Biological Anthropology & Human Evolution

The big picture

It is just after dawn at a field camp on the eastern shore of Lake Turkana in northern Kenya, and the light is still low enough to be useful. You are sitting on an upturned crate at a plywood table with a fragment of cranial bone in front of you, about the size of your palm, the same brown as the sediment it came out of yesterday afternoon. You have a pair of sliding calipers and a waterproof notebook, and the entire intellectual apparatus of a science is bearing down on this one small object: how thick is the bone, how curved is the vault, how does it compare with the twenty other fragments in the box, and what, if anything, can honestly be said about the animal it came from.

Two hundred meters away a colleague is photographing and describing the layer the fragment eroded out of, because a bone without a layer is a bone without a date. In a lab in Nairobi a technician is preparing sediment for isotope analysis that will say something about whether this place was grassland or woodland. In Leipzig a team will try to recover protein sequences from a tooth out of the same deposit and will very likely fail, because this is equatorial Africa and heat destroys molecules. In a university basement somewhere, your measurements will be run against reference samples of chimpanzees, gorillas, and modern humans. And two valleys over, a different anthropologist is measuring the height and arm circumference of living children, because human biology did not stop happening when the fossils ran out.

All of that is biological anthropology, and this lesson is about why those wildly different activities belong to one field. By the end you should be able to say what the discipline studies, how it divides its labor, what makes its approach distinctive, and how it decides that a claim about the human past or present is actually supported.

A definition worth keeping

Biological anthropology, also called physical anthropology in older usage and in many department names, is the study of humans as biological organisms: our evolution, our variation, our biology as it interacts with our environments and our cultures, and the biology of our closest living and extinct relatives. Every clause in that sentence is doing work. Evolution gives the field its explanatory framework. Variation gives it its subject matter, since a science of humans that studied only an imagined average human would miss most of what is interesting. Environment and culture appear because human biology is never expressed in a vacuum. And relatives appear because you cannot understand what is distinctive about a species by looking at that species alone.

Notice what the definition does not say. It does not say the field studies bones, although many biological anthropologists do. It does not say the field studies fossils, although some do. Those are materials, not the subject. The subject is the organism and its history. A researcher measuring the resting metabolic rate of Tsimane forager-horticulturalists in Bolivia and a researcher describing a 3.5 million year old jaw are doing the same science with different evidence, and both are asking, in the end, what kind of animal we are and how we got that way.

Key idea: Biological anthropology studies humans and our relatives as evolving, varying biological organisms; bones, fossils, genes, and living bodies are its evidence, not its subject.

One field of four

In the North American tradition, biological anthropology is one of anthropology's four fields, alongside cultural anthropology (living societies and what their practices mean), archaeology (the human past through material remains), and linguistic anthropology (language as social life). That four-field structure was an argument before it was an organizational chart. It says that a human being is simultaneously an evolved primate, a member of a society, an inheritor of a material past, and a speaker, and that studying any one of those in isolation gives a distorted answer.

The overlap with archaeology is especially heavy, and the two share sites, field seasons, funding, and often the same trench. If you have taken an introduction to archaeology, you already have some of what this course needs: stratigraphy, context, and the dating methods you will meet again in Module 3, where this course deliberately declines to re-derive radiocarbon calibration and instead points you at how the same techniques get applied to hominin sites. Where archaeology asks what people made and did, biological anthropology asks what people were: what their bodies could do, what they ate, how they grew, how they were related, and how they changed across generations.

Key idea: The four-field structure is a claim that human beings must be studied as evolved organisms, social actors, makers of things, and speakers at once.

The subfields, and what each one actually does

Inside biological anthropology, the work divides into recognizable specialties. They share theory and constantly borrow each other's methods, but the daily labor looks very different.

Subfield	What it studies	Typical evidence
Paleoanthropology	The fossil record of humans and our extinct relatives	Fossil bone and teeth, sediments, associated fauna, dates
Primatology	Living non-human primates: anatomy, ecology, behavior, cognition	Field observation, fecal hormones and genetics, captive studies
Human biology and variation	How living human populations differ and adapt	Anthropometry, physiology, growth data, allele frequencies
Molecular anthropology	Evolutionary history read from DNA and proteins	Modern genomes, ancient DNA, protein sequences
Skeletal biology and bioarchaeology	Human remains from archaeological contexts	Cemetery populations, pathology, isotopes, trauma
Forensic anthropology	Human remains in medicolegal and human rights cases	Recent skeletons, scenes of recovery, case records
Human behavioral ecology	How ecology shapes reproduction, foraging, and cooperation	Time allocation, energetics, demography in living groups

An honest note about that table: the boundaries leak. The person who studies Neanderthal diet may use isotope methods developed by bioarchaeologists, a comparative baseline built by primatologists, and a chronology built by archaeologists. Increasingly the interesting work happens in the seams, which is one reason the field changes fast.

Key idea: The subfields are divisions of labor rather than separate sciences, and most current advances come from combining their methods on the same problem.

The biocultural approach

Here is the idea that most distinguishes biological anthropology from human biology as practiced in a medical school. Human biology and human culture are not two systems that occasionally touch. They are one system, and each continually reshapes the other. The field calls this the biocultural approach, and the classic demonstration is worth walking through slowly, because once you see it you will see it everywhere.

In West Africa, the sickle-cell allele of the beta-globin gene occurs at frequencies that would be lethal nonsense in most environments: people who inherit two copies develop sickle-cell disease. It persists because a single copy confers substantial protection against falciparum malaria. So far, that is ordinary biology. Now add the culture. In the 1950s the anthropologist Frank Livingstone argued that the high frequencies were not ancient but relatively recent, and that they tracked the spread of slash-and-burn yam agriculture. Clearing forest created standing water in sunlit pools, which suited the mosquito Anopheles gambiae, which raised malaria transmission, which raised the payoff to carrying one sickle allele. A farming decision changed a landscape, the landscape changed an insect population, the insect changed the selective environment, and the human gene pool moved. Culture was not a bystander to that evolution. Culture caused it.

The same logic runs through much of this course. Dairying made lactase persistence advantageous. Cooking, once invented, plausibly relaxed selection on chewing anatomy. Clothing and shelter allowed a tropical primate to live in the Arctic, which changed what selection acted on. Industrial food environments interact with metabolic physiology in ways that produce disease patterns no ancestral population experienced. If you take one habit of mind from this lesson, take this one: when you see a human biological pattern, ask what people were doing.

Key idea: Human biology and culture form a single feedback system, so cultural practices such as farming and dairying are causes of human evolution, not merely its background.

How the science actually works, and where it strains

Biological anthropology reasons like other historical and comparative sciences. You state a question, specify what evidence would count for and against a proposed answer, gather data under a design you can describe to a skeptic, and then let the data have their say. The comparative method does much of the heavy lifting: to know whether a human trait is unusual, you compare humans with other primates; to know whether a fossil trait is derived, you compare it with the condition in the group's ancestors and cousins.

Two structural strains deserve naming up front, because they explain most of the arguments you will read about in this course. The first is sample size. Some hominin species are known from a handful of specimens, occasionally from one. When you have a single skull, you cannot tell whether an odd feature marks a new species or is simply an individual quirk, a pathology, or the result of being crushed under sediment for two million years. Living species help calibrate this: measure enough gorillas and you learn how much variation a single primate species can contain, and the answer is usually more than beginners expect.

The second strain is bias in what survives. Fossilization requires unusual conditions, so the record over-represents environments where bones get buried fast and under-represents wet forests, which is exactly where a large part of ape and possibly early hominin evolution happened. Ancient DNA survives best in cold, dry places, which is why we have superb genomes from Siberian caves and almost nothing from equatorial Africa. When you read that a species first appears at a certain date, hold it lightly: first appearance in the record is a statement about discovery and preservation as much as about biology.

This is also why the field publicly changes its mind, and does so often. A new fossil, a new dating run, or a new genome regularly overturns textbook claims. A text course has its own limits. It cannot put a cast of a cranium in your hands, and handling casts is how most students finally understand what a sagittal crest or a supraorbital torus is. It cannot take you to a primate enclosure to watch a social group for two hours. Where a hands-on step matters, the activities in this course will send you toward the closest substitute available at a desk.

Key idea: Small samples and a biased preservation record are permanent features of this science, which is why careful practitioners report uncertainty and revise conclusions when new evidence arrives.

