Common section



Questions about Angkor ■ Angkor’s environment ■ Angkor’s rise ■ The great city ■ Magnificent engineering ■ Angkor’s decline ■

In 2008 I finally fulfilled a long-held dream of visiting the ruins of the medieval Southeast Asian city of Angkor in modern Cambodia. While I already knew that Angkor was big, no advance reading could have prepared me for experiencing at first hand the reality of its enormous scale. At its peak around a thousand years ago, it was the world’s most extensive city, among the most populous ones, and the capital of the largest and most powerful empire in Southeast Asia (the Khmer Empire). Its temples, such as Angkor Wat, include the largest religious monuments of the pre-modern world. Along with China’s Great Wall, Angkor’s large size makes it one of the ancient constructions most readily visible from space.

Yet, by the 19th century there remained only about eight small villages dispersed over the central area formerly covered by this vast city. Cambodia today has become Southeast Asia’s poorest country. I know of no modern nation so nostalgically identified with the vanished glory of its archaeological past as is Cambodia, whose national flag displays an image of Angkor. The devolution of a sprawling metropolis into a largely empty landscape with scattered villages surely deserves to be termed a collapse. How did an environment initially supporting poor farmers spawn such a huge city and empire, and then fade again?

When I published my book Collapse in 2005, I devoted only four sentences to Angkor because too little information was available then to tell a coherent story of the city’s collapse. Now, thanks to a flood of recent information from aerial radar surveys, ground surveys, excavations, and tree-ring measurements, we understand better what happened, even though many questions remain unanswered. It turns out that Angkor was not a unique phenomenon, as it at first seems. It was “just” the largest example of a type of city that no longer exists today but that formerly occurred more widely in seasonally wet tropical environments. That now unfamiliar type consisted of a low-density city, much more spread out than even my notoriously diffuse home city of Los Angeles: a city with farmland and farmhouses in close proximity to palaces and temples, with people living at lower population densities than in our familiar modern cities that are all city and no farmland, but living at higher densities than in densely populated purely rural modern landscapes. Other examples of such low-density cities besides Angkor used to exist in Sri Lanka, Java, Thailand, Vietnam, and Myanmar. They existed especially in the Maya homeland of Mexico and Honduras, which supported Tikal, Copán, and the other great Maya cities that are the most familiar examples of this now vanished urban model. But even Tikal, the largest well-surveyed Maya city, was only about one-fifth of Angkor in extent. All of these low-density cities collapsed before they could be visited and described by Europeans. Could understanding of Angkor’s decline also illuminate the Classic Lowland Maya collapses that I described in Chapter 5?

In the back of my mind as I visited Angkor was another question about its broader relevance. Sadly, Cambodia today is famous not only for its glorious ancient past, but also for its horrible recent past. From 1975 to 1979, under the paranoid dictatorship of Pol Pot and his Khmer Rouge, Cambodia was the site of the largest genocide since World War II, when Cambodians killed over a million of their fellow Cambodians, somewhere between one-seventh and one-third of their country’s entire population. While some victims were “merely” starved to death, others were tortured to death, or killed by their own parents. My Cambodian guides at Angkor in 2008 included one man who had lived through the horrors of Pol Pot and reluctantly answered our questions about what it was like then. Any people with any hint of usefulness for anything other than being a farm laborer—people wearing eyeglasses, people able to speak a second language besides Khmer, people with education—were killed. In a radical remaking of Cambodian society that dwarfed even the re-makings of North Korea and Albania and sought to turn back the clock a thousand years to the days of Angkor, cities were evacuated; money, religion, markets, private property, and businesses were abolished; hospitals, schools, and stores were closed; no books or newspapers could be published; and everyone had to wear black clothes, eat in communal kitchens, and be a rice farmer.

