The curse of Akkad.

The world's first empire was established forty-three hundred years ago, between the Tigris and Euphrates Rivers. The details of its founding, by Sargon of Akkad, have come down to us in a form somewhere between history and myth. Sargon--Sharru-kin, in the language of Akkadian--means "true king"; almost certainly, though, he was a usurper. As a baby, Sargon was said to have been discovered, Moses-like, floating in a basket. Later, he became cup- bearer to the ruler of Kish, one of ancient Babylonia's most powerful cities. Sargon dreamed that his master, Ur-Zababa, was about to be drowned by the goddess Inanna in a river of blood. Hearing about the dream, Ur-Zababa decided to have Sargon eliminated. How this plan failed is unknown; no text relating the end of the story has ever been found.

Until Sargon's reign, Babylonian cities like Kish, and also Ur and Uruk and Umma, functioned as independent city-states. Sometimes they formed brief alliances--cuneiform tablets attest to strategic marriages celebrated and diplomatic gifts exchanged--but mostly they seem to have been at war with one another. Sargon first subdued Babylonia's fractious cities, then went on to conquer, or at least sack, lands like Elam, in present-day Iran. He presided over his empire from the city of Akkad, the rains of which are believed to fie south of Baghdad. It was written that "daily five thousand four hundred men ate at iris presence," meaning, presumably, that he maintained a huge standing army. Eventually, Akkadian hegemony extended as far as the Khabur plains, in northeastern Syria, an area prized for its grain production. Sargon came to be known as "king of the world"; later, one of his descendants enlarged this title to "king of the four corners of the universe.”

Akkadian rule was highly centralized, and in this way anticipated the administrative logic of empires to come. The Akkadians levied taxes, then used the proceeds to support a vast network of local bureaucrats. They introduced standardized weights and measures---the gur equalled roughly three hundred litres--and imposed a uniform dating system, under which each year was assigned the name of a major event that had recently occurred: for instance, "the year that Sargon destroyed the city of Mari." Such was the level of systematization that even the shape and the layout of accounting tablets were imperially prescribed. Akkad's wealth was reflected in, among other things, its art work, the refinement and naturalism of which were unprecedented.

Sargon ruled, supposedly, for fifty-six years. He was succeeded by his two sons, who reigned for a total of twenty-four years, and then by a grandson, Naramsin, who declared himself a god. Naram-sin was, in turn, succeeded by his son. Then, suddenly, Akkad collapsed. During one three-year period, four men each, briefly, claimed the throne. "Who was king? Who was not king?" the register known as the Sumerian King List asks, in what may be the first recorded instance of political irony.

The lamentation "The Curse of Akkad" was written within a century of the empire's fall. It attributes Akkad's I demise to an outrage against the gods. I Angered by a pair of inauspicious oracles, Naram-sin plunders the temple of I Enlil, the god of wind and storms, who, in retaliation, decides to destroy both him and his people:

For the first time since cities were built and
  founded,
The great agricultural tracts produced no
  grain,
The inundated tracts produced no fish,
The irrigated orchards produced neither
  syrup nor wine,
The gathered clouds did not rain, the
  masgurum did not grow.
At that time, one shekel's worth of oil was
  only one-half quart,
One shekel's worth of grain was only one-
  half quart…
These sold at such prices in the markets of
  all the cities!
He who slept on the roof, died on the roof,
He who slept in the house, had no burial,
People were flailing at themselves from
  hunger.

For many years, the events described in "The Curse of Akkad" were thought, like the details of Sargon's birth, to be purely fictional.

In 1978, after scanning a set of maps at Yale's Sterling Memorial Library, a university archeologist named Harvey Weiss spotted a promising-looking mound at the confluence of two dry riverbeds in the Khabur plains, near the Iraqi border. He approached the Syrian government for permission to excavate the mound, and, somewhat to his surprise, it was almost immediately granted. Soon, he had uncovered a lost city, which in ancient rimes was known as Shekhna and today is called Tell Leilan.

Over the next ten years, Weiss, working with a team of students and local laborers, proceeded to uncover an acropolis, a crowded residential neighborhood reached by a paved road, and a large block of grain-storage rooms. He found that the residents of Tell Leilan had raised barley and several varieties of wheat, that they had used carts to transport their crops, and that in their writing they had imitated the style of their more sophisticated neighbors to the south. Like most cities in the region at the time, Tell Leilan had a rigidly organized, state-run economy: people received rations--so many lines of barley and so many of oil—based on how old they were and what kind of work they performed. From the time of the Akkadian empire, thousands of similar potsherds were discovered, indicating that residents had received their radons in mass-produced, one-line vessels. After examining these and other artifacts, Weiss constructed a time line of the city's history, from its origins as a small farming village (around 5000 B.C.), to its growth into an independent city of some thirty thousand people (2600 B.C.), and on to its reorganization under imperial rule (2300 B.C.).