Why the field is worth your time

There is a practical argument and a deeper one. The practical argument is that biological anthropology answers questions people care about and get wrong constantly: what humans are adapted for, why populations differ, whether racial categories describe biology, what our diets were, why childbirth is hard, why some medicines work differently in different people. The deeper argument is the one Sherwood Washburn made when he pushed the field toward evolutionary questions in the mid twentieth century. Humans are the only species that asks what kind of animal it is. Biological anthropology is the discipline that tries to answer with evidence rather than with a story we would like to be true, and, as the next lesson shows in uncomfortable detail, it did not always succeed.

Common misconceptions

• Biological anthropology is the study of bones. Bones are one kind of evidence. The field also studies living people, genomes, hormones, growth, primate behavior, and energetics.
• It is the same thing as paleontology. Paleontology studies ancient life generally; biological anthropology studies humans, our ancestors, and our primate relatives, and includes living populations.
• Human evolution stopped when agriculture began. Several of the best-documented cases of selection in humans, including lactase persistence and high-altitude adaptation, occurred within the last ten thousand years.
• Because culture matters, biology does not. The biocultural point is the opposite: culture is an evolutionary force acting on human biology, which makes biology more relevant, not less.
• The fossil record is a straight line of ancestors. It is a scattered sample of a branching bush, most of whose branches left no descendants and many of whose members we have not found.

Recap

• Biological anthropology studies humans and our relatives as evolving, varying organisms, past and present.
• It is one of anthropology's four fields and works especially closely with archaeology.
• Its subfields include paleoanthropology, primatology, human biology, molecular anthropology, bioarchaeology, forensic anthropology, and human behavioral ecology.
• The biocultural approach treats culture as a cause of human biological change, as in the sickle-cell and yam-farming case.
• Small samples and preservation bias are permanent constraints, so honest practitioners report uncertainty and revise.

Sources

1. American Association of Biological Anthropologists. (n.d.). About AABA and the field. bioanth.org
2. Smithsonian National Museum of Natural History. (n.d.). What does it mean to be human? Human Origins Program. humanorigins.si.edu
3. Britannica. (2024). Physical anthropology. britannica.com
4. Wikipedia contributors. (2025). Biological anthropology. en.wikipedia.org

The big picture

In a storage room at the University of Pennsylvania Museum stand shelves of human crania, each with a number inked onto the bone. There are around a thousand of them. They were assembled in the 1830s and 1840s by a Philadelphia physician named Samuel George Morton, who wanted to measure the interior volume of skulls from around the world and use the results to rank human groups. Some of the skulls were taken from the graves of enslaved people in Cuba and the United States. Some were taken from Native American burials by collectors who did not ask. None of these people consented, and almost none of them have names in the ledger.

Morton's instruments worked. His conclusion was fixed before he lifted the first one. What follows traces what was done with those skulls, what it authorized in American law, who took it apart, and what remains unresolved, because the answer to each of those questions is why the discipline now teaches human variation the way it does.

Measuring skulls to rank people

Eighteenth and nineteenth century European naturalists began by sorting humans into varieties. Linnaeus, in the tenth edition of Systema Naturae in 1758, listed four geographic varieties of Homo sapiens and attached temperaments and characters to each, which is not taxonomy but stereotype in Latin. Johann Friedrich Blumenbach in the 1770s and 1790s proposed five varieties and coined the term Caucasian on aesthetic grounds after a skull from the Caucasus he found beautiful. Blumenbach, to his credit, insisted humans were a single species and that the varieties graded into one another. Others took the categories and dropped the gradation.

Morton industrialized the project. Starting in the 1830s, he filled cranial cavities with white mustard seed and later with lead shot, poured the filling into a graduated cylinder, and recorded the volume in cubic inches. He published Crania Americana in 1839 and Crania Aegyptiaca in 1844. He then arranged group averages into a ranking, with Europeans on top, and treated the ranking as a measure of innate capacity. Two moves happened here, and it matters that you can separate them. Measuring the internal volume of a skull is a legitimate, repeatable physical operation. Treating group average cranial volume as a measure of intellectual worth is not a measurement at all; it is an assumption smuggled in as a result.

Key idea: Nineteenth-century craniometry combined a real, repeatable measurement with an unexamined assumption that group averages in skull volume revealed innate mental worth, and the assumption did all the political work.

Polygenism, slavery, and the American School

Morton's numbers fed a live political argument. Most naturalists were monogenists, holding that all humans descend from a common origin. A vocal group, later called the American School, argued instead for polygenism: that human races were separately created species. Josiah Nott, an Alabama physician, and George Gliddon published Types of Mankind in 1854, dedicated to Morton's memory, and the volume was read in the United States as scientific support for slavery. Nott was explicit that his goal included defending the racial order of the South. Louis Agassiz, the most famous naturalist in America, endorsed polygenism after a personal encounter with Black servants in Philadelphia that he described in a letter to his mother in frankly revolted terms, and he commissioned daguerreotypes of enslaved people at a South Carolina plantation to document racial types.

Note what is happening methodologically. The scientific question and the political stake were entangled, the researchers were not disinterested, the data were collected under coercion, and the framework guaranteed the answer. Darwin, when he published in 1859, cut the ground out from under polygenism by giving a mechanism for a single origin with subsequent divergence, and by the 1870s the polygenist position had collapsed as biology. Its social conclusions simply migrated into evolutionary language and kept going.

The Gould dispute, and why it is a good lesson in method

In 1978 and again in his 1981 book The Mismeasure of Man, Stephen Jay Gould reanalyzed Morton's data and argued that Morton had unconsciously biased his results: mis-selecting specimens, making arithmetic errors that ran one direction, and packing the seed more or less firmly depending on expectation. The account became famous as a demonstration that bias can operate without fraud.

Then in 2011, a team led by Jason Lewis physically re-measured a large portion of the Morton collection and published in PLoS Biology. They found Morton's own measurements largely accurate and reproducible, and they argued that Gould's reanalysis contained errors of its own and that Gould had not, in fact, re-measured any skulls. Subsequent commentators, including Michael Weisberg and a group led by Jonathan Kaplan, pushed back on the pushback, noting that the 2011 team's sample and analysis also had problems and that Morton's group summaries were still shaped by which skulls he happened to obtain, which was not random at all.

Where does that leave you? With something more useful than a hero and a villain. Morton's individual measurements appear to have been mostly honest. His samples were unrepresentative, his groupings were arbitrary, his interpretations were racist, and, crucially, the entire research program rested on a premise that later evidence demolished: cranial capacity within the normal human range does not predict intelligence, and average brain volume tracks body size, sex, and climate far more than anything else. The dispute is worth learning because it shows a field arguing in public about its own history using measurements anyone can repeat. That is what correction looks like, including when it is messy.

Key idea: The Morton controversy shows that a study can have reproducible numbers and still be worthless, because sampling and interpretation, not just measurement, carry a research program's assumptions.

Eugenics: when the ranking became law

Francis Galton, Darwin's half-cousin, coined eugenics in 1883 for the project of improving human heredity by controlling who reproduces. In the United States the movement acquired institutions: Charles Davenport's Eugenics Record Office opened at Cold Spring Harbor in 1910 and collected family pedigrees purporting to show the heritability of pauperism, criminality, and feeblemindedness. State fairs ran Fitter Families contests. Anthropometry and intelligence testing, both dressed in the authority of measurement, supplied the evidence.

The consequences were legal and enormous. Indiana passed the first compulsory sterilization law in 1907, and about thirty states followed. In 1927 the United States Supreme Court upheld Virginia's law in Buck v. Bell, in an opinion by Oliver Wendell Holmes Jr. that ended with the notorious line, "Three generations of imbeciles are enough." Carrie Buck was neither feebleminded nor the child of a feebleminded mother; she had been raped and institutionalized to hide it. Roughly sixty thousand people were sterilized under such laws in the United States, disproportionately poor women, disabled people, and women of color, and the practice continued in some states into the 1970s. The Immigration Act of 1924 set national origin quotas built on arguments about the hereditary quality of southern and eastern Europeans, supported by testimony from eugenicists including Harry Laughlin.

Then it was exported. German race hygienists cited California's sterilization program as a working model, and the 1933 German sterilization law drew directly on it. Anthropologists were not bystanders. Eugen Fischer, whose 1913 study of mixed-ancestry families in German South West Africa was conducted in the shadow of the Herero and Nama genocide, later headed the Kaiser Wilhelm Institute for Anthropology; his colleague Otmar von Verschuer supervised Josef Mengele, who sent specimens from Auschwitz back to Berlin. Living people were exhibited as specimens too: Sarah Baartman in Europe, and Ota Benga at the Bronx Zoo in 1906.

Key idea: Eugenics was not a fringe movement but mainstream science policy, and its anthropometric and testing methods produced sterilization laws, immigration quotas, and a template that Nazi Germany explicitly borrowed.

