WHEN COOKING BEGAN

"The introduction of cooking may well have been the decisive factor in leading man from a primarily animal existence into one that was more fully human.”
—CARLETON S. COON, The History of Man

Archaeologists are divided about the origins of cooking. Some suggest that fire was not regularly used for cooking until the Upper Paleolithic, about forty thousand years ago, a time when people were so modern that they were creating cave art. Others favor much earlier times, half a million years ago or before. A common proposal lies between those extremes, advocated especially by physical anthropologist Loring Brace, who has long noted that people definitely controlled fire by two hundred thousand years ago and argues that cooking started around the same time. As the wide range of views shows, the archaeological evidence is not definitive. Archaeology offers only one safe conclusion: it does not tell us what we want to know. But though we cannot solve the problem of when cooking began by relying on the faint traces of ancient fires, we can use biology instead. In the teeth and bones of our ancestors we find indirect evidence of changes in diet and the way it was processed.

Yet the archaeological data leave no doubt that controlling fire is an ancient tradition. In the most recent quarter of a million years, there is sparkling evidence of fire control, and even occasionally of cooking, by both our ancestors and our close relatives the Neanderthals. The most informative sites tend to be airy caves or rock shelters, many of them in Europe. In Abri Pataud in France’s Dordogne region, heat-cracked river cobblestones from the late Aurignacian period, around forty thousand years ago, show that people boiled water by dropping hot rocks in it. At Abri Romani near Barcelona, a series of occupations dating back seventy-six thousand years includes more than sixty hearths together with abundant charcoal, burnt bones, and casts of wooden objects possibly used for cooking. More than ninety-three thousand years ago in Vanguard Cave, Gibraltar, three separate episodes of burning can be distinguished in a single hearth. Neanderthals heated pinecones on these fires and broke them open with stones, much as contemporary hunter-gatherers have been recorded doing, to eat the seeds.

Our ancestors were using fire in the Middle East and Africa as well. In a cave at Klasies River Mouth, a coastal site in South Africa from sixty thousand to ninety thousand years ago, burnt shells and fish bones lie near family-size hearths that appear to have been used for weeks or months at a time. Between 109,000 and 127,000 years ago in the Sodmein Cave of Egypt’s Red Sea Mountains, people appear responsible for huge fires with three distinct superimposed ash layers and the burnt bones of an elephant. Charred logs, together with charcoal, reddened areas, and carbonized grass stems and plants, date to 180,000 years ago at Kalambo Falls in Zambia. Back to 250,000 years ago in Israel’s Hayonim Cave, there are abundant hearths with ash deposits up to 4 centimeters (1.6 inches) thick. Such sites show that people have been controlling fire throughout the evolutionary life span of our species, Homo sapiens, which is considered to have originated about two hundred thousand years ago.

Because evidence about controlling fire is inconsistent before the last quarter of a million years, it is often argued that the control of fire was unimportant or absent until that time. But that idea is now particularly shaky because the older part of the record, going back in time from a quarter of a million years ago, has been improving in quality. Two sites in particular give tantalizing hints of what earlier people were doing with fire.

An ancient fireplace at Beeches Pit archaeological site in England securely dated to four hundred thousand years ago lies on the gently sloping bank of an ancient pond. Eight hand axes attest to the presence of humans. Dark patches about one meter (three feet) in diameter with reddened sediments at the margins show where burning occurred. Tails of ashlike material lead down from the fires toward the pond, while the upper side contains numerous pieces of flint. The flints have been knapped, or broken by a sharp blow, and many are burnt. A team led by archaeologist John Gowlett fitted the flint pieces together, and one of the various refits showed that someone had been knapping a heavy core (1.3 kilograms, or 2.9 pounds) until a flaw became obvious. The knapper abandoned it, and two flakes from the series fell forward and were burnt, indicating that the toolmaker apparently had been squatting next to a warming blaze.

Another four-hundred-thousand-year-old site, at Schöningen in Germany, has yielded more than a half dozen superb throwing spears carved from spruce and pine, together with the remains of at least twenty-two horses that appear to have died at the same time as one another, apparently killed by humans. Cut marks show that people removed meat from the horses. At the same site were numerous pieces of burnt flint, four large reddened patches about one meter in diameter that appear to have been fireplaces, and some pieces of burnt wood including a shaped stick, also made from spruce, that had been charred at one end as if it had been used as a poker, or perhaps held over coals to cook strips of meat. This exceptional lakeshore find by archaeologist Hartmut Thieme represents the earliest evidence of group hunting. Thieme suggests that after people killed the horse group, they found themselves with far more food than they could consume at the time. They settled for several days and built the fires along the lakeshore to dry as much meat as possible.

