ancient Greek mathematician, Heron of Alexandria. Heron's formula says that the area, x, of a triangle with side lengths (a, b, c) must solve the following polynomial: 16x2 + a4 + b4 + c4 −2a2b2 −2a2c2 −2b2c2 = 0. The volume of a tetrahedron has to satisfy a similar—but more complicated—polynomial. Connelly and, independently, Sabitov came up with the idea that the volume of any polyhedron might also solve some version of Heron's polynomial. If so, then the volume of a polyhedron with fixed side lengths could only change by jumping from one solution of the polynomial to another. But if the motion of the polyhedron is gradual, the volume cannot change suddenly. “It has no choice but to remain constant,” says Gluck.

    To prove that such polynomials exist for more complex polyhedra, Connelly and Sabitov found a way to divide these figures into tetrahedra, eliminate the edges of component tetrahedra that aren't actually edges of the final figure, and merge the known polynomials for the tetrahedra into a single polynomial for the entire shape. Even for a simple figure like an octahedron, the resulting polynomial involves 16th powers of the volume. Sabitov, in 1996, was the first to produce an algorithm that yields a polynomial for a general polyhedron, but last year's joint paper by Sabitov, Connelly, and Walz greatly simplifies the proof.

    The proof still leaves plenty of mysteries. For one thing, the Bellows Conjecture is surprisingly sensitive to the kind of space the figure inhabits. It does not hold in two dimensions: A flexible four-sided figure, for example, can change its area without changing the side lengths. Connelly and Walz believe they can prove the Bellows Conjecture in four dimensions, but in higher dimensions, Connelly admits, “we're stuck.”

    Sabitov remains optimistic that in three dimensions, at least, mathematicians will soon have a complete understanding of how edge lengths determine the shape of polyhedra—not just their volume but also whether they can flex, and by how much. In the future, he says, “there will be a chapter of geometry titled ‘The solution of polyhedra’ as we now have ‘The solution of triangles.’”

Log in to view full text

Via your Institution

Log in through your institution

Log in through your institution

Navigate This Article