The correction, and who made it

The first serious blow came from inside anthropology. Franz Boas, working for the Dillingham immigration commission, published a study in 1912 comparing the cephalic index of immigrants with that of their American-born children. The head shape changed measurably within a single generation. If the supposedly fixed racial marker was responsive to the environment in one generation, then racial types were not fixed biological essences. Boas spent decades attacking typological race thinking and trained the students who carried the argument forward. Honesty requires adding that Boas himself participated in the era's worst collecting practices, including arranging for the remains of Inuit people brought to New York to enter museum collections, in a case that left the boy Minik Wallace to discover his father's skeleton on display. The correction was not made by saints, and that is part of the point.

After the Second World War the argument moved fast. Ashley Montagu published Man's Most Dangerous Myth: The Fallacy of Race in 1942 and drafted the first UNESCO Statement on Race in 1950, which declared race a social myth rather than a biological fact and was revised in 1951 under pressure from biologists who wanted the biology stated more precisely. Sherwood Washburn's 1951 call for a new physical anthropology pushed the field from measuring and classifying toward asking evolutionary questions about process and function. In 1962 Frank Livingstone published an article whose title said it plainly, arguing that there are no races, only clines, and in 1972 Richard Lewontin published the apportionment analysis showing that the great majority of human genetic variation lies within populations rather than between the groups called races. You will work through that evidence in detail in Module 6.

Formal statements followed. The American Anthropological Association issued a Statement on Race in 1998. The American Association of Physical Anthropologists adopted a statement on the biological aspects of race in 1996 and replaced it in 2019 with a much broader Statement on Race and Racism that acknowledged the discipline's own complicity. In 2021 the association renamed itself the American Association of Biological Anthropologists, in part to shed the associations of the older label.

Key idea: The correction combined new evidence (plasticity studies, clines, genetic apportionment) with new theory (population thinking from the modern synthesis) and, eventually, institutional acknowledgment of the field's complicity.

What is still unfinished

Do not read this as a story that ended. Museums and universities still hold enormous numbers of human remains taken without consent. In the United States the Native American Graves Protection and Repatriation Act of 1990 created a legal process for return, and rules strengthened in 2024 required institutions to obtain descendant consent before displaying or researching covered remains, but compliance has been slow and thousands of individuals remain unrepatriated. Penn announced in 2021 that it would repatriate the crania of enslaved people in the Morton collection, and in the same year the university faced a separate scandal over the retention and classroom use of remains of a child killed in the 1985 police bombing of the MOVE house in Philadelphia.

Genomics raised the same questions in new form. Members of the Havasupai Tribe gave blood samples for diabetes research in the 1990s and later learned the samples had been used for studies of schizophrenia, inbreeding, and migration they had never agreed to; the case settled in 2010 with the return of samples and a payment. In 2017 the San of southern Africa issued their own code of research ethics after decades of being studied. The phrase for the underlying problem is helicopter science: researchers arriving, extracting data or samples, publishing elsewhere, and leaving nothing behind. Current best practice puts descendant and source communities in the design of the research, not at the end of it, and the field is still arguing about how to do that well.

Common misconceptions

• Scientific racism was a fringe view held by a few cranks. It was mainstream, institutionally funded, taught in universities, and written into immigration and sterilization law.
• Gould proved Morton faked his data. Re-measurement found Morton's numbers largely accurate; what failed was his sampling and his interpretive framework, and Gould's own analysis has been criticized in turn.
• Bigger brains mean higher intelligence. Within the normal human range, cranial capacity does not predict cognitive ability; it tracks body size, sex, and climate.
• Eugenics ended in 1945. Compulsory sterilizations continued in some United States jurisdictions into the 1970s, and coerced sterilization cases have surfaced far more recently.
• The discipline has fully cleaned this up. Repatriation is incomplete, consent practices are uneven, and the field openly describes this work as ongoing.

Recap

• Craniometry measured real things but rested on the unexamined premise that skull volume revealed innate worth.
• Polygenism and the American School supplied scientific cover for slavery; Darwinian common descent undercut it, though its conclusions persisted in new language.
• The Gould-Morton dispute shows that reproducible numbers cannot rescue a study whose sampling and interpretation are biased.
• Eugenics turned this science into sterilization laws, immigration quotas, and a model that German race hygienists adopted.
• Boas's plasticity study, Montagu and UNESCO, Washburn's new physical anthropology, Livingstone's clines, and Lewontin's apportionment drove the correction, formalized in AAA and AABA statements.
• Repatriation, consent, and equitable partnership remain unfinished business, not solved history.

Sources

1. American Association of Biological Anthropologists. (2019). AABA statement on race and racism. bioanth.org
2. Lewis, J. E., DeGusta, D., Meyer, M. R., Monge, J. M., Mann, A. E., & Holloway, R. L. (2011). The mismeasure of science: Stephen Jay Gould versus Samuel George Morton on skulls and bias. PLoS Biology, 9(6), e1001071. doi.org
3. Britannica. (2024). Eugenics. britannica.com
4. National Park Service. (n.d.). Native American Graves Protection and Repatriation Act. U.S. Department of the Interior. nps.gov
5. Wikipedia contributors. (2025). Scientific racism. en.wikipedia.org

The big picture

Daphne Major is a small volcanic cone in the Galapagos with almost nothing on it: cactus, some scrubby bushes, guano, and a population of medium ground finches. Starting in 1973, Peter and Rosemary Grant went there year after year and did something tedious and brilliant. They caught, banded, and measured nearly every finch on the island, generation after generation, recording beak length, beak depth, and body size to a fraction of a millimeter.

Then 1977 happened. Almost no rain fell. Small soft seeds ran out first. What was left were the large, hard, spiky fruits of a plant called Tribulus, which only birds with deep, strong beaks could crack. About eighty percent of the finches on the island died. The survivors were, on average, measurably larger-beaked than the population that entered the drought, and their offspring inherited that difference. Average beak depth in the population shifted by roughly four percent in a single generation. Then in 1983 an El Nino year drowned the island in rain, small soft seeds exploded in abundance, and selection ran the other way.

That is evolution measured in the field, on a named island, in banded birds, by counting. It is worth starting here because evolutionary theory can sound abstract until you see it as bookkeeping about who survived and who reproduced. This lesson gives you the theory as working biological anthropologists use it: the four forces that change populations, the synthesis that unified them, the species concepts that make fossil naming so contentious, and the independent lines of evidence that make the whole framework the most productive idea in biology.

What had to be in place first

Two ideas had to arrive before natural selection was even thinkable. The first was deep time. Charles Lyell's Principles of Geology, published in the early 1830s, argued that the ordinary processes you can watch today, erosion, deposition, and uplift, are sufficient to explain the Earth's features given enough time, and that the required time was immense. The second was extinction, established by Georges Cuvier around 1800 from fossil elephants that matched no living species. Together they gave a world that was old and in which species came and went.

Jean-Baptiste Lamarck deserves better than the punchline he usually gets. He proposed, in 1809, a real mechanism for change over time, at a moment when most naturalists held species to be fixed. His mechanism, the inheritance of characteristics acquired during an individual's life, turned out to be wrong for most traits, but his insistence that species transform was right and unpopular. And Thomas Malthus supplied the last piece in 1798 with a grim observation about populations: they reproduce faster than resources grow, so most offspring must die.

The logic of natural selection

Charles Darwin and Alfred Russel Wallace independently assembled these pieces into the same argument, presented jointly in 1858, with Darwin's On the Origin of Species following in 1859. Strip it to its bones and it is a short deduction from things anyone can check.

1. Organisms produce more offspring than can survive to reproduce.
2. Individuals within a population vary in their traits.
3. Some of that variation is heritable, passed from parent to offspring.
4. Therefore, the heritable variants that happen to help their bearers survive and reproduce in the current environment will become more common in the next generation.

That is the whole engine. Notice what it does not require: no goal, no plan, no striving. Notice also the three conditions that must all hold. If there is no variation, selection has nothing to act on. If the variation is not heritable, the advantage dies with the individual. If the trait does not affect survival or reproduction, nothing happens to its frequency. The Grants' finches satisfied all three, which is why the drought produced a measurable shift.

Fitness in this framework is not health, strength, or athleticism. It is relative reproductive success: how many offspring an individual leaves compared with others in the same population, counted in the same currency. An adaptation is a heritable trait that exists at its current frequency because it raised fitness in the past environment. That last clause matters. Adaptations are backward-looking. A trait shaped by a Pleistocene environment need not be helpful now, which is a point Module 6 will press hard when we get to evolutionary medicine.