Prior to half a million years ago, there is no evidence for the control of fire in Europe, but ice covered Britain for much of the time between five hundred thousand and four hundred thousand years ago, and glaciers would have swept away most evidence of any earlier occupations. Farther south, however, fire-using is strongly attested at 790,000 years ago. In a well-dated site called Gesher Benot Ya’aqov, next to Israel’s Jordan River, hand axes and bones were first discovered in the 1930s, and in the 1990s, Naama Goren-Inbar found burnt seeds, wood, and flint. Olives, barley, and grapes were among the species of seeds found burned. The flint fragments were grouped in clusters, suggesting they had fallen into campfires. Nira Alperson-Afil analyzed these dense concentrations. She concluded that the early humans who made these fires “had a profound knowledge of fire-making, enabling them to make fire at will.”

Gesher Benot Ya’aqov is the oldest site offering confident evidence of fire control. Before then we find only provocative hints. Archaeological sites between a millionand a million and a half years old include burnt bones (at Swartkrans in South Africa), lumps of clay heated to the high temperatures associated with campfires (Chesowanja, near Lake Baringo in Kenya), heated rocks in a hearthlike pattern (Gadeb in Ethiopia), or colored patches with appropriate plant phytoliths inside (Koobi Fora, Kenya). But the meaning of such evidence as indicating human control of fire is disputed. Some archaeologists find it totally unconvincing, regarding natural processes such as lightning strikes as likely explanations for the apparent use of fire. Others accept the idea that humans controlled fire in the early days of Homo erectus as well established. Overall, these hints from the Lower Paleolithic tell us only that in each case the control of fire was a possibility, not a certainty.

Evidence of humans controlling fire is hard to recover from early times. Meat can be cooked easily without burning bones. Fires might have been small, temporary affairs, leaving no trace within a few days of exposure to wind and rain. Even now hunter-gatherers such as the Hadza, who live near the Serengeti National Park in northern Tanzania, may use a fire only once, and they often leave no bones or tools at the fire site, so archaeologists would not be able to infer human activity even if they could detect where burning had occurred. The caves and shelters that preserve relatively recent evidence of fire use tend to be made of soft rock, such as limestone, which erodes quickly, so the half-lives of caves average about a quarter of a million years, leaving increasingly few opportunities to find traces of fire use from earlier periods. From the past quarter of a million years there are sites of human occupation where people must have used fire, yet there is no sign of it.

There are also mysterious reductions in the frequency of finding evidence of fire, such as one that followed an interglacial period in Europe from 427,000 years ago to 364,000 years ago, when fire evidence was relatively abundant. In short, while humans have certainly been using fire for hundreds of thousands of years, archaeology does not tell us exactly when our ancestors began to do so.

The inability of the archaeological evidence to tell when humans first controlled fire directs us to biology, where we find two vital clues. First, the fossil record presents a reasonably clear picture of the changes in human anatomy over the past two million years. It tells us what were the major changes in our ancestors’ anatomy, and when they happened. Second, in response to a major change in diet, species tend to exhibit rapid and obvious changes in their anatomy. Animals are superbly adapted to their diets, and over evolutionary time the tight fit between food and anatomy is driven by food rather than by the animal’s characteristics. Fleas do not suck blood because they happen to have a proboscis well designed for piercing mammalian skin; they have the proboscis because they are adapted to sucking blood. Horses do not eat grass because they happen to have the right kind of teeth and guts for doing so; they have tall teeth and long guts because they are adapted to eating grass. Humans do not eat cooked food because we have the right kind of teeth and guts; rather, we have small teeth and short guts as a result of adapting to a cooked diet.

Therefore, we can identify when cooking began by searching the fossil record. At some time our ancestors’ anatomy changed to accommodate a cooked diet. The change must mark when cooking became not merely an occasional activity but a predictable daily occurrence, because until then our ancestors would have sometimes had to resort to eating their food raw—and therefore could  not adapt to cooking. The time when our ancestors became adapted to cooked food also marks the time when fire was controlled so effectively that it was never lost again.