Wherever Weiss and his team dug, they also encountered a layer of dirt that contained no signs of human habitation. This layer, which was more than three feet deep, corresponded to the years 2200 to 1900 B.C., and it indicated that, around the time of Akkad's fall, Tell Leilan had been completely abandoned. In 1991, Weiss sent soil samples from Tell Leilan to a lab for analysis. The results showed that, around the year 2200 B.C., even the city's earthworms had died out. Eventually, Weiss came to believe that the lifeless soil of Tell Leilan and the end of the Akkadian empire were products of the same phenomenon---a drought so prolonged and so severe that, in his words, it represented an example of "climate change.”

Weiss first published his theory, in the journal Science, in August, 1993. Since then, the list of cultures whose demise has been linked to climate change has continued to grow. They include the Classic Mayan civilization, which collapsed at the height of its development, around 800 A.D.; the Tiwanaku civilization, which thrived near Lake Titicaca, in the Andes, for more than a millennium, then disintegrated around 1100 A.D.; and the Old Kingdom of Egypt, which collapsed around the same time as the Akkadian empire. (In an account eerily reminiscent of "The Curse of Akkad," the Egyptian sage Ipuwer described the anguish of the period: "Lo, the desert claims the land. Towns are ravaged…Food is lacking…Ladies suffer like maidservants. Lo, those who were entombed are cast on high grounds.") In each of these cases, what began as a provocative hypothesis has, as new information has emerged, come to seem more and more compelling. For example, the notion that Mayan civilization had been undermined by climate change was first proposed in the late nineteen-eighties, at which point there was little climatological evidence to support it. Then, in the mid-nineteen-nineties, American scientists studying sediment cores from Lake Chichancanab, in north-central Yucatan, reported that precipitation patterns in the region had indeed shifted during the ninth and tenth centuries, and that this shift had led to periods of prolonged drought. More recently, a group of researchers examining ocean-sediment cores collected off the coast of Venezuela produced an even more detailed record of rainfall in the area. They found that the region experienced a series of severe, "multi-year drought events" beginning around 750 A.D. The collapse of the Classic Mayan civilization, which has been described as "a demographic disaster as profound as any other in human history,” is thought to have cost millions of lives.

The climate shifts that affected past cultures predate industrialization by hundreds---or, in the case of the Akkadians, thousands--of years. They reflect the climate system's innate variability and were caused by forces that, at this point, can only be guessed at. By contrast, the climate shifts predicted for the coming century are attributable to forces that are now well known. Exactly how big these shifts will be is a matter of both intense scientific interest and the greatest possible historical significance. In this context, the discovery that large and sophisticated cultures have already been undone by climate change presents what can only be called an uncomfortable precedent.

The Goddard Institute for Space Studies, or GISS, is situated just south of Columbia University's main campus, at the corner of Broadway and West 112th Street. The institute is not well marked, but most New Yorkers would probably recognize the building: its ground floor is home to Tom's Restaurant, the coffee shop made famous by "Seinfeld.”

GISS, an outpost of NASA, started out, forty-four years ago, as a planetary-research center; today, its major function is making forecasts about climate change. GISS employs about a hundred and fifty people, many of whom spend their days working on calculations that may—or may not--end up being incorporated in the institute's climate model. Some work on algorithms that describe the behavior of the atmosphere, some on the behavior of the oceans, some on vegetation, some on clouds, and some on making sure that all these algorithms, when they are combined, produce results that seem consistent with the real world. (Once, when some refinements were made to the model, rain nearly stopped falling over the rain forest.) The latest version of the GISS model, called ModelE, consists of a hundred and twenty-five thousand lines of computer code.

GISS's director, James Hansen, occupies a spacious, almost comically cluttered office on the institute's seventh floor. (I must have expressed some uneasiness the first time I visited him, because the following day I received an e-mail assuring me that the office was "a lot better organized than it used to be.") Hansen, who is sixty-three, is a spare man with a lean face and a fringe of brown hair. Although he has probably done as much to publicize the dangers of global warming as any other scientist, in person he is reticent almost to the point of shyness. When I asked him how he had come to play such a prominent role, he just shrugged. "Circumstances," he said.