Key idea: Natural selection follows deductively from overproduction, heritable variation, and differential reproduction; it requires no foresight and produces adaptations to past, not future, environments.

Selection has more than one shape

Directional selection shifts a trait's average in one direction, as beak depth did in 1977. Stabilizing selection kills off both extremes and keeps the average where it is, which is what usually happens to human birth weight: very small and very large newborns both historically faced higher mortality. Disruptive selection favors both extremes over the middle. Sexual selection acts on traits that improve mating success even when they hurt survival, which is why male peafowl carry absurd tails and why the degree of body size difference between male and female primates is such a useful clue about mating systems.

The most important variety for anthropology is balancing selection, which actively maintains variation instead of eliminating it. The textbook human case is the sickle-cell allele. People with two normal alleles are vulnerable to falciparum malaria. People with two sickle alleles develop sickle-cell disease. People with one of each are substantially protected against severe malaria and do not have the disease. In a malarial environment, the heterozygote is the fittest genotype, and the arithmetic of that guarantees that both alleles persist together at an equilibrium frequency, because heterozygote parents keep producing homozygotes of both kinds. Selection here does not purify the population. It keeps a costly allele in circulation because the alternative is worse.

Key idea: Selection can push a trait, hold it steady, split it, or actively preserve two alleles at once, as heterozygote advantage does for sickle cell in malarial regions.

The other three forces

Mutation is the ultimate source of all genetic variation, and everything else merely shuffles what mutation supplies. Human germline point mutations occur at roughly one per hundred million base pairs per generation, which works out to somewhere around 50 to 70 new mutations in each newborn. Most are neutral, falling in regions where they change nothing. Some are harmful. A very small minority are beneficial in a given environment. Mutation includes not only single base substitutions but insertions, deletions, duplications of whole genes, and rearrangements of chromosomes, and gene duplication in particular is a major route to evolutionary novelty because a spare copy is free to change.

Genetic drift is change in allele frequencies from random sampling alone. Imagine a population where a neutral allele is at fifty percent. The individuals who happen to reproduce are a sample of the population, and samples wobble. In a group of ten thousand, the wobble is negligible. In a group of twenty, an allele can vanish or reach fixation in a few generations for no reason but luck. Two special cases matter constantly in human history. A founder effect occurs when a small group establishes a new population and carries an unrepresentative slice of the parent gene pool, which is why certain alleles are common in Ashkenazi Jewish, Afrikaner, French Canadian, and Finnish populations. A bottleneck occurs when a population crashes and recovers from few survivors, permanently reducing diversity. Human populations show drift signatures everywhere, and effective population size, the size of an idealized population that would drift at the observed rate, has been estimated for our long-term ancestry at only around ten thousand. That number will matter enormously in Module 6.

Gene flow is the movement of alleles between populations through migration and mating. It makes populations more similar and works directly against drift and against local selection. For humans it is the single most underrated force. Our species has never stopped moving and interbreeding, at every scale from the neighboring valley to the transcontinental, which is a large part of why human populations are so genetically similar and why sharp biological boundaries between them do not exist.

Key idea: Mutation creates variation, drift removes it randomly and fastest in small populations, gene flow spreads it between populations, and selection sorts it non-randomly; all four operate at once.

The modern synthesis, and why anthropology cared

Darwin had no working theory of inheritance, and after Mendel's work was rediscovered in 1900 there was a genuine standoff. Biometricians studying continuous traits like height thought Mendelian genetics with its discrete factors could not explain smooth variation, while Mendelians thought the biometricians were measuring noise. The resolution came from mathematics. Between roughly 1918 and 1932, R. A. Fisher, J. B. S. Haldane, and Sewall Wright showed that many genes of small effect, each behaving in a perfectly Mendelian way, produce exactly the continuous variation the biometricians measured, and they built the algebra of how selection, drift, mutation, and gene flow change allele frequencies. That field is population genetics, and it is the technical core of Lesson 4.

The modern synthesis of the late 1930s and 1940s extended that mathematics across biology: Theodosius Dobzhansky in genetics, Ernst Mayr in systematics and speciation, George Gaylord Simpson in paleontology, G. Ledyard Stebbins in botany. Its central conceptual move was population thinking: a species is not an ideal type with imperfect individual copies, but a population of varying individuals described by frequencies. Reread that sentence with the previous lesson in mind. Typological thinking is precisely what racial science ran on. When Sherwood Washburn called in 1951 for a new physical anthropology, he was demanding that the field stop classifying types and start studying processes and populations, and that demand was the modern synthesis arriving in anthropology.

Key idea: The modern synthesis unified genetics, systematics, and paleontology around population thinking, which replaced the typological reasoning that had underwritten racial classification.

What is a species, and why fossil names are fought over

Mayr's biological species concept defines species as groups of actually or potentially interbreeding natural populations that are reproductively isolated from other such groups. It is useful, and it is untestable on fossils, because you cannot breed a jawbone. Paleoanthropologists therefore work with morphospecies, groups defined by shared anatomy and separated by more difference than you would expect within one living species. That standard is a judgment call, and it is why the field contains lumpers, who see one variable species where splitters see three.

Two complications make it worse. First, sexual dimorphism inflates apparent variation: if you did not know gorillas were one species, the difference between a large male and a small female skull might convince you they were two. Second, the discovery of Neanderthal and Denisovan interbreeding with our own ancestors showed that hominin lineages that were distinct enough to be named separately were not reproductively isolated. Under a strict biological species concept, that is awkward. Most researchers now treat the boundary as genuinely fuzzy, and you should read every hominin species name in this course as a hypothesis under active argument rather than a fact.

Why the evidence is overwhelming

Evolution is supported by many independent lines that converge on the same answer, which is the strongest kind of scientific case, because no single flaw can bring the whole structure down.

Line of evidence	What it shows	Example
Fossils	Sequences of forms with intermediate anatomy, in the right rock ages	Bipedal apes with small brains preceding large-brained humans
Comparative anatomy	Homologous structures modified from a shared plan	The same bone sequence in a human arm, a bat wing, and a whale flipper
Vestigial and imperfect structures	Design constraints inherited from ancestors	The recurrent laryngeal nerve's long detour; the human coccyx
Biogeography	Species distributions match geological and dispersal history	Lemurs only on Madagascar; marsupial radiation in Australia
Molecular data	Similarity tracks relatedness across independent genes	Human chromosome 2 shows the fusion of two ape chromosomes, telomere remnants included
Shared errors	Broken genes and viral insertions in identical positions	The disabled vitamin C gene shared by humans and other primates
Observation	Evolution measured in real time	Grant finches; antibiotic resistance; lactase persistence in the last ten thousand years

One point about vocabulary, because it causes real confusion. In everyday speech a theory is a hunch. In science a theory is the opposite: a well-substantiated explanatory framework that organizes a large body of tested observations and generates new predictions. Germ theory, atomic theory, and plate tectonics are theories in the same sense. Calling evolution a theory ranks it with the most secure ideas in science, not the least.

A word on the obvious question. This course teaches evolution because it is the framework that explains and predicts the biological evidence, and because nothing in genetics, medicine, or anthropology makes coherent sense without it. It takes no position on theology. Theodosius Dobzhansky, who wrote the sentence about nothing in biology making sense except in the light of evolution and who did as much as anyone to build the modern synthesis, was a practicing Russian Orthodox Christian who wrote about his faith at length. So is Francis Collins, who led the Human Genome Project. Where you land is yours to work out. What follows is the evidence and the reasoning.

Common misconceptions

• Evolution means progress toward better organisms. Selection tracks local, current environments; a change that helps in a drought may hurt in a wet year, as Daphne Major showed within a decade.
• Individuals evolve. Individuals develop and die with the genes they were born with; populations evolve, because it is allele frequencies that change across generations.
• Survival of the fittest means the strongest survive. Fitness is relative reproductive success, and it is often achieved by cooperating, hiding, or reproducing early rather than by strength.
• Humans evolved from chimpanzees. Humans and chimpanzees share a common ancestor that was neither; chimpanzees have been evolving for exactly as long as we have.
• Mutations are always harmful. Most are neutral, some are harmful, and a small fraction are beneficial; without them there would be no variation for selection to act on.
• Evolution is just a theory. In science a theory is a tested explanatory framework, not a guess, and evolution is supported by many independent converging lines of evidence.