Yet visitors today are struck by Cambodians as being nice, meek, gentle, peaceful people. How could people normally so mild have seesawed into such thoroughgoing savagery? Underneath that meek exterior, many Cambodians must have been seething with repressed fury. Is there anything about Cambodia’s history, and about the world society that enabled Cambodians to erect such a mighty city and empire, that could help us understand the plight and the explosion of modern Cambodia?

Several features of the environment in which the Khmer Empire developed are crucial for understanding the empire’s rise and fall and its capital’s glory. At its maximum extent, the empire controlled one-third of mainland Southeast Asia. While its heartland was the basin of the Lower Mekong River in Cambodia, it expanded to encompass much of the territory of the adjacent modern countries of Vietnam, Laos, and Thailand between latitudes 9 and 20 degrees north. As modern Americans who fought there during the Vietnam War remember painfully well, this is a hot tropical environment in which temperatures are almost constantly above 70 degrees Fahrenheit, even at dawn after the coldest night of the coldest month of the year. Warm clothes and fires for warmth are only rarely necessary.

Angkor’s major environmental challenges were related to rain and water. The area has a monsoon climate, a phrase that evokes images of torrential rain. Actually, the mean annual rainfall of Angkor is only 59 inches, which is barely higher than that of New York City. Compared to some of my New Guinea study sites and my home in Los Angeles, with annual rainfalls of 400 and 15 inches respectively, Angkor is neither especially wet nor especially dry.

Instead, Angkor’s water problems stem from the fact that rainfall varies predictably between seasons and unpredictably from year to year. Most of the rain comes in the so-called summer monsoon from June to November. The winter months, from December to May, are relatively dry, limiting the growing season for crops unless one can store rain falling in the rainy season—which the Khmer did do, as we shall see. As for variation between years, annual rainfall can be as low as 38 inches, or as high as 91 inches. That would expose Angkor’s inhabitants to the opposite risks of crop failures from droughts or from flooding, unless they had systems for storing rain in wet years to release in dry years and for controlling and quickly disposing of rain runoff during flood seasons. As we shall see, the Khmer water management system also coped with those problems for many centuries, until it finally became overwhelmed by extreme weather swings between severe droughts and severe floods.

To the south of Angkor are low mountains rising from the coast of the Gulf of Thailand. As a result, the Khmer Empire was oriented inland rather than toward maritime trade, and only in 1960 did Cambodia finally complete its first coastal seaport accessible to deep-draft oceangoing ships. To the north of Angkor at a distance of about 12 miles are the Kulen Hills, so steep that their slopes are prone to massive soil erosion if they become deforested. (That also emerged as a problem for the Khmer Empire.) From the Kulen Hills southward beyond Angkor, the terrain is extremely flat, with an average slope of only 0.1%. That created big problems of controlling water flows across the plains for the Khmer engineers of Angkor. It also created big problems of sanitation, because water flows across those flat plains were so slow. Rivers and channels in the plains supplied water for drinking, cooking, and bathing, but also functioned as the area’s sewers. A Chinese visitor to Angkor at its height commented on the frequency of cases of dysentery, which (the visitor wrote) proved fatal for 90% of the victims.

An outstanding feature of the Khmer heartland, and probably the main reason for the location of the Khmer capital at Angkor, is Southeast Asia’s largest lake, called the Tonle Sap (Lake Tonle), which constitutes an expansion of the Tonle River, one of the four branches of the Mekong River. In the dry season the shallow lake shrinks to an area of 1,000 square miles. But in the wet season the volume of water coming down the Mekong’s main branch is so large that the dry-season flow of water from the lake down the Tonle River reverses, and floodwaters back up the river into the lake, whose area quadruples to 4,000 square miles. That means 3,000 square miles of floodplain that is seasonally watered by natural flooding and is ideal for growing rice.