Hansen first became interested in climate change in the mid-nineteen- seventies. Under the direction of James Van Allen (for whom the Van Allen radiation belts are named), he had written his doctoral dissertation on the climate of Venus. In it, he had proposed that the planet, which has an average surface temperature of eight hundred and sixty-seven degrees Fahrenheit, was kept warm by a smoggy haze; soon afterward, a space probe showed that Venus was actually insulated by an atmosphere that consists of ninety-six per cent carbon dioxide. When solid data began to show what was happening to greenhouse-gas levels on earth, Hansen became, in his words, "captivated." He decided that a planet whose atmosphere could change in the course of a human lifetime was more interesting than one that was going to continue, for all intents and purposes, to broil away forever. A group of scientists at NASA had put together a computer program to try to improve weather forecasting using satellite data. Hansen and a team of half a dozen other researchers set out to modify it, in order to make longer-range forecasts about what would happen to global temperatures as greenhouse gases continued to accumulate. The project, which resulted in the first version of the GISS climate model, took nearly seven years to complete.

At that time, there was little empirical evidence to support the notion that the earth was warming. Instrumental temperature records go back, in a consistent fashion, only to the mid-nineteenth century. They show that average global temperatures rose through the first half of the twentieth century, then dipped in the nineteen-fifties and sixties. Nevertheless, by the early nineteen-eighties Hansen had gained enough confidence in his model to begin to make a series of increasingly audacious predictions. In 1981, he forecast that "carbon dioxide warming should emerge from the noise of natural climate variability" around the year 2000. During the exceptionally hot summer of 1988, he appeared before a Senate subcommittee and announced that he was "ninety-nine per cent" sure that "global warming is affecting our planet now.” And in the summer of 1990 he offered to bet a roomful of fellow-scientists a hundred dollars that either that year or one of the following two years would be the warmest on record. To qualify, the year would have to set a record not only for land temperatures but also for sea-surface temperatures and for temperatures in the lower atmosphere. Hansen won the bet in six months.

Like all climate models, GISS's divides the world into a series of boxes. Thirty-three hundred and twelve boxes cover the earth's surface, and this pattern is repeated twenty times moving up through the atmosphere, so that the whole arrangement might be thought of as a set of enormous checkerboards stacked on top of one another. Each box represents an area of four degrees latitude by five degrees longitude. (The height of the box varies depending on altitude.) In the real world, of course, such a large area would have an incalculable number of features; in the world of the model, features such as lakes and forests and, indeed, whole mountain ranges are reduced to a limited set of properties, which are then expressed as numerical approximations. Time in this grid world moves ahead for the most part in discrete, half-hour intervals, meaning that a new set of calculations is performed for each box for every thirty minutes that is supposed to have elapsed in actuality. Depending on what part of the globe a box represents, these calculations may involve dozens of different algorithms, so that a model run that is supposed to simulate climate conditions over the next hundred years involves more than a quadrillion separate operations. A single run of the G1SS model, done on a supercomputer, usually takes about a month.

Very broadly speaking, there are two types of equations that go into a climate model. The first group expresses fundamental physical principles, like the conservation of energy and the law of gravity. The second group describes—the term of art is "parameterize"—patterns and interactions that have been observed in nature but may be only partly understood, or processes that occur on a small scale, and have to be averaged out over huge spaces. Here, for example, is a tiny piece of ModelE, written in the computer language FORTRAN, which deals with the formation of clouds:

C**** COMPUTE THE AUTOCONVERSION
RATE OF CLOUD WATER TO PRECIPITATION
   RHO= 1 .E5 * PL(L)/(RGAS *TL(L))
TEM--RHO*WMX(L)/(WCONST*FCLD+
  1.E-20)
IF(LHX.EQ.LHS) TEM=RHO*WMX(L)/
  (WMUI*FCLD+1.E-20)
TEM =TEM*TEM
IF(TEM.GT.10.) TEM=10.
CM1=CM0
IF(BANDF) CM1=CM0 *CBF
IF(LHX.EQ.LHS) CM1=CM0
CM=CM1 *(1.-1./EXP(TEM*TEM))+1.
  *100.*(PREBAR(L+1)+
*PRECNVL(L+ 1)* BYDTSTC)
IF(CM.GT.BYDTSrC) CM=BYDTSrC
PREP(L)=WMX(L) *CM
END IF
C* * * * FORM CLOUDS ONLY IF RH GT RH00
219 IF(RH1(L).LT.RH00(L)) GO TO 220.