Recap

• Natural selection follows from overproduction, heritable variation, and differential reproduction, and it requires no foresight.
• Selection comes in directional, stabilizing, disruptive, sexual, and balancing forms; sickle cell is the classic human case of heterozygote advantage.
• Mutation supplies variation, drift removes it by chance in small populations, gene flow shares it between populations, and all four forces act together.
• The modern synthesis unified biology around population thinking, which anthropology adopted through Washburn's new physical anthropology.
• Species concepts are workable for living organisms and contested for fossils, which is why hominin names are hypotheses.
• Fossils, anatomy, biogeography, molecules, shared errors, and direct observation independently converge on evolution, which is what a scientific theory means.

Sources

1. Britannica. (2024). Evolution. britannica.com
2. National Human Genome Research Institute. (n.d.). Talking glossary of genomic and genetic terms. National Institutes of Health. genome.gov
3. Smithsonian National Museum of Natural History. (n.d.). Human evolution evidence. Human Origins Program. humanorigins.si.edu
4. Britannica. (2024). Natural selection. britannica.com
5. Wikipedia contributors. (2025). Modern synthesis (20th century). en.wikipedia.org

The big picture

In 1908 the Cambridge geneticist Reginald Punnett was stuck. Someone at a meeting had objected that brachydactyly, a dominant condition producing short fingers, ought to spread until three people in four had it, purely because it was dominant. Punnett was sure this was wrong and could not show why, so he put the problem to his cricket partner, the mathematician G. H. Hardy. Hardy dispatched it in a short letter to Science that year and apologized for bothering the readership with what he called "mathematics of the multiplication-table type." A Stuttgart physician named Wilhelm Weinberg had presented the same result a few months earlier and went uncredited in English for decades.

Their answer was that dominance has nothing to do with frequency. In a population where nothing evolutionary is happening, allele frequencies stay exactly where they are, and genotype frequencies settle into a fixed pattern after one generation of random mating. That result is the null model against which every claim in this course about a changing human population gets measured, and you will work it three times with real numbers before this lesson ends. First, the genetics it rests on.

The physical inheritance

Your genome is about 3.1 billion base pairs of DNA, packaged into 23 pairs of chromosomes: 22 pairs of autosomes plus one pair of sex chromosomes. Chimpanzees, gorillas, and orangutans have 24 pairs, and the difference is not a problem for common ancestry but evidence for it, because human chromosome 2 carries the signature of a fusion of two ancestral ape chromosomes, telomere sequences stranded in its interior and a second, inactivated centromere.

Only about one to two percent of the genome codes for protein, in something on the order of twenty thousand genes. Much of the rest regulates when and where those genes are switched on, and a good deal of the interesting difference between species turns out to live in regulation rather than in protein sequence. A gene is a stretch of DNA with a function; an allele is one of the alternative versions of that stretch found in a population. A locus is the address on the chromosome. You inherit one allele at each autosomal locus from each parent, so you are homozygous at that locus if the two match and heterozygous if they do not.

One process deserves special attention because Module 5 depends on it. During meiosis, chromosomes exchange segments in recombination, and every generation of recombination chops inherited stretches into shorter pieces. That means the length of a shared DNA segment is a clock. Long unbroken stretches of Neanderthal-like sequence in a 45,000 year old Siberian genome told researchers that the interbreeding had happened only a few thousand years earlier. Short scattered fragments in living people tell you it was long ago. Recombination turns ancestry into something datable.

Key idea: Genes come in alternative alleles at fixed loci, and recombination shortens inherited segments each generation, so segment length dates when ancestry entered a lineage.

Mendel with a real human example

Gregor Mendel's two rules still hold. Segregation says the two alleles at a locus separate into different gametes, so each parent passes on one at random. Independent assortment says alleles at loci on different chromosomes are inherited independently of one another. Textbooks usually illustrate this with pea color; let us use the ABO blood group instead, because it is a genuine human polymorphism that shows several ideas at once.

The ABO locus has three common alleles: A, B, and O. A and B each produce a different sugar antigen on the red blood cell surface; O produces neither, because the O allele carries a deletion that disables the enzyme. A and B are codominant, which means a person with one of each expresses both. O is recessive to both. Three alleles yield six genotypes and four phenotypes.

Genotype	Blood type	Note
AA or AO	Type A	Two genotypes, one phenotype: dominance hides the O
BB or BO	Type B	Same pattern on the other side
AB	Type AB	Codominance: both antigens expressed
OO	Type O	Recessive phenotype requires two copies

Two lessons hide in that table. First, dominance is a relationship between an allele and a phenotype, not a statement that one allele is stronger, better, or more common. Type O is the recessive phenotype and it is also the most common blood type on Earth. Second, you cannot always read genotype from phenotype: type A people come in two genetic flavors, and this is exactly why counting phenotypes and counting alleles are different exercises.

Many traits anthropologists care about are not single-locus at all. Height, skin pigmentation, and body proportions are polygenic, influenced by many loci of small effect plus substantial environmental input. Some loci are pleiotropic, affecting several unrelated traits at once, which constrains what selection can do without side effects. Keep those two words handy; they will do work in Module 6.

Two special pieces of DNA, and their limits

Mitochondrial DNA is a small circular genome of about 16,569 base pairs inside the mitochondria. It is inherited only from the mother, it does not recombine, it mutates fast, and there are hundreds to thousands of copies in every cell. That last fact makes it precious in ancient DNA work, where nuclear DNA may be gone but mitochondrial fragments survive. The non-recombining portion of the Y chromosome gives the same kind of clean, single-line record on the father's side.

Now the caveat, which is one of the most misunderstood points in all of human genetics. A single non-recombining locus gives you the history of that locus, which is a gene tree, and a gene tree is not a population tree. When the 1987 study by Rebecca Cann, Mark Stoneking, and Allan Wilson traced human mitochondrial lineages to a common ancestor in Africa, headlines produced Mitochondrial Eve, a lone founding mother. That is not what the result means. It means that among the mitochondrial lineages carried by living people, all trace back to one woman who lived among a population of many thousands of contemporaries, all of whom may well be your ancestors through other lines. Every gene in your genome has its own coalescent history and its own most recent common ancestor, at a different time and often on a different continent. Read single-locus results as one thread, never as the fabric.

Key idea: Mitochondrial DNA and the Y chromosome give clean uniparental lineages, but a single locus yields a gene tree, not a population history, and Mitochondrial Eve was never the only woman alive.

Counting alleles: the first calculation

Populations evolve, individuals do not, so the quantity that matters is the frequency of each allele in a gene pool. Counting it is arithmetic. Take the MN blood group, a simple two-allele codominant system where every genotype is directly visible as a phenotype. Suppose you type 200 people and find 98 MM, 84 MN, and 18 NN.

Each person carries two alleles, so there are 400 alleles in the sample. Count the M alleles: each MM person contributes two and each MN person contributes one, giving 2 times 98 plus 84, which is 280. Count the N alleles: 2 times 18 plus 84, which is 120. Check: 280 plus 120 equals 400. Now convert to frequencies. The frequency of M, which we call p, is 280 divided by 400, or 0.70. The frequency of N, which we call q, is 120 divided by 400, or 0.30. And p plus q equals 1, as it must when there are only two alleles.

Hardy-Weinberg: the null model

In 1908 G. H. Hardy and Wilhelm Weinberg independently showed that in a population where nothing evolutionary is happening, allele frequencies stay the same and genotype frequencies settle into a fixed relationship after one generation of random mating. With two alleles at frequencies p and q, the expected genotype frequencies are p squared for one homozygote, 2pq for the heterozygote, and q squared for the other homozygote, and those three add to 1.

The Hardy-Weinberg equilibrium holds only if five conditions are met: no selection, no mutation, no gene flow, random mating with respect to the locus, and a population large enough that drift is negligible. No real population satisfies all five. That is not a flaw. It is the whole point. Hardy-Weinberg is a null model, the genetic equivalent of asking what a coin would do if it were fair, so that when the data depart from expectation you know something is going on and can go looking for it.

Worked example one. Return to the MN data, where p equals 0.70 and q equals 0.30. Expected frequencies are p squared equals 0.49, 2pq equals 2 times 0.70 times 0.30 equals 0.42, and q squared equals 0.09. Multiply each by the 200 people sampled: 98 MM, 84 MN, 18 NN. Compare with what you actually observed: 98, 84, 18. The fit is exact, so this population is consistent with equilibrium at this locus, and there is no evidence here that any evolutionary force is acting on it.