A second benefit of Lake Tonle for nearby Angkor’s inhabitants was as a transport artery to the Mekong River and the sea. The remaining advantage was the lake’s high biological productivity, supporting per cubic yard the greatest concentration of freshwater fish in the world, thanks to the sediments carried into the lake each year with the Mekong River floods. An early French visitor to the lake wrote, “The fish in it are so incredibly abundant that when the water is high they are actually crushed under the boats, and the play of oars is frequently impeded by them.” The lake accounts for most of Cambodia’s fisheries today, supplies most of the dietary protein of Cambodians, and is responsible for their enjoying one of the world’s highest levels of fish consumption.

While rice was the food staple of the Khmer in the past as it is of modern Cambodians, much of the Cambodian plain rates as land of only medium to poor quality for growing rice. The soils are mostly sandy and low in nutrient content. Of the three main methods for growing rice in Cambodia, one, practiced especially in hilly uplands, is swidden, or slash-and-burn, agriculture, wholly dependent on rain falling on the fields for watering the crop. The second and most extensive method, but still not especially productive by Chinese and Japanese standards, is rice paddies on flat terrain flooded by rain. The most productive method, which yielded most of the rice consumed by the ancient Khmer, is termed flood-retreat farming: rice is planted in fields into which water stored in upstream impoundments is released.

Thus, Angkor’s environment offered some advantages—especially those associated with the lake, and with the large area of flat plain available in the Lower Mekong Basin. But it also posed problems. The Khmer solved those problems brilliantly for many centuries, thereby succeeding in establishing a great city and empire, but eventually became defeated by the problems.

Who are the Khmer, and how did their empire arise? Today the Khmer constitute 90% of Cambodia’s population, and they already dominated the area of Angkor at least 1,400 years ago, to judge from preserved Khmer inscriptions on stone. The Khmer language belongs to the Austroasiatic language family, consisting of about 150 languages spoken in scattered areas from India to North Vietnam and the Malay Peninsula and mostly surrounded by speakers of other language families (especially the Sino-Tibetan and Tai-Kadai families). That fragmented distribution suggests that the latter two families have been encroaching on Austroasiatic lands, and indeed we know that the Thais and Vietnamese have been expanding in historic times. The only other Austroasiatic language besides Khmer likely to be familiar to most readers of this book is the distantly related Vietnamese language, which has become much more heavily modified through contact with Chinese (e.g., in becoming a tonal language) than has the Khmer language.

Until as recently as 5,000 years ago, all peoples of tropical Southeast Asia were stone tool-using hunter-gatherers, as was formerly true of everybody everywhere in the world until the origins of agriculture. Rice farming reached Cambodia from southern China by 2000 B.C., and the subsequent arrival of efficient iron tools around 500 B.C. fueled increased food production and a population explosion. By around A.D. 200, archaeological excavations in Cambodia revealed the existence of modest-sized towns and small kingdoms. From A.D. 245 onward, Chinese imperial records describe a cunning, malicious people of a land named “Funan” which evidently lies in modern Cambodia, which sent trade missions or tribute to China, and whose kings rode on the backs of elephants. Despite those contacts with China, Cambodia became much more strongly influenced by contacts with India, whence (beginning by around A.D. 300) arrived the Hindu and Buddhist religions, writing derived from South India’s Brahmi script, and use of the Sanskrit language for religious texts. The first inscription in the Khmer language itself dates to A.D. 611.

Political power in those earliest Indianized kingdoms of Cambodia was concentrated in the lower Mekong Delta, permitting access to coastal trade. By around A.D. 600, though, power shifted inland, where competing kingdoms were building surprisingly large towns, temples, and reservoirs. In A.D. 802 the independent kingdoms were finally unified under King Jayavarman the Second, regarded as the founder of the Khmer Empire, who chose the Angkor region for his capital’s site. For the next five centuries the empire was ruled by a succession of 24 kings, all with long Indianized names like “Udayadityavarman the Second,” “Dharamindravarman,” and “Jayavarmadiparameshvara.” Every half century or so, a king’s death led to a contest over his succession, and the empire disintegrated into several pieces before being put back together again.

Successive kings outbid each other by taking the large building projects of pre-unification kings as models and scaling them up into huge, then gigantic, and finally world-record humongous projects. For instance, the third king after unification, Indravarman the First, inspired by the big reservoirs of pre-unification kingdoms, broke previous records by launching the construction, five days after he was crowned, of a rectangular reservoir 2.3 miles long and 0.5 miles wide, modestly named after himself as the Indratataka (“Sea of Indra”). His successor, King Yashovarman the First, then built another rectangular reservoir eight times larger, 4.7 by 1.1 miles, also modestly named after himself as the Yashodharatataka, or East Baray. Another century had to pass before King Suryavarman the First could barely top that record with the 5-by-1.4-mile West Baray, one of the largest structures built by humans before the modern industrial era. Two centuries later, King Jayavarman the Seventh, busy with other construction such as the city of Angkor Thom and the Bayon temple, had to swallow his pride and attach his name to a humble new reservoir, the Jayatataka (“Sea of Jaya”), measuring a mere 2.2 by 0.6 miles. These are among the Khmer structures visible to aliens out there who might be viewing the Earth from space.

At the same time as Indianized names, writing, and religion were thriving at Angkor, Chinese influence continued. The Khmer sent embassies to the imperial court in China, which reciprocated by sending embassies and products to Angkor. Stone carvings at Angkor depict Chinese inventions such as pontoon bridges and multi-projectile artillery. Remains of Chinese Tang, Song, and later pottery are scattered all over Angkor.

Angkor’s grandiose royal courts and construction did not come cheaply to the empire’s peasants, who were taxed in the form of rice deliveries and labor. It’s estimated that construction of the West Baray would have taken the efforts of 200,000 peasants working for three years. One king required 4,000 concubines, and just one medium-sized temple had to be staffed by 1,000 administrators, over 600 dancers, 95 professors, and assorted others, adding up to a total of 12,640 functionaries, all of whom had to be fed. When I first heard these numbers, I began to get a glimmering of how centuries of exploitation of Cambodian peasants by extravagant elites could have had something to do with the repressed fury that exploded under Pol Pot.

We shouldn’t be deceived by all those temples, statues of Buddha, and beautiful reservoirs into thinking of Angkor’s kings as a peace-loving bunch. The Khmer were constantly fighting against each other and against their western neighbors the Thai, their eastern neighbors the Cham of South Vietnam, and their northeastern neighbors the Vietnamese of North Vietnam. The empire eventually conquered not just the areas of modern Cambodia and Laos but also much of southern Vietnam and Thailand and a sliver of southeastern Myanmar. Abundant stone carvings depict in vivid detail artillery, shields, armor, war chariots, cavalry mounted on horses and elephants, infantry battles, and naval warfare using ships with grappling hooks and dozens of rowers. Like ancient Rome, Angkor became filled with war plunder, including bronze, silver, and gold from conquered towns and shrines.

The distribution of population in urbanized societies familiar to us is inhomogeneous and hierarchical. That is, within a given area most of the landscape is used for farmland or industry and occupied at low population density, while a much smaller fraction of the landscape is urban and occupied at high density. The urban areas form a hierarchy, with at the top a large metropolis, below which in size and population come some medium-sized cities, then small cities, towns, and villages, surrounded by farmland which is clearly distinct from cities.

However, in the Khmer Empire, at least in its well-studied central core, all of those middle levels of hierarchy were missing: the only major city was Angkor, below which there were only small provincial centers. In that vast urban core the distinction between urban areas and farmland was blurred or absent. Angkor was instead a low-density city in which rice fields lay immediately outside the walls of temples, and the city itself consisted largely of rice fields with farmers’ houses spread out in clusters and lines across the landscape. Angkor’s area, of about 400 square miles, was thus far greater than that of the familiar high-density pre-industrial cities of Eurasia, such as 19th-century Tokyo (known as Edo), medieval Constantinople, 7th-century Baghdad, Rome during the Roman Empire, and other European cities before the 16th century, all of which were less than 40 square miles in area and mostly less than 10 square miles. While Angkor’s population density was lower than that of these high-density cities, its vastly greater area resulted in Angkor’s total population, estimated at about 750,000, approached that of a great contemporary Chinese capital.

It is becoming apparent to archaeologists that such low-density agriculturally based pre-industrial cities used to be a more widespread phenomenon in the moist tropics than we had realized. The growing list of examples includes the great Classic Lowland Maya cities until the 9th century A.D., such as Tikal and Copán; the Sri Lanka cities of Anuradhapura and Pollonaruwa, from around the 4th century B.C. until the 12th century A.D.; Bagan in 13th-century Myanmar; two cities of Angkor’s enemies, the Cham capital of My Son in Vietnam until the 13th century and the Thai capital of Sukhotai from A.D. 1238 to 1438; and perhaps also the 9th- and 10th-century Javan center around the temples of Borobudur and Prambanan. We have no European eyewitness account of any of those low-density tropical cities at their peaks, because all had declined or been abandoned by the time that Europeans began to explore the world, around A.D. 1500. Evidently, in the long run there was something apparently unstable about this city model: what was it? I shall return to this question for the case of Angkor.

Despite the lack of contemporary European descriptions of Angkor, we do have a lengthy account by a Chinese visitor, the commercial attaché Zhou Daguan, who spent a year in Angkor, from A.D. 1295 to 1296, at the end of King Jayavarman the Eighth’s reign. By astonishing good fortune, a copy of part of Zhou’s account was rediscovered in Beijing in the 19th century. His detailed descriptions of daily life at Angkor complement the inscriptions that tell us about the temple administrations and the depictions of ceremonies and wars on Angkor’s temple bas-reliefs that provide our only other information. Imagine how much more we would understand about ancient Maya life if we had an account such as Zhou’s for the Maya city of Tikal at its prime! To give you a flavor of what Zhou saw, here are some quotes from his Memoirs on the Customs of Cambodia:

On the extravagantly luxurious architecture: “In the center of the capital is a gold tower (the Bayon). . . . To the east of it is a gold bridge flanked by two gold lions . . . about a Li [0.3 miles] north of the gold tower there is a bronze tower [the Baphuon] . It is even taller than the gold tower, and an exquisite sight.... Ten Li [three miles] to the east of the city wall lies the East Lake [East Baray]. It is about 100 Li [30 miles] in circumference. In the middle of it is a bronze reclining Buddha with water continually flowing from its navel.” Zhou’s account was confirmed by the spectacular rediscovery of part of that colossal Buddha statue in 1936, actually in the West Baray; Zhou seems to have reversed his compass directions.

As a commercial attaché, Zhou was especially interested in trade between Angkor and China. He lists the main exports from Angkor to China, in descending order of preference, as: bright blue kingfisher feathers, elephant tusks, rhinoceros horns, beeswax, wood incense, cardamom, resin, lacquer, medicinal oil, and pepper. Angkor’s main imports from China were gold, silver, silk fabrics, tin goods, lacquered trays, celadon ware, mercury, paper, and saltpeter for making gunpowder.

On slaves: “If young and strong, slaves may be worth 100 pieces of cloth; when old and feeble, they can be had for 30 or 40 pieces. They are permitted to lie down or be seated only beneath the floor of the house. To perform their tasks they may go upstairs, but only after they have knelt, bowed to the ground, and joined their hands in reverence.... If they have committed some misdemeanor, they bow their heads and take the blows without daring to make the least movement. If a slave should run away and be captured, a blue mark would be tattooed on his face; moreover, an iron collar will be fitted to his neck, or shackles to his arms or legs.”

On interrogation and punishment of criminals: “If an object is missing, and accusations brought against someone who denies the charge, oil is brought to boil in a kettle and the suspected person forced to plunge his hand into it. If he is truly guilty, the hand is cooked to shreds; if not, skin and bones are unharmed. Such is the amazing ways of these barbarians.... [For punishment of serious offenses], a ditch is dug into which the criminal is placed, earth and stones are thrown back and heaped high, and all is over.”

On the strong sex drive of Cambodian women: “One or two days after giving birth to a child they are ready for intercourse: if a husband is not responsive he will be discarded. When a husband is called away on matters of business, they endure his absence for a while; but if he is gone as much as 10 days, the wife is apt to say, ‘I am no ghost; how can I be expected to sleep alone?’ Though their sexual impulses are very strong, it is said some of them remain faithful.”

Zhou’s brief dismissive assessment of Khmer military skills: “Generally speaking, these people have neither discipline nor strategy.”

The Khmer also had books of their own. Religious books were made of palm-leaf fronds inscribed with a stylus, and the incisions were then filled with black pigment. Secular books were written with either white chalk pencils or black ink on accordion-like screenfolds of either black or white paper, respectively. Alas, those materials are perishable in a hot humid climate, and all of Angkor’s books without exception are lost. We can only guess what those books would have told us about Khmer history, society, science, and philosophy. It’s as if we were trying to evaluate the ancient Greeks despite having lost all the writings of Homer, Plato, Aristotle, Thucydides, Herodotus, Sappho, and Sophocles.

At the time when France established a protectorate over Cambodia in 1863, Angkor was largely overgrown by jungle, but the huge temples and reservoirs and major canals were visible. Over the following century, as French archaeologists cleared and mapped the area and reconstructed ruined structures, a debate arose over the reservoirs’ function. In the 1980s British and American scholars began to favor the view that they were ornamental and used just for rituals, while French scholars (especially after the work of Bernard-Philippe Groslier) viewed Angkor as a “hydraulic city” dependent on the reservoirs for irrigating rice fields. A strong objection to Groslier’s view was that the reservoirs appeared to lack inlets and outlets, and that lack would have made them useless for distributing water to fields.

The resolution of this debate had to await the end of the Cambodian civil war and Pol Pot era, the application of new mapping techniques, and the launch in 2002 of a joint project between Australian, French, and Cambodian archaeologists. The research teams involved are from the University of Sydney and the École Française d’Extrême-Orient (teams headed by Roland Fletcher and Christophe Pottier respectively), in collaboration with the Cambodian authority that manages Angkor. A key advance involved the use of aerial radar imaging, which can penetrate clouds and detect variations in surface roughness and vegetation and moistness, and thereby recognize features invisible to observers on the ground. Those radar images can then be followed up by ground searches for remains of bricks, ceramics, and other direct evidence of crumbled and buried structures. The first radar images of Angkor, obtained in 1994 from the space shuttle Endeavor, were greatly expanded by a NASA/Jet Propulsion Laboratory (JPL) airborne radar survey in September 2000 after intensive fieldwork by Christophe Pottier in the 1990s. When the radar images are combined with aerial photography and other surveys, the results are a high-resolution map for the whole of the 400-square-mile core area of the Greater Angkor urban complex.

The map reveals a web of canals with raised banks serving as roads, reservoirs of all sizes from the gigantic baray down to small ponds for each house, and a fine grid of rice fields hidden below the present-day land surface with its modern rice fields. The entire landscape from Lake Tonle north to the Kulen Hills was cleared of its original forest and converted into a low-density city whose land was devoted to rice production, houses, and temples. From the central area with the largest temples radiated six highways on large embankments, fitted with bridges. Ground surveys located the long-sought inlets and outlets of the baray, connecting them to the canal network. In addition to the famous major temples such as Angkor Wat, Baphuon, and Bayon, there were hundreds of minor local temples, each on top of a square mound 65 feet long on each side, surrounded by a moat that was crossed by a causeway on the eastern side.

Angkor’s landscape was divided into three zones, each with different roles in water management. The northeastern zone, including the Kulen Hills, served to collect water from the rivers running off from the hills. The central zone, including the big baray, stored the collected water. The southwestern zone was crisscrossed by canals that either distributed water over the rice fields or disposed of it quickly to the lake, depending on needs at the moment. One large channel, following the shortest route from the West Baray to the lake, would have served to dump excess water after heavy rains. The canals included right-angle turns and cross channels to slow the water flow, decrease erosion of the channel banks, remove suspended sediment, and prevent channels from silting up. The system’s plan and its components’ dates allow one to trace its development from the 8th to the 14th century A.D., as rivers successively farther north and west became tapped for collecting water.

The whole system represents engineering on a vast scale. Rivers were re-engineered to flow north to south instead of their original natural course of northeast to southwest. The direction of flow in one stretch of river was reversed. The big reservoirs were on a modern industrial scale, with in one case banks over 300 feet wide and 30 feet high. The resulting system is so complicated that we are only beginning to understand its operation, and there remain large elements whose function is still a mystery. Obviously, it was designed to manage water, and not just to look pretty and serve as ritual sites, as some archaeologists formerly assumed. The system’s most obvious purpose was for risk management: to smooth out fluctuations in water availability between seasons and years, and to ensure adequate water supply for rice farming even if the monsoon rains should be poor or fail in one year. The system thus served to guarantee the food supply for the capital’s inhabitants, and it illustrates systems used throughout the empire to fuel its population and its power.

By the early 1200s the Khmer Empire under King Jayavarman the Seventh was Southeast Asia’s largest and most powerful state, with its capital at Angkor. In the 1860s the French arrived to find the Khmer kingdom small and weak, with its capital 140 miles south of Angkor at Phnom Penh and only a handful of small villages in Angkor’s former urban area. What happened between the early 1200s and 1860s to produce that result?

A century after Jayavarman the Seventh, the empire underwent a slow decline. While its gold towers and pompous ceremonies still impressed the Chinese visitor Zhou Daguan in 1295-1296, construction of big stone monuments had ceased, the last traditional temple was dedicated in 1295, and the last Sanskrit text was inscribed in 1327. At some time after the early 15th century, new capitals began to develop east and south of Lake Tonle in the vicinity of Phnom Penh, and people seem to have gradually moved away from Angkor. While a Khmer ruler in the 1600s still boasted that he had re-gilded the towers of Angkor Wat, Angkor’s urban area became abandoned after about 1660, and the Khmer kingdom continued to shrink.

One cause of the Khmer decline was the rise of powerful enemies. From the 1200s, both the Thais and the Vietnamese pressed southward, the former on the west from southern China and the latter on the east of Angkor, squeezing the Khmer Empire in a vice. The Thais claimed to have captured Angkor for a while in the 15th century, while the Vietnamese seized the Mekong Delta from the Khmer in the 1700s.

Another cause was a change in the focus of the Khmer economy, from its original inland agricultural emphasis to an increasing involvement in the maritime trade along Southeast Asia’s coast between China, India, and the Islamic heartland. That would have provided a motive for shifting the capital from inland Angkor toward Phnom Penh, with a more direct connection to the coast down the Mekong River. But the Vietnamese expansion later made that Khmer maritime involvement more difficult.

A further big factor that now needs to be added to our understanding of Angkor’s decline is a change in climate. Studies of the water network have yielded seemingly contradictory evidence of both floods and droughts in the 14th and 15th centuries. On the one hand, the big canals to the south of Angkor became filled with coarse-grained sand, implying heavy rains and strong flooding. On the other hand, the exit channels of the big baray were blocked, while the East Baray exit was reconstructed to make it narrower and was then converted from an outlet to an inlet—sure signs of water shortages and attempts to keep the reservoir water levels high. The resolution to this paradoxical combination of floods and droughts has just emerged from publication of a 979-year-long record of tree-ring widths, similar to the tree-ring records that have illuminated climate change in the Anasazi area as I discussed in Chapter 4. These records show that monsoon rainfall in Southeast Asia became much more variable after A.D. 1350, with severe droughts from about 1336 to 1374 and from 1400 to 1425, and exceptionally heavy rainfall in some years between and just before and after those two droughts. The period 1322 to 1453 contains a disproportionate number of both the driest years and the wettest years of the last millennium.

We can now place Angkor’s decline within the framework of the five sets of factors that I proposed in this book’s prologue for understanding a society’s success or failure. First, the Khmer did inflict unintentional damage on their environment: they deforested the Angkor plain and the slopes of the Kulen Hills. Without trees to slow rain runoff, heavy monsoons eroded soil and dislodged sediment that was swept into canals, and floods incised the Siem Riep River’s channel. That river is now about 20 feet below the Angkorian land surface. Second, climate change exposed the Angkor area to conditions both drier and wetter than Angkor’s system was designed to accommodate. Third, the Khmer, like the Roman Empire and the Greenland Norse, faced growing problems from hostile neighbors. Fourth, friendly trade partners played a role, by offering the Khmer maritime economic opportunities more attractive than the inland opportunities available at Angkor, but then those opportunities became restricted. Finally, the Khmer Empire responded to the attractions and the problems of the Angkor environment by becoming committed to an increasingly huge, complex, and hard-to-maintain water management system from which there was no going back. All five of those factors interacted: climate change and erosion weakened the Khmer to the point where they could no longer resist their enemies, no longer maintain and improve their water management system, and turned away from an agricultural economy to maritime trade until shifting trade routes and political power made that, too, less profitable.

We can also place the Khmer decline within the spectrum of collapses, from quick and lethal to slow and non-lethal. At the former extreme lies the end of Greenland’s Western Settlement, where everybody may have died in a single winter. Angkor’s decline seems to lie toward the opposite extreme: it extended over centuries, people gradually moved away, and there is no evidence of massive die-offs of people. But the result, nevertheless, was unequivocally a collapse: scattered villages in the former metropolitan heartland, on the site of what once had been a world-class city and the capital of the region’s most powerful empire.

For all of us fascinated by the glory and the mystery of Angkor, these are exciting times. We have gained much detailed knowledge about the city in the last decade, building on the previous century of archaeological investigation. But big questions remain unsolved, and the next decade promises to be even more exciting. Here is my short list of five sets of questions to which I would love to know the answers by the year 2020:

Where did Angkor’s population get its wood for construction and fuel? The city’s 750,000 people must have needed huge quantities of wood for building houses and other structures, and to make charcoal for cooking. Yet the original forest of the Angkor plains and much of the Kulen Hills was cut down, and it boggles the mind to imagine that the trees then planted by people around their houses could have met the needs of such a large population.

How did the Khmer manage water? They surely had methods for recording the monsoons and predicting the resulting water flows. What roles did the shrines in the middle of the baray play in managing the canal network? How did the Khmer move water around the landscape without pumps? The West Baray was not dug into the ground but was built on the ground, and cleverly positioned so that the intake canals provided a water surface several feet above ground level at the exit. How did Khmer water engineers shift water flows from one channel to another as needed? Did the canals have movable gates?

How did the whole water management system function? It includes many big structures that have been located but whose purpose remains unknown.

Why has it taken so long for Angkor to regain its formerly dense human population, after the city became largely abandoned in the 1600s? The same puzzle arises for the Southern Maya Lowlands, which were still sparsely populated upon Cortés’s arrival in 1524, six or seven centuries after the Classic Maya collapse.

Finally, I mentioned at the beginning of this chapter that there used to be numerous other low-density cities besides Angkor in the seasonally wet tropics elsewhere in the world, but that all of those cities had declined before the European expansion of the last five centuries. What was the Achilles’ heel that made Angkor and all those other cities unsustainable in the long run?

If you find an error or have any questions, please email us at Thank you!