All climate models treat the laws of physics in the same way, but, since they parameterize phenomena like cloud formation differently, they come up with different results. (At this point, there are some fifteen major climate models in operation around the globe.) Also, because the real-world forces influencing the climate are so numerous, different models tend, like medical students, to specialize in different processes. GISS's model, for example, specializes in the behavior of the atmosphere, other models in the behavior of the oceans, and still others in the behavior of land surfaces and ice sheets.

Last fall, I attended a meeting at GISS which brought together members of the institute's modelling team. When I arrived; about twenty men and five women were sitting in battered chairs in a conference room across from Hansen's office. At that particular moment, the institute was performing a series of runs for the U.N. Intergovernmental Panel on Climate Change. The runs were overdue, and apparently the I.P.C.C. was getting impatient. Hansen flashed a series of charts on a screen on the wall summarizing some of the results obtained so far.

The obvious difficulty in verifying any particular climate model or climate-model run is the prospective nature of the results. For this reason, models are often run into the past, to see how well they reproduce trends that have already been observed. Hansen told the group that he was pleased with how ModelE had reproduced the aftermath of the eruption of Mt. Pinatubo, in the Philippines, which took place in June of 1991. Volcanic eruptions release huge quantifies of sulfur dioxide--Pinatubo produced some twenty million tons of the gas---which, once in the stratosphere, condenses into tiny sulfate droplets. These droplets, or aerosols, tend to cool the earth by reflecting sunlight back into space. (Man-made aerosols, produced by burning coal, oil, and biomass, also reflect sunlight and are a countervailing force to greenhouse warming, albeit one with serious health consequences of its own.) This cooling effect lasts as long as the aerosols remain suspended in the atmosphere. In 1992, global temperatures, which had been rising sharply, fell by half of a degree. Then they began to climb again. ModelE had succeeded in simulating this effect to within nine-hundredths of a degree. "That's a pretty nice test," Hansen observed laconically.

One day, when I was talking to Hansen in his office, he pulled a pair of photographs out of his briefcase. The first showed a chubby-faced five-year-old girl holding some miniature Christmas-tree lights in front of an even chubbier—faced five-month-old baby. The girl, Hansen told me, was his granddaughter Sophie and the boy was his new grandson, Connor. The caption on the first picture read, "Sophie explains green-house warming." The caption on the second photograph, which showed the baby smiling gleefully, read, "Connor gets it.”

When modellers talk about what drives the climate, they focus on what they call "forcings." A forcing is any ongoing process or discrete event that alters the energy of the system. Examples of natural forcings include, in addition to volcanic eruptions, periodic shifts in the earth's orbit and changes in the sun's output, like those linked to sun-spots. Many climate shifts of the past have no known forcing associated with them; for instance, no one is certain what brought about the so-called Little Ice Age, which began in Europe some five hundred years ago. A very large forcing, meanwhile, should produce a commensurately large--and obvious--effect. One GISS scientist put it to me this way: "If the sun went supernova, there's no question that we could model what would happen.”

Adding carbon dioxide, or any other greenhouse gas, to the atmosphere by, say, burning fossil fuels or levelling forests is, in the language of climate science, an anthropogenic forcing. Since pre-industrial times, the concentration of CO2 in the earth's atmosphere has risen by roughly a third, from 280 parts per million to 378 p.p.m. During the same period, concentrations of methane, an even more powerful (but more short—lived) greenhouse gas, have more than doubled, from .78 p.p.m, to 1.76 p.p.m. Scientists measure forcings in terms of watts per square metre, or w/m2, by which the), mean that a certain number of watts of energy have been added (or, in the case of a negative forcing, subtracted) for every single square metre of the earth's surface. The size of the green-house forcing is estimated, at this point, to be 2.5 w/m2. A miniature Christmas light gives off about four tenths of a watt of energy, mostly in the form of heat, so that, in effect (as Sophie supposedly explained to Connor), we have covered the earth with tiny bulbs, six for every square metre. These bulbs are burning twenty-four hours a day, seven days a week, year in and year out.