Worked example two, the useful one. Often you can only see the recessive phenotype, and you want to know how many carriers are hidden. Cystic fibrosis affects roughly 1 in 2,500 newborns in populations of northern European ancestry. Only people with two copies are affected, so q squared equals 1 divided by 2,500, which is 0.0004. Take the square root: q equals 0.02. Then p equals 1 minus 0.02, or 0.98. The carrier frequency is 2pq, which is 2 times 0.98 times 0.02, or 0.0392. That is about 1 in 25 people. Sit with that result. A condition that appears in one newborn in 2,500 is carried, invisibly, by roughly four percent of the population. This single calculation is why genetic counseling exists, and it is the standard way public health estimates carrier burden for recessive conditions.

Worked example three, the anthropological one. Now consider a hemoglobin locus in a malarial region of West Africa, with the normal allele A at p equals 0.80 and the sickle allele S at q equals 0.20. Among newborns, Hardy-Weinberg predicts, per 1,000 births: 640 AA, 320 AS, and 40 SS. Now suppose you survey 1,000 surviving adults in the same population and find 590 AA, 400 AS, and 10 SS.

Genotype	Expected per 1,000	Observed adults	Interpretation
AA	640	590	Deficit: many died of malaria
AS	320	400	Excess: protected heterozygotes survive best
SS	40	10	Large deficit: most died of sickle-cell disease

The observed adults are nowhere near the expectation, and the pattern is not random noise. Both homozygotes are under-represented and the heterozygote is over-represented, which is the exact fingerprint of balancing selection through heterozygote advantage. Notice how the reasoning ran: build the null, compare, and let the size and shape of the departure point to the responsible force. A deficit of heterozygotes instead would have suggested inbreeding or population subdivision; an excess of one homozygote might suggest directional selection or gene flow from a population with a different frequency.

Key idea: Hardy-Weinberg is a null model whose violation is the finding: the direction and shape of the departure from expectation identify which evolutionary force is acting.

Scale, and a preview

A few numbers to carry forward. Any two unrelated humans are about 99.9 percent identical in DNA sequence, differing at roughly three million single-nucleotide positions out of three billion. Humans and chimpanzees are about 98.8 percent identical at aligned single-nucleotide sites, closer to 96 percent when insertions and deletions are counted. And when Richard Lewontin partitioned human genetic variance in 1972, he found that about 85 percent of it lies among individuals within any given population, with only a small remainder distinguishing the continental groups people call races. Genome-scale work has repeatedly confirmed that apportionment. Module 6 will take that result apart carefully, including the serious statistical objection raised against it, because it is the single most important quantitative fact in the race debate and it deserves more than a slogan.

A last word on epigenetics, since it is often oversold. Chemical marks such as DNA methylation and histone modification change how genes are expressed without changing sequence, and they are central to normal development and to how environments get under the skin within a lifetime. Whether such marks are reliably transmitted across human generations is a genuinely open question: the Dutch Hunger Winter cohort and the Overkalix records are suggestive, but they are confounded and the molecular mechanism in humans is not established. Epigenetics is not a revival of Lamarck, and a course that told you otherwise would be selling you something.

Common misconceptions

• Dominant alleles are more common or stronger. Dominance describes how a phenotype appears in heterozygotes; type O blood is recessive and is the world's most common type.
• Mitochondrial Eve was the only woman alive at the time. She is the most recent common ancestor of living mitochondrial lineages, one woman among thousands of contemporaries who are also our ancestors.
• Hardy-Weinberg describes real populations. It describes a population in which nothing is happening; its value lies entirely in measuring how far reality departs from it.
• A 99.9 percent identical genome means humans are essentially uniform. That remaining fraction is roughly three million differences per pair of people, more than enough for enormous variation.
• Epigenetics means acquired traits are inherited. Most marks are reset between generations, and transgenerational transmission in humans remains unproven.

Recap

• Alleles are alternative versions of a locus; recombination shortens inherited segments each generation, making segment length a clock for admixture.
• ABO shows codominance, recessiveness, and the gap between phenotype and genotype in one real human system.
• Mitochondrial DNA and the Y chromosome give clean single lineages but only gene trees, not population histories.
• Allele frequencies are counted directly from genotype counts; p plus q equals 1 for a two-allele locus.
• Hardy-Weinberg predicts p squared, 2pq, and q squared under five idealized conditions, and departures identify the force at work.
• The carrier calculation from an affected frequency of 1 in 2,500 gives roughly 1 carrier in 25, which is why the null model has practical clinical value.

Sources

1. National Human Genome Research Institute. (n.d.). Talking glossary of genomic and genetic terms. National Institutes of Health. genome.gov
2. MedlinePlus Genetics. (n.d.). Understanding genetics. U.S. National Library of Medicine. medlineplus.gov
3. Cann, R. L., Stoneking, M., & Wilson, A. C. (1987). Mitochondrial DNA and human evolution. Nature, 325, 31-36. doi.org
4. Wikipedia contributors. (2025). Hardy-Weinberg principle. en.wikipedia.org
5. National Human Genome Research Institute. (n.d.). Genomic variation fact sheet. National Institutes of Health. genome.gov

The big picture

It is nine at night in a dry forest in western Madagascar and you are walking a narrow trail with a headlamp strapped to your forehead, sweeping the beam slowly across the canopy at head height. Then you catch it: two small points of light, orange and steady, staring back at you from four meters up. The eyeshine belongs to a grey mouse lemur, an animal that weighs about sixty grams and could sit in your cupped palm. When you creep closer it does not flee; it swivels its head, watches you with enormous forward-facing eyes, then reaches out and closes a hand around a branch with a genuine opposable thumb and nails, not claws, on its fingers.

Everything you need to define an order of mammals is in that thirty seconds. The eyeshine comes from a tapetum lucidum, a reflective layer behind the retina that most nocturnal mammals have and that monkeys and apes have lost. The forward-facing eyes give overlapping visual fields and depth perception. The grasping hand with flat nails is the primate hallmark. And the animal watching you rather than bolting is doing something that the largest brain-to-body ratio in its size class allows.

This lesson introduces the order we belong to. You will learn what a primate is, why the definition is a list rather than a single feature, how the order divides into groups, and how ecology shapes the way primates live together. Everything here is groundwork for the next lesson, where behavior and culture take the stage, and for Module 4, where you will read fossils by comparing them with the living animals you meet here.

A definition made of tendencies

Ask what makes a mammal a bat and the answer is easy: powered flight. Ask what makes a mammal a primate and there is no single answer, because primates are anatomically generalized. What they have instead is a suite of tendencies, most of which relate to life in trees, to seeing rather than smelling, and to investing heavily in few offspring. The British anatomist Wilfrid Le Gros Clark assembled the standard list in the mid twentieth century, and it still works.

• Grasping extremities. Opposable thumbs and, in all primates except humans, opposable big toes; flat nails instead of claws on most digits; sensitive tactile pads with dermal ridges.
• Generalized limbs. A retained collarbone, separate radius and ulna allowing forearm rotation, and joints capable of many movements rather than one specialized gait.
• Vision over smell. Forward-facing eyes with overlapping fields giving stereoscopic depth perception, a bony postorbital bar or full postorbital closure protecting the eye, color vision in most diurnal species, and a reduced snout with reduced olfactory apparatus.
• Large brains for body size. Especially expanded visual and association areas.
• Generalized dentition. Four tooth types retained, with fewer teeth than the ancestral mammal condition, and molars suited to varied diets rather than to one food.
• Slow life histories. Long gestation, usually single births, prolonged infant dependence, late maturity, and long lifespans relative to body size.
• Sociality. Most primates live in stable social groups and much of their behavior is directed at other members of those groups.

Why did this package evolve? Three hypotheses compete, and the honest answer is that they are not exclusive. The arboreal hypothesis holds that the traits are adaptations to moving in a three-dimensional world of branches, where grasping hands and depth perception prevent fatal mistakes. Matt Cartmill's visual predation hypothesis points out that squirrels manage trees perfectly well with claws and lateral eyes, and notes that forward-facing eyes and grasping hands are also found in cats and owls and in insect-hunting marsupials, suggesting the traits evolved for stalking insects at close range on fine terminal branches. Robert Sussman's angiosperm coevolution hypothesis ties the origin of primates to the radiation of flowering plants and their fruits and flowers, which put small foods at the end of small branches. Most researchers now expect the true answer combines predation on insects among terminal branches of angiosperms in a nocturnal ancestor.

Key idea: Primates are defined by a suite of tendencies rather than one trait, and the leading explanations for that suite, arboreal life, visual predation, and angiosperm coevolution, are complementary rather than mutually exclusive.

The nested groups

Classification is a hypothesis about relationships, and modern primate taxonomy is built to reflect descent. The order divides first into two suborders.