Anthropologists have sometimes suggested that humans could have controlled fire for reasons such as warmth and light for many millennia before starting to use it for cooking.

However, many animals show a spontaneous preference for cooked food over raw. Would prehuman ancestors have preferred cooked food also? Evolutionary anthropologists Victoria Wobber and Brian Hare tested chimpanzees and other apes in the United States, Germany, and Tchimpounga, a Congolese sanctuary. Across the different locations, despite different diets and living conditions, the apes responded similarly. No apes preferred any food raw.

They ate sweet potatoes and apples with equal enthusiasm whether raw or cooked, but they preferred their carrots, potatoes, and meat to be cooked. The Tchimpounga chimpanzees were particularly informative because there was no record of them having eaten meat previously, yet they showed a strong preference for cooked meat over raw meat. The first of our ancestors to control fire would likely have reacted the same way. Cooked food would have suited their palate the first time they tried it, just as a taste for cooked food, with its immediate benefits, is shared by a wide range of wild and domestic species. Chimpanzees in Senegal do not eat the raw beans of Afzelia trees, but after a forest fire has passed through the savanna, they search under Afzelia trees and eat the cooked seeds.

Why are wild animals pre-adapted in this way to appreciate the smells, tastes, and textures of cooked food? The spontaneous preference for cooked food implies an innate mechanism for recognizing high-energy foods. Many foods change their taste when cooked, becoming sweeter, less bitter, or less astringent, so taste could play a role in this preference, as some evidence suggests. Koko is a gorilla who learned to communicate with humans, and she prefers her food cooked. Cognitive psychologist Penny Patterson asked her why: “I asked Koko while the video was rolling if she liked her vegetables better cooked (specifying my left hand) or raw/fresh (indicating my right hand). She touched my left hand (cooked) in reply. Then I asked why she liked vegetables better cooked, one hand standing for ‘tastes better,’ the other ‘easier to eat.’ Koko indicated the ‘tastes better’ option.”

When primates eat, sensory nerves in the tongue perceive not only taste but also particle size and texture. Some of the brain cells (neurons) responsive to texture converge with taste neurons in the amygdala and orbito-frontal-cortex of the brain, allowing a summed assessment of food properties. This sensory-neural system enables primates to respond instinctively to a wide range of food properties other than merely taste, including such factors as grittiness, viscosity, oiliness, and temperature.

In 2004 such abilities in the human brain were reported for the first time. A team led by psychologist Edmund Rolls found that when people had foods of a particular viscosity in their mouths, specific brain regions were activated.

Those regions partly overlapped with regions of taste cortex that register sweetness. The picture emerging from such studies is that hard-wired responses to properties such as taste, texture, and temperature are integrated in the brain with learned responses to the sight and smell of food. So the mechanisms that allow animals to assess the quality of raw foods directly apply to cooked foods and allow them to choose foods of a good texture for easy digestion.

Rolls’s studies suggest that the proximate reasons chimpanzees and many other species like their meat and potatoes cooked may be the same as in humans. We identify foods that have high caloric value not just by their being sweet, but also by their being soft and tender. Our ancestors were surely prepared by their pre-existing sensory and brain mechanisms to like cooked foods in the same way. A long delay between the first control of fire and the first eating of cooked food is therefore deeply improbable.

A long delay between the adoption of a major new diet and resulting changes in anatomy is also unlikely. Studies of Galapagos finches by Peter and Rosemary Grant showed that during a year when finches experienced an intense food shortage caused by an extended drought, the birds that were best able to eat large and hard seeds—those birds with the largest beaks—survived best. The selection pressure against small-beaked birds was so intense that only 15 percent of birds survived and the species as a whole developed measurably larger beaks within a year. Correlations in beak size between parents and offspring showed that the changes were inherited. Beak size fell again after the food supply returned to normal, but it took about fifteen years for the genetic changes the drought had imposed to reverse.

The Grants’ finches show that anatomy can evolve very quickly in response to dietary changes. In the case of the drought year in the Galapagos, the change in diet was temporary and therefore so was the change in anatomy. Other data show that if an ecological change is permanent, the species also changes permanently, and again the transition is fast. Some of the clearest examples come from animals confined on islands that have been newly created by a rise in sea level. In fewer than eight thousand years, mainland boa constrictors that occupied new islands off Belize shifted their diets away from mammals and toward birds, spent more time in trees, became more slender, lost a previous size difference between females and males, and were reduced to a fifth of their original body weight.