If greenhouse gases were held constant at today's levels, it is estimated that it would take several decades for the full impact of the forcing that is already in place to be felt. This is because raising the earth's temperature involves not only warming the air and the surface of the land but also melting sea ice, liquefying glaciers, mad, most significant, heating the oceans--all processes that require tremendous amounts of energy (Imagine trying to thaw a gallon of ice cream or warm a pot of water using an Easy-Bake oven.) It could be argued that the delay that is built into the system is socially useful, because it enables us--with the help of climate models--to prepare for what lies ahead, or that it is socially disastrous, because it allows us to keep adding CO2 to the atmosphere while fobbing the impacts off on our children and grandchildren. Either way, if current trends continue, which is to say, if steps are not taken to reduce emissions, carbon dioxide levels will probably reach 500 parts per million-nearly double pre-industrial levels--sometime around the middle of the century. By that point, of course, the forcing associated with greenhouse gases will also have increased, to four watts per square metre and possibly more. For comparison's sake, it is worth keeping in mind that the total forcing that ended the last ice agenda forcing that was eventually sufficient to melt mile-thick ice sheets and raise global sea levels by four hundred feet--is estimated to have been just six and a half watts per square metre.

There are two ways to operate a climate model. In the first, which is known as a transient run, greenhouse gases are slowly added to the simulated atmosphere--just as they would be to the real atmosphere--and the model forecasts what the effect of these additions will be at any given moment. In the second, greenhouse gases are added to the atmosphere all at once, and the model is run at these new levels until the climate has fully adjusted to the forcing by reaching a new equilibrium. Not surprisingly, this is known as an equilibrium run. For doubled CO2, equilibrium runs of the GISS model predict that average global temperatures will rise by 4.9 degrees Fahrenheit. Only about a third of this increase is directly attributable to more greenhouse gases; the rest is a result of indirect effects, the most important among them being the so-called "water-vapor feedback." (Since warmer air holds more moisture, higher temperatures are expected to produce an atmosphere containing more water vapor, which is itself a greenhouse gas.) GISS's forecast is on the low end of the most recent projections; the Hadley Centre model, which is run by the British Met Office, predicts that for doubled CO2 the eventual temperature rise will be 6.3 degrees Fahrenheit, while Japan's National Institute for Environmental Studies predicts 7.7 degrees.

In the context of ordinary life, a warming of 4.9, or even of 7.7, degrees may not seem like much to worry about; in the course of a normal summer's day, after all, air temperatures routinely rise by twenty degrees or more. Average global temperatures, however, have practically nothing to do with ordinary life. In the middle of the last glaciation, Manhattan, Boston, and Chicago were deep under ice, and sea levels were so low that Siberia and Alaska were connected by a land bridge nearly a thousand miles wide. At that point, average global temperatures were roughly ten degrees colder than they are today. Conversely, since our species evolved, average temperatures have never been much more than two or three degrees higher than they are right now.

This last point is one that climatologists find particularly significant. By studying Antarctic ice cores, researchers have been able to piece together a record both of the earth's temperature and of the composition of its atmosphere going back four full glacial cycles. (Temperature data can be extracted from the isotopic composition of the ice, and the makeup of the atmosphere can be reconstructed by analyzing tiny bubbles of trapped air.) What this record shows is that the planet is now nearly as warm as it has been at any point in the last four hundred and twenty thousand years. A possible consequence of even a four-or five-degree temperature rise--on the low end of projections for doubled CO2--is that the world will enter a completely new climate regime, one with which modern humans have no prior experience. Meanwhile, at 378 p.p.m., CO2 levels are significantly higher today than they have been at any other point in the Antarctic record. It is believed that the last time carbon-dioxide levels were in this range was three and a half million years ago, during what is known as the mid-Pliocene warm period, and they likely have not been much above it for tens of millions of years. A scientist with the National Oceanic and Atmospheric Ad- ministration (NOAA) put it to meanly half-jokingly--this way: "It's true that we've had higher CO2 levels before. But, then, of course, we also had dinosaurs.”

David Rind is a climate scientist who has worked at GISS since 1978. Rind acts as a trouble-shooter for the institute's model, scanning reams of numbers known as diagnostics, trying to catch problems, and he also works with GISS'S Climate Impacts Group. (His office, like Hansen's, is filled with dusty piles of computer printouts.) Although higher temperatures are the most obvious and predictable result of increased CO2, other, second-order consequences--rising sea levels, changes in vegetation, loss of snow cover--are likely to be just as significant. Rind's particular interest is how CO2 levels will affect water supplies, because, as he put it to me, "you can't have a plastic version of water.”