Group	Members	Distinguishing features
Strepsirrhini	Lemurs of Madagascar; lorises, pottos, and galagos of Africa and Asia	Moist rhinarium, tapetum lucidum, dental toothcomb, grooming claw, stronger reliance on smell
Haplorhini: Tarsiiformes	Tarsiers of island Southeast Asia	Enormous immobile eyes, no tapetum, huge hind limbs, entirely faunivorous
Haplorhini: Platyrrhini	New World monkeys of Central and South America	Sideways-facing nostrils, three premolars, prehensile tail in some, entirely arboreal
Haplorhini: Cercopithecoidea	Old World monkeys of Africa and Asia	Downward nostrils, two premolars, ischial callosities, tails never prehensile
Haplorhini: Hominoidea	Gibbons and siamangs; orangutans, gorillas, chimpanzees, bonobos, humans	No tail, broad shallow chest, mobile shoulder, Y-5 molar pattern, larger brains

A dental formula gives the number of incisors, canines, premolars, and molars in one quadrant of the mouth, upper over lower or simply as a sequence. Old World monkeys, apes, and humans share the catarrhine formula 2.1.2.3, which is thirty-two teeth in total. Most New World monkeys have 2.1.3.3, keeping a third premolar, which gives thirty-six teeth. Marmosets and tamarins are 2.1.3.2. This unglamorous detail is genuinely useful: a fragment of jaw with three premolars is not a catarrhine, and that single observation can place a fossil on a continent.

Within the Old World monkeys, two subfamilies split by diet. Cercopithecines, including baboons, macaques, and guenons, have cheek pouches and eat a wide omnivorous range. Colobines, including colobus monkeys, langurs, and the proboscis monkey, have complex sacculated stomachs with bacterial fermentation chambers that let them digest mature leaves and even seeds full of toxins. That is convergent, in miniature, with what a cow does.

Key idea: Primate classification is nested and reflects descent, and small anatomical details such as premolar count or stomach structure carry real information about relationships and diet.

Where humans sit

Here is a sentence students often resist and should not: humans are apes. Not descended from apes at a distance, not ape-like. We are hominoids by every anatomical and molecular criterion, nested inside the great apes, and inside that group closer to chimpanzees and bonobos than either is to gorillas. Genomic comparisons put the human lineage and the chimpanzee and bonobo lineage as sisters, with gorillas branching earlier, then orangutans, then gibbons.

Classification is nested rather than sequential, which is the point people miss. Saying humans are apes is like saying a robin is a bird and a bird is a dinosaur: it is a statement about ancestry, not about resemblance. It also means that phrases like humans evolved from monkeys are wrong twice over, since our ancestors were not any living monkey and we never left the catarrhine group we belong to. If a taxonomist insisted on strict consistency, the group containing all catarrhines would include us, which is precisely why arguments about whether humans are monkeys are arguments about naming conventions and not about biology.

Ecology: what a primate eats determines almost everything

Primate diets fall into broad categories: frugivores concentrate on fruit, folivores on leaves, insectivores or faunivores on insects and small animals, gummivores on tree exudates, and many species are omnivores drawing on all of these seasonally. The single most useful principle linking diet to body size is the Jarman-Bell relationship. Small animals have high metabolic demands per unit of body mass and small guts, so they must eat concentrated, easily digested foods such as insects and gum. Large animals have lower relative demands and large guts with long retention times, so they can subsist on abundant, low-quality bulk such as mature leaves. This is why there are no leaf-eating mouse lemurs and no insect-specialist gorillas, and why body size is one of the first things a primatologist wants to know.

Diet in turn shapes ranging. Fruit is patchy, seasonal, and defensible; leaves are abundant and everywhere. Frugivores therefore travel farther, hold larger home ranges, and are more likely to compete directly over resource patches. Folivores can afford smaller ranges and shorter days. Add activity pattern (diurnal, nocturnal, or cathemeral, meaning active in irregular bouts across the whole day) and locomotor mode (vertical clinging and leaping, quadrupedalism on branches or on the ground, brachiation, knuckle-walking, or bipedalism), and you have most of what determines a species' daily life.

Key idea: Body size sets what quality of food a primate must eat, food distribution sets how far it must travel, and travel and competition set the conditions under which social groups form.

Social organization and the socioecological model

Primates are unusual among mammals in how thoroughly social they are. The forms that sociality takes are limited and recognizable.

Organization	Composition	Examples
Solitary or dispersed	Individuals forage alone; ranges overlap; contact by scent and call	Orangutans, many galagos and lorises
Pair-bonded	One adult male, one adult female, and dependent offspring	Gibbons, titi monkeys, indri
Polyandrous or cooperative	One breeding female with multiple males helping to rear twins	Marmosets and tamarins
One-male, multi-female	A single breeding male with several females; other males in bachelor groups	Gorillas, hamadryas and gelada units, many colobines
Multi-male, multi-female	Many adults of both sexes with dominance hierarchies	Savanna baboons, macaques, capuchins
Fission-fusion	A large community that splits and reforms in changing subgroups	Chimpanzees, bonobos, spider monkeys

The socioecological model explains this variety with a two-step argument that is worth memorizing because it generates testable predictions. Female primates distribute themselves according to the distribution of food and safety, because female reproductive success is limited mainly by energy. Males then distribute themselves according to the distribution of females, because male reproductive success is limited mainly by access to mates. Add two more pressures, predation risk, which favors grouping, and infanticide risk from incoming males, which favors female association with protective males, and you can predict a good deal about which social system a species will have.

One correlation matters for the fossil record. Sexual dimorphism, the size difference between males and females, tends to be low in pair-bonded species like gibbons and high in species with intense male competition over groups of females, such as gorillas and baboons. That relationship lets paleoanthropologists reason from the size range of a fossil sample toward a mating system. It is also, as Module 4 will show, exactly the inference that has generated some of the field's longest-running arguments, because you can only estimate dimorphism if you already know which specimens belong to one species.

How many primates are there, and does the number mean anything?

Depending on whose taxonomy you consult, there are somewhere between about 300 and more than 500 living primate species. That is a startlingly wide range for a well-studied order, and it is not because new animals keep being discovered, though some are. It is because the species concept applied has shifted. Under a phylogenetic species concept, any diagnosably distinct population can be named, and several genera have been split extensively, particularly the lemurs. Critics call this taxonomic inflation and note that conservation funding follows species counts. Defenders answer that the older lumped categories hid real, independently evolving populations. This is the same argument you met over fossil names in Lesson 3, playing out among living animals where you can at least go and look.

Common misconceptions

• Humans are not apes, just related to them. Humans are hominoids nested inside the great apes; the classification reflects ancestry, not resemblance.
• All primates live in trees. Many, including baboons, geladas, patas monkeys, and humans, are substantially or entirely terrestrial.
• Monkeys are a step on the way to apes. Living monkeys are not ancestral to apes; both descend from a shared catarrhine ancestor and have been evolving separately since.
• Prehensile tails are a general monkey feature. Only some New World monkeys have them; no Old World monkey or ape does, and apes have no tail at all.
• Bigger primate groups always mean more advanced social life. Group size tracks ecology, predation, and food distribution, not sophistication.

Recap

• Primates are defined by a suite of traits: grasping hands, nails, generalized limbs, stereoscopic vision, reduced smell, large brains, slow life histories, and sociality.
• The arboreal, visual predation, and angiosperm coevolution hypotheses each explain part of that suite and are probably all partly right.
• The order splits into strepsirrhines and haplorhines, and within haplorhines into tarsiers, platyrrhines, cercopithecoids, and hominoids.
• Dental formulas, nostril orientation, and stomach structure carry real taxonomic and dietary information.
• Body size determines dietary quality, food distribution determines ranging, and both feed into the socioecological model of group formation.
• Sexual dimorphism correlates with mating system, which is why fossil size ranges are so heavily argued about.

Sources

1. Britannica. (2024). Primate. britannica.com
2. Duke Lemur Center. (n.d.). About lemurs. Duke University. lemur.duke.edu
3. Smithsonian National Museum of Natural History. (n.d.). Human family tree and primate relatives. Human Origins Program. humanorigins.si.edu
4. Wikipedia contributors. (2025). Primate. en.wikipedia.org

The big picture

In the Tai Forest of Cote d'Ivoire you hear it before you see anything: a hard, flat crack echoing through the canopy, then a pause, then another. It sounds like someone splitting kindling. When you finally get close, what you find is a chimpanzee sitting beside an exposed tree root, holding a stone in one hand and placing a panda nut into a worn depression in the root with the other. The nut takes several precise blows to open, and too much force ruins the kernel. Nearby, a four-year-old is failing at the same task with obvious frustration. She has been watching adults do this since infancy, and she will not be competent for several more years.