According to evolutionary biologist Stephen Jay Gould, this rate of change is not unusual. Drawing from the fossil record, he suggested that fifteen thousand to twenty thousand years may be about the average time one species takes to make a complete evolutionary transition to another. While a species that takes many years to mature, such as our ancestors, would take longer to evolve than a rapidly growing species, such rapid rates of evolution are sharply inconsistent with some previous interpretations of the effects of cooking. Loring Brace suggested that the use of fire for softening meat began around 250,000 to 300,000 years ago, followed by a supposed drop in tooth size that began about 100,000 years ago. This would mean that for at least the first 150,000 years after cooking was adopted, human teeth showed no response. Because such a long delay before adapting to a major new influence does not fit the animal pattern, we can conclude that Brace’s idea is wrong. The adaptive changes brought on by the adoption of cooking would surely have been rapid.

In addition to following quickly, the changes would have been substantial. We can infer this from pairs of species in which lesser differences in diet have large effects. Take chimpanzees and gorillas, two ape species that often share the same forest habitat. In many ways their diets are very similar. Both choose ripe fruits when they are available. Both also supplement their diets with fibrous foods, such as piths and leaves. There is only one important difference in their food choice. When fruits are scarce, gorillas rely on foliage alone, whereas chimpanzees continue to search for fruit every day. Unlike gorillas, chimpanzees never survive only on piths and leaves—presumably because they are physiologically unable to do so.

The relative ability of these two apes to rely on foliage might at first glance appear to be a trivial matter—especially compared to the introduction of cooking. But many consequences follow from it. To find their vital fruits, chimpanzees must travel farther than gorillas, so they are more agile and smaller. There are differences in distributional range. Unlike chimpanzees, gorillas successfully occupy high-altitude forests without fruits, such as the Virunga Volcanoes of Rwanda, Uganda, and the Democratic Republic of Congo. Chimpanzees are limited to lower altitudes. Like other primates that are able to rely on a leaf diet, gorillas mature earlier, start having babies at a younger age, and reproduce faster.

Grouping patterns of these species also differ strikingly as a result of the difference in diet. The terrestrial foliage gorillas rely on is easily found and occurs in big patches, allowing their groups to be stable all year. But during food-poor seasons, chimpanzees are driven to travel alone or in small groups as they search for rare fruits. The difference in grouping patterns has further consequences. Gorillas form long-lasting bonds between females and males, whereas chimpanzees do not.

More than the relatively slight dietary difference that distinguishes gorillas from chimpanzees, cooked food had multiple differences from raw food. Effects of cooking include extra energy, softer food, fireside meals, a safer and more diverse set of food species, and a more predictable food supply during periods of scarcity. Cooking would therefore be expected to increase survival, especially of the vulnerable young. It should also have increased the range of edible foods, allowing extension into new biogeographical zones. The anatomical differences between a cooking and a precooking ancestor should be at least as big as those between a chimpanzee and a gorilla. So whenever cooking was adopted, its effects should be easy to find. We can expect the origin of cooking to be signaled by large, rapid changes in human anatomy appropriate to a softer and more energy-rich diet.

The search for such changes proves to be rather simple. Before two million years ago, there is no suggestion for the control of fire. Since then there have been only three periods when our ancestors’ evolution was fast and strong enough to justify changes in the species names. They are the times that produced Homo erectus (1.8 million years ago), Homo heidelbergensis (800,000 years ago), and Homo sapiens (200,000 years ago). These are therefore the only times when it is reasonable to infer that cooking could have been adopted.

Most recent was the evolution of Homo sapiens from an ancestor that is now usually called Homo heidelbergensis. It was a gentle process that began in Africa as early as three hundred thousand years ago and was largely complete by around two hundred thousand years ago. The transition was too recent to correspond to the origin of cooking, however, because Homo heidelbergensis was already using fire at Beeches Pit, Schöningen, andelsewhere four hundred thousand years ago. Nor does the transition to Homo sapiens show the kinds of change we are looking for. Homo heidelbergensis was merely a more robust form of human than Homo sapiens, with a large robust form of human than Homo sapiens, with a large face, less rounded head, and slightly smaller brain. Most of the differences between these two species are too small and not obviously related to diet. We can be confident that cooking began more than three hundred thousand years ago, before Homo sapiens emerged.