One afternoon, when I was talking to Rind in his office, he mentioned a visit that President Bush's science adviser, John Marburger, had paid to GISS a few years earlier. "He said, 'We're really interested in adaptation to climate change,' " Rind recalled. "Well, what does 'adaptation' mean?" He rummaged through one of his many file cabinets and finally pulled out a paper that he had published in the Journal of Geophysical Research entitled "Potential Evapo—transpiration and the Likelihood of Future Drought." In much the same way that wind velocity is measured using the Beaufort scale, water availability is measured using what's known as the Palmer Drought Severity Index. Different climate models offer very different predictions about future water availability; in the paper, Rind applied the criteria used in the Palmer index to GISS's model and also to a model operated by NOAA's Geophysical Fluid Dynamics Laboratory. He found that as carbon-dioxide levels rose the world began to experience more and more serious water shortages, starting near the equator and then spreading toward the poles. When he applied the index to the GISS model for doubled CO2, it showed most of the continental United States to be suffering under severe drought conditions. When he applied the index to the G.F.D.L. model, the results were even more dire. Rind created two maps to illustrate these findings. Yellow represented a forty—to—sixty-per-cent chance of summertime drought, ochre a sixty-to-eighty-per-cent chance, and brown an eighty-to-a-hundred-per- cent chance. In the first map, showing the GISS results, the Northeast was yellow, the Midwest was ochre, and the Rocky Mountain states and California were brown. In the second, showing the G.F.D.L. results, brown covered practically the entire country.

"I gave a talk based on these drought indices out in California to water-resource managers," Rind told me. "And they said, 'Well, if that happens, forget it.' There's just no way they could deal with that.”

He went on, "Obviously, if you get drought indices like these, there's no adaptation that's possible. But let's say it's not that severe. What adaptation are we talking about? Adaptation in 2020? Adaptation in 2040? Adaptation in 2060? Because the way the models project this, as global warming gets going, once you've adapted to one decade you're going to have to change everything the next decade.

"We may say that we're more technologically able than earlier societies. But one thing about climate change is it's potentially geopolitically destabilizing. And we're not only more technologically able; we're more technologically able destructively as well. I think it's impossible to predict what will happen. I guess--though I won't be around to see it--I wouldn't he shocked to find out that by 2100 most things were destroyed." He paused. "That's sort of an extreme view.”

On the other side of the Hudson River and slightly to the north of GISS, the Lamont-Doherty Earth Observatory occupies what was once a week-end estate in the town of Palisades, New York. The observatory is an outpost of Columbia University, and it houses, among its collections of natural artifacts, the world's largest assembly of ocean-sediment cores--more than thirteen thousand in all The cores are kept in steel compartments that look like drawers from a firing cabinet, only longer and much skinnier. Some of the cores are chalky, some are clayey, and some are made up almost entirely of gravel. All can be coaxed to yield up--in one way or an-other--information about past climates.

Peter deMenocal is a paleoclimatologist who has worked at Lamont-Doherty for fifteen years. He is an expert on ocean cores, and also on the climate of the Pliocene, which lasted from roughly five million to two million years ago. Around two and a half million years ago, the earth, which had been warm and relatively ice-free, started to cool down until it entered an era--the Pleistocene of recurring glaciations. DeMenocal has argued that this transition was a key event in human evolution: right around the time that it occurred, at least two types of hominids--one of which would eventually give rise to us--branched off from a single ancestral line. Until quite recently, paleoclimatologists like deMenocal rarely bothered with anything much closer to the present day; the current interglacial--the Holocene--which began some ten thousand years ago, was believed to be, climatically speaking, too stable to warrant much study. In the mid-nineties, though, de-Menocal, motivated by a growing concern over global warming--and a concomitant shift in government research funds--decided to look in detail at some Holocene cores. What he learned, as he put it to me when I visited him at Lamont-Doherty last fall, was "less boring than we had thought.”

One way to extract climate data from ocean sediments is to examine the remains of what lived or, perhaps more pertinently, what died and was buried there. The oceans are rich with microscopic creatures known as foraminifera. There are about thirty planktonic species in all, and each thrives at a different temperature, so that by counting a species' prevalence in a given sample it is possible to estimate the ocean temperatures at the time the sediment was formed. When de-Menocal used this technique to analyze cores that had been collected off the coast of Mauritania, he found that they contained evidence of recurring cool periods; every fifteen hundred years or so, water temperatures dropped for a few centuries before climbing back up again. (The most recent cool period corresponds to the Little Ice Age, which ended about a century and a half ago.) Also, perhaps even more significant, the cores showed profound changes in precipitation. Until about six thousand years ago, northern Africa was relatively wet--dotted with small lakes. Then it became dry, as it is today. DeMenocal traced the shift to periodic variations in the earth's orbit, which, in a generic sense, are the same forces that trigger ice ages. But orbital changes occur gradually, over thousands of years, and northern Africa appears to have switched from wet to dry all of a sudden. Although no one knows exactly how this happened, it seems, like so many climate events, to have been a function of feedbacks---the less rain the continent got, the less vegetation there was to retain water, and so on until, finally, the system just flipped. The process provides yet more evidence of how a very small forcing sustained over time can produce dramatic results.