That scene contains almost every theme of this lesson: a difficult technique, a local tradition that neighboring chimpanzee communities do not share, years of social learning, and a set of hard questions about what any of it tells us about ourselves. This lesson covers how primate behavior is studied, what primates can do cognitively, what chimpanzee culture is and is not, why using living apes as stand-ins for our ancestors is a methodological trap, and why most of the animals in this lesson are now in serious trouble.

How you actually study a wild primate

The discipline changed on a specific day in 1960, when Jane Goodall watched a chimpanzee she called David Greybeard strip leaves from a twig, push it into a termite mound, and pull out termites to eat. Toolmaking had been treated as the boundary of humanity. Louis Leakey's response, when she cabled him, was that anthropology would now have to redefine tools, redefine humans, or admit chimpanzees to the club.

Goodall's method also mattered. Getting useful data requires habituation: months to years of a researcher being present until the animals treat them as a neutral part of the forest rather than as a predator. Once habituated, primatologists use structured sampling rather than writing down whatever catches the eye. Jeanne Altmann's 1974 paper on observational methods is the field's turning point here. In focal animal sampling you pick one individual and record everything it does for a fixed period, which prevents the systematic bias toward big, dramatic, male-typical behavior that ad libitum watching produces. In scan sampling you record what every visible animal is doing at set intervals. Behaviors are coded against a predefined ethogram so that different observers can be compared. Modern work adds hormones and DNA extracted non-invasively from feces and urine, which is how paternity, stress physiology, and kinship get measured without touching an animal.

Habituation has costs that the field now takes seriously. Habituated primates are easier for poachers to approach, and they are vulnerable to human respiratory viruses; outbreaks traced to human contact have killed chimpanzees at long-studied sites. Standard protocols now require distance rules, masks, quarantine periods, and health screening for staff and tourists.

Key idea: Systematic sampling methods, not casual observation, made primatology a science, and habituation, which makes those methods possible, carries real risks of disease and poaching for the animals studied.

Social life: hierarchy, grooming, reconciliation, and conflict

Most primate groups are structured by dominance hierarchies, which are not simply about who is strongest. In many macaque and baboon societies females inherit rank from their mothers, so a small young female may outrank a large adult because of who her mother is. High rank buys priority of access to food and mates, but the returns vary by species and are often modest, while the costs of maintaining rank can be substantial.

Grooming is the currency of primate social life. It removes parasites, but the time invested vastly exceeds hygienic need, and it is exchanged strategically for tolerance, coalition support, and access to infants. Frans de Waal and Angeline van Roosmalen documented in 1979 that chimpanzees who fight seek each other out afterward for contact, an act they called reconciliation, and later work found consolation of victims by uninvolved bystanders in apes. These findings shifted the field's picture of primate societies from a scoreboard of competition to a network of relationships that individuals actively repair.

Aggression is real too. Sarah Blaffer Hrdy's work on Hanuman langurs at Mount Abu in the 1970s documented males killing unweaned infants after taking over a group, and she argued this was a reproductive strategy: a female whose infant dies resumes cycling sooner, which benefits the new male. The interpretation was fiercely contested, with critics arguing that infanticide was a pathology of crowded or disturbed populations rather than an adaptation. Broad comparative analyses across mammals have since supported the sexual selection interpretation as a general pattern, but individual cases remain arguable, and this is a good example of a claim that is well supported in aggregate while still being genuinely uncertain in particulars. Chimpanzees also conduct lethal intergroup violence, documented at Gombe and elsewhere, and whether human provisioning contributed to those observations was debated for years before a multi-site analysis found the pattern held independent of human interference.

What primate minds can do, stated carefully

Great apes pass the mirror mark test devised by Gordon Gallup in 1970, touching a mark on their own body visible only in a reflection; most monkeys do not, although the result has been complicated by training studies and by claims for other animals. Chimpanzees track what others can and cannot see and adjust their behavior accordingly, which is well established through competitive food paradigms. Whether they represent false beliefs, the classic test of a full theory of mind, remains disputed: anticipatory looking studies have reported positive results and other laboratories have questioned the interpretation. Chimpanzees at the Kyoto primate center, especially an individual named Ayumu, perform remarkably at brief numerical memory tasks, though critics point to extensive practice as an explanation for the human comparison. A captive chimpanzee named Santino at a Swedish zoo was documented collecting and caching stones in the morning to throw at visitors later, which is a reasonable candidate for planning for a future motivational state.

The ape language projects deserve a straight account, because they are usually reported either as triumph or as fraud and were neither. Early attempts to teach vocal speech failed for anatomical reasons. Sign-based projects with the chimpanzee Washoe and the gorilla Koko produced substantial vocabularies of signs used to request and label. Herbert Terrace's project with Nim Chimpsky was designed to test whether apes produce sentences, and his frame-by-frame analysis of the video concluded that most of Nim's multi-sign strings were prompted or imitative rather than syntactic. The bonobo Kanzi, working with lexigrams, produced the most impressive results, including comprehension of novel spoken English requests at a level roughly compared to a two-year-old child. The honest summary is this: apes can acquire large symbol vocabularies and considerable comprehension, and no ape has demonstrated the productive, recursive syntax that human children generate spontaneously by age three. Researchers still disagree about how to weigh the comprehension results, and you should be suspicious of anyone who reports this literature without mentioning the disagreement.

Key idea: Primate cognition research supports substantial abilities in self-recognition, perspective taking, memory, and symbol use, while the evidence for false-belief understanding and for human-like syntax remains contested.

Tools and traditions

Tool use in the wild is now documented across many species, and it varies by community in ways that ecology alone cannot explain.

Species and site	Tool behavior	Why it matters
Chimpanzees, Gombe	Termite fishing with modified twigs	The 1960 observation that broke the toolmaking boundary
Chimpanzees, Tai and Bossou	Hammer and anvil nut cracking	Takes years to learn; absent in nearby communities with the same nuts
Chimpanzees, Fongoli	Sharpened branches used to jab bushbabies in tree hollows	Tool-assisted vertebrate hunting, performed mostly by females and juveniles
Capuchins, Serra da Capivara	Stone pounding that incidentally produces sharp-edged flakes	Complicates the archaeological signature of intentional flaking
Long-tailed macaques, Thailand	Stone hammers for shellfish and nuts	Independent evolution of stone tool traditions outside the ape lineage
Orangutans, Suaq Balimbing	Stick tools for extracting seeds and honey	Present in one swamp population and absent in others

In 1999 Andrew Whiten and colleagues, pooling decades of data from seven long-term chimpanzee field sites, identified thirty-nine behavior patterns that were customary at some sites and absent at others without an ecological or genetic explanation. They called the result culture. The paper's method of exclusion, ruling out ecology and genetics and attributing the remainder to social learning, was criticized by Kevin Laland and others as too weak, since you cannot confidently exclude subtle ecological differences you have not measured. The stronger evidence came afterward, from controlled diffusion experiments in which a technique seeded in one individual spread through a group and persisted, and from field experiments showing the same.

What separates chimpanzee traditions from human culture is worth stating precisely, because it is not intelligence. Human culture is cumulative: innovations accumulate across generations so that no individual could reinvent the technology they use, a property sometimes called the ratchet effect. Chimpanzee traditions are stable but show little accumulation of complexity over time. The likely reason is transmission fidelity. Human teaching, imitation of process rather than result, and language allow high-fidelity copying; chimpanzees mostly acquire techniques by watching and reinventing, which loses detail each generation. Deliberate teaching has been documented rarely, including mothers at Tai adjusting nut-cracking tools for infants.

Key idea: Chimpanzees have genuine socially transmitted traditions, but human culture is distinguished by cumulative complexity built on high-fidelity transmission, not by the mere presence of learned behavior.

What living primates can and cannot tell us about us

Now the methodological warning, and it is one of the most important ideas in the course. There is a strong temptation to use a living species as a stand-in for our ancestors: to say that early hominins lived like savanna baboons, or that the common ancestor behaved like a chimpanzee. This is a referential model, and it fails for a simple reason. Chimpanzees have been evolving for exactly as long as we have. The common ancestor of humans and chimpanzees was not a chimpanzee, and the discovery of Ardipithecus, with its small canines and its odd mixture of features, suggested to its describers that the ancestor may have been considerably less chimpanzee-like than the field had assumed.

The deeper problem is that referential models let you choose your conclusion. Build your account of human nature on common chimpanzees and you get male coalitions, territorial killing, and dominance. Build it on bonobos, whose females form coalitions that constrain male aggression and who use sexual behavior to defuse tension, and you get something almost opposite. Both species are equally related to us. Any argument