Homo heidelbergensis evolved from Homo erectus in Africa from eight to six hundred thousand years ago. The timing of the erectus-heidelbergensis transition provides a reasonably comfortable fit with the archaeological data on the control of fire becoming particularly scarce. The main changes in anatomy from Homo erectus to Homo heidelbergensis were an increase in cranial capacity (brain volume) of around 30 percent, a higher forehead, and a flatter face. These are smaller modifications than the differences between a chimpanzee and a gorilla, and the modifications show little correspondence to changes in the diet. So this Pleistocene transition does not look favorable. It is a possibility for when cooking began, but not a promising one.

The only other option is the original change, from habilines to Homo erectus. This shift happened between 1.9 million and 1.8 million years ago and involved much larger changes in anatomy than any subsequent transitions. Recall that in many ways habilines were apelike. Like the australopithecines, they appear to have had two effective styles of locomotion. They walked upright and can be reconstructed as having had sufficiently strong and mobile reconstructed as having had sufficiently strong and mobile arms to be good climbers. Their small size must have helped them in trees. They are estimated to have stood about 1 to 1.3 meters tall (3 feet 3 inches to 4 feet 3 inches) and appear to have weighed about the same as a chimpanzee, around thirty-two kilograms (seventy pounds) for a female and thirty-seven kilograms (eighty-one pounds) for a male. Despite their small bodies, they had much bigger chewing teeth than in any subsequent species of the genus Homo: the surface areas of three representative chewing teeth decreased by 21 percent from habilines to early Homo erectus. Habilines’ larger teeth imply a bulky diet that required a lot of chewing.

Homo erectus did not exhibit the apelike features of the habilines. In the evolution of Homo erectus from habilines, we find the largest reduction in tooth size in the last six million years of human evolution, the largest increase in body size, and a disappearance of the shoulder, arm, and trunk adaptations that apparently enabled habilines to climb well. Additionally, Homo erectus had a less flared rib cage and a narrower pelvis than the australopithecines, both features indicating that they had a smaller gut. There was a 42 percent increase in cranial capacity. Homo erectus was also the first species in our lineage to extend its range beyond Africa: it was recorded in western Asia by 1.7 million years ago, Indonesia in Southeast Asia by 1.6 million years ago, and Spain by 1.4 million years ago. The reduction in tooth size, the signs of increased energy availability in larger brains and bodies, the indication of smaller guts, and the ability to exploit new kinds of habitat all support the idea that cooking was responsible for the evolution of Homo erectus.

Even the reduction in climbing ability fits the hypothesis that Homo erectus cooked. Homo erectus presumably climbed no better than modern humans do, unlike the agile habilines. This shift suggests that Homo erectus slept on the ground, a novel behavior that would have depended on their controlling fire to provide light to see predators and scare them away. Primates hardly ever sleep on the ground. Smaller species sleep in tree holes, in hidden nests, on branches hanging over water, on cliff ledges, or in trees so tall that no ground predator is likely to reach them.

Great apes mostly build sleeping platforms or nests. The only nonhuman primate that regularly sleeps on the ground is the largest species of great ape, gorillas. Gorillas are safer on the ground than Homo erectus would have been because gorillas live in forests with few predators and they are relatively enormous. The most frequent ground sleepers are adult males, weighing around 127 kilograms (286 pounds). Smaller gorillas often sleep in trees.

The late Pliocene and early Pleistocene periods in Africa were rich in predators. In wooded areas from 4 million to 1.5 million years ago, our ancestors would have found saber-toothed cats. There was Megantereon, the size of a leopard, and Dinofelis, as big as a lion. In more open habitats there was the scimitar cat Homotherium, equally large. An extinct kind of lion and spotted hyena lived alongside our early ancestors, while modern lions and leopards have been present since at least 1.8 million years ago. There were also many large animals such as elephants, rhinoceroses, and buffalo-like ungulates that could stumble unawares onto an unconscious biped. The African woodlands would have been a very dangerous place to sleep on the ground.