"We were kind of surprised by what we found," deMenocal told me about his work on the supposedly stable Holocene. "Actually, more than surprised. It was one of these things where, you know, in life you take certain things for granted, like your neighbor's not going to be an axe murderer. And then you discover your neighbor is an axe murderer.”

Not long after deMenocal began to think about the Holocene, a brief mention of his work on the climate of Africa appeared in a book produced by National Geographic. On the facing page, there was a piece on Harvey Weiss and his work at Tell Leilan. DeMenocal vividly remembers his reaction. "I thought, Holy cow, that's just amazing!" he told me. "It was one of these cases where I lost sleep that night, I just thought it was such a cool idea.”

DeMenocal also recalls his subsequent dismay when he went to learn more. "It struck me that they were calling on this climate-change argument, and I wondered how come I didn't know about it," he said. He looked at the Science paper in which Weiss had originally laid out his theory. "First of all, I scanned the list of authors and there was no paleoclimatologist on there," deMenocal said. "So then I started reading through the paper and there basically was no paleoclimatology in it." (The main piece of evidence Weiss adduced for a drought was that Tell Leilan had filled with dust.) The more deMenocal thought about it, the more unconvincing he found the data, on the one hand, and the more compelling he found the underlying idea, on the other. "I just couldn't leave it alone," he told me. In the summer of 1995, he went with Weiss to Syria to visit Tell Lellan. Subsequently, he decided to do his own study to prove--or disprove--Weiss's theory.

Instead of looking in, or even near, the ruined city, deMenocal focussed on the Gulf of Oman, nearly a thousand miles downwind. Dust from the Mesopotamian floodplains, just north of Tell Leilan, contains heavy concentrations of the mineral dolomite, and since arid soil produces more wind-borne dust, deMenocal figured that if there had been a drought of any magnitude it would show up in gulf sediments. "In a wet period, you'd be getting none or very, very, low amounts of dolomite, and during a dry period you'd be getting a lot," he explained. He and a graduate student named Heidi Cullen developed a highly sensitive test to detect dolomite, and then Cullen assayed, centimetre by centimetre, a sediment core that had been extracted near where the Gulf of Oman meets the Arabian Sea.

"She started going up through the core," DeMenocal told me. "It was like nothing, nothing, nothing, nothing, nothing. Then one day, I think it was a Friday afternoon, she goes, 'Oh, my God.' It was really classic." DeMenocal had thought that the dolomite level, if it were elevated at all, would be modestly higher; instead, it went up by tour hundred per cent. Still, he wasn't satisfied. He decided to have the core re-analyzed using a different marker: the ratio of strontium 86 and strontium 87 isotopes. The same spike showed up. When deMenocal had the core carbon-dated, it turned out that the spike lined up exactly with me period of Tell Leilan's abandonment.

Tell Leilan was never an easy place to live. Much like, say, western Kansas today, the Khabur plains received enough annual rainfall--about seventeen inches--to support cereal crops, but not enough to grow much else. "Year-to-year variations were a real threat, and so they obviously needed to have grain storage and to have ways to buffer themselves," deMenocal observed. "One generation would tell the next, 'Look, there are these things that happen that you've got to be prepared for.' And they were good at that. They could manage that. They were there for hundreds of years.”

He went on, "The thing they couldn't prepare for was the same thing that we won't prepare for, because in their case they didn't know about it and because in our case the political system can't listen to it. And that is that the climate system has much greater firings in store for us than we think.”

Shortly before Christmas, Harvey Weiss gave a lunchtime lecture at Yale's Institute for Biospheric Studies. The title was "What Happened in the Holocene," which, as Weiss explained, was an allusion to a famous archeology text by V. Gordon Childe, entitled "What Happened in History." The talk brought together archeological and paleoclimatic records from the Near East over the last ten thousand years.

Weiss, who is sixty years old, has thinning gray hair, wire-rimmed glasses, and an excitable manner. He had prepared for the audience--mostly Yale professors and graduate students--a handout with a time line of Mesopotamian history. Key cultural events appeared in black ink, key climatological ones in red. The two alternated in a rhythmic cycle of disaster and innovation. Around 6200 B.C., a severe global cold snap--red ink--produced aridity in the Near East. (The cause of the cold snap is believed to have been a catastrophic flood that emptied an enormous glacial lake---called Lake Agassiz--into the North Atlantic.) Right around the same time--black ink--farming villages in northern Mesopotamia were abandoned, while in central and southern Mesopotamia the art of irrigation was invented. Three thousand years later, there was another cold snap, after which settlements in northern Mesopotamia once again were deserted. The most recent red event, in 2200 B.C., was followed by the dissolution of the Old Kingdom in Egypt, the abandonment of villages in ancient Palestine, and the fall of Akkad. Toward the end of his talk, Weiss, using a PowerPoint program, displayed some photographs from the excavation at Tell Leilan. One showed the wall of a buildings-probably in- tended for administrative offices-that had been under construction when the rain stopped. The wall was made from blocks of basalt topped by rows of mud bricks. The bricks gave out abruptly; as if construction had ceased from one day to the next.

The monochromatic sort of history that most of us grew up with did not allow for events like the drought that destroyed Tell Leilan. Civilizations fell, we were taught, because of wars or barbarian invasions or political unrest. (Another famous text by Childe bears the exemplary tide "Man Makes Himself.") Adding red to the time line points up the deep contingency of the whole enterprise. Civilization goes back, at the most, ten thousand years, even though, evolutionarily speaking, modern man has been around for at least ten times that long. The climate of the Holocene was not boring, but at least it was dull enough to allow people to sit still. It is only after the immense climatic shifts of the glacial epoch had run their course that writing and agriculture finally emerged.

Nowhere else does the archeological record go back so far or in such detail as in the Near East. But similar red-and- black chronologies can now be drawn up for many other parts of the world: the Indus Valley, where, some four thousand years ago, the Harappan civilization suffered a decline after a change in monsoon patterns; the Andes, where, fourteen hundred years ago, the Moche abandoned their dries in a period of diminished rainfall; and even the United States, where the arrival of the English colonists on Roanoke Island, in 1587, coincided with a severe regional drought. (By the time English ships returned to resupply the colonists, three years later, no one was left.) At the height of the Mayan civilization, population density was five hundred per square mile, higher than it is in most parts of the U. S. today. Two hundred years later, much of the territory occupied by the Mayans had been completely depopulated. You can argue that man through culture creates stability, or you can argue, just as plausibly, that stability is for culture an essential precondition.

After the lecture, I walked with Weiss back to his office, which is near the center of the Yale campus, in the Hall of Graduate Studies. This past year, Weiss decided to suspend excavation at Tell Leilan. The site lies only fifty miles from the Iraqi border, and, owing to the uncertainties of the war, it seemed like the wrong sort of place to bring graduate students. When I visited, Weiss had just returned from a trip to Damascus, where he had gone to pay the guards who watch over the site when he isn't there. While he was away from his office, its contents bad been piled up in a corner by repairmen who had come to fix some pipes. Weiss considered the piles disconsolately; then unlocked a door at the back of the room.

The door led to a second room, much larger than the first. It was set up like a library, except that instead of books the shelves were stacked with hundreds of cardboard boxes. Each box contained fragments of broken pottery from Tell Leilan. Some were painted, others were incised with intricate designs, and still others were barely distinguishable from pebbles. Every fragment had been inscribed with a number, indicating its provenance.

I asked what he thought life in Tell Leilan had been like. Weiss told me that that was a "corny question," so I asked him about the city's abandonment. "Nothing allows you to go beyond the third or fourth year of a drought, and by the fifth or sixth year you're probably gone," he observed. "You've given up hope for the rain, which is exactly what they wrote in 'The Curse of Akkad.'" I asked to see something that might have been used in Tell Leilan's last days. Swearing softly; Weiss searched through the rows until he finally found one particular box. It held several potsherds that appeared to have come from identical bowls. They were made from a greenish-colored clay, had been thrown on a wheel, and had no decoration. Intact, the bowls had held about a litre, and Weiss explained that they had been used to mete out rations----probably wheat or barley--to the workers of Tell Leilan. He passed me one of the fragments. I held it in my hand for a moment and tried to imagine the last Akkadian who had touched it. Then I passed it back (This is the second part of a three-part article.)

PHOTO (COLOR): When the Akkadian empire collapsed, around 2200 B. C., even the earthworms died. Photograph by Joseph Coscia, Jr.

~~~~~~~~

By Elizabeth Kolbert