Extrapolating from the behavior of living primates in predator-rich environments, the australopithecines and habilines surely slept in trees. Their habitats were well wooded and their upper-body anatomy suggests they climbed well. But what did Homo erectus do? The famous “Turkana boy,” a beautifully preserved specimen of Homo erectus dated between 1.51 and 1.56 million years ago provides excellent evidence that they climbed relatively poorly. Physical anthropologists Alan Walker and Pat Shipman have described the Turkana boy as committed to locomotion on the ground. His finger bones had lost the curved, robust shape of australopithecine fingers. His shoulder blade had the modern form, giving no indication of being adapted to the stresses of climbing with the arm above the shoulder. The Turkana boy is so well preserved that Walker was able to study the vestibular system of the inner ear, responsible for balance. Species that climb regularly have a large and characteristically shaped vestibular system. The Turkana boy’s is different from that of species that climb, but closely resembles the modern human system.

So the Turkana boy, like other Homo erectus, could not have climbed well and he therefore would have found it difficult to make the type of nest great apes sleep in. Chimpanzees take about five minutes to build their nests by standing on all fours where the nest is taking shape, bending branches toward themselves. They break some of the bigger ones and weave the branches together to form a platform that they finish off with a few leafy twigs that serve as cushions or pillows to make it comfortable. Making a nest depends on being able to move around easily on the end of a swaying branch. The long legs and flat feet of humans such as Homo erectus and modern people do not allow such agility. For a mother with a small infant, the gymnastic challenges of making a nest would have been particularly difficult given her need to cradle while she swayed in the tree.

Homo erectus therefore must have slept on the ground. But to do so in the dark of a moonless night seems impossibly dangerous. Homo erectus was as poorly defended a creature as we are, unable to sprint fast and dependent on weapons for any success in fighting. Surprised by a Dinofelis or a pack of hyenas at midnight, they would have been vulnerable.

If Homo erectus used fire, however, they could sleep in the same way as people do nowadays in the savanna. In the bush, people lie close to the fire and for most or all of the night someone is awake. When a sleeper awakens, he or she might poke at the fire and chat a while, allowing another to fall asleep. In a twelve-hour night with no light other than what the fire provides, there is no need to have a continuous eight-hour sleep. An informal system of guarding easily emerges that allows enough hours of sleep for all while ensuring the presence of an alert sentinel. To judge from records of attacks by jaguars, modern huntergatherers are safer in camp at night than they are on the hunt by day.

The control of fire could explain why Homo erectus losttheir climbing ability. The normal assumption is that when long legs were favored, perhaps as a result of the increasing importance of long-distance travel as humans searched for meat, it was harder for humans to climb efficiently, and Homo erectus therefore abandoned the trees. But since that argument does not explain how Homo erectus could sleep safely, I prefer an alternative hypothesis: having controlled fire, a group of habilines learned that they could sleep safely on the ground. Their new practice of cooking roots and meat meant that food obtained from trees was less important than it had been when raw food was the only option. When they no longer needed to climb trees to find food or sleep safely, naturalselection rapidly favored the anatomical changes that facilitated long-distance locomotion and led to living completely on the ground.

Two kinds of evidence thus point independently to the origin of Homo erectus as the time when cooking began. First, anatomical changes related to diet, including the reduction in tooth size and in the flaring of the rib cage, were larger than at any other time in human evolution, and they fit the theory that the nutritional quality of the diet improved and the food consumed was softer. Second, the loss of traits allowing efficient climbing marked a commitment to sleeping on the ground that is hard to explain without the control of fire.

The only alternative is the traditional theory that cooking was first practiced by beings that already looked like us—physically human members of the genus Homo. If this were true, by the time our ancestors adopted cooking, Homo erectus had long ago adapted to a soft, easily chewed diet of high caloric density. But as we have seen, cold-processing techniques such as grinding and blending provide relatively poor energy even when carried out by raw-foodists with modern equipment.

For more than 2.5 million years our ancestors have been cutting meat off animal bones, and the impact was huge. A diet that included raw meat as well as plant foods pushed our forebears out of the australopithecine rut, initiated the evolution of their larger brains, and probably inspired a series of food-processing innovations. But according to the evidence carried in our bodies, it would take the invention of cooking to convert habilines into Homo erectus, and launch the journey that has led without any major changes to the anatomy of modern humans.

By Richard Wrangham in "Catching Fire - How Cooking Made Us Human", Basic Books (an member of Perseus Books Group), New York, USA, 2009, excerpts pp.92-114 (chapter 4). Adapted and illustrated to be posted by Leopoldo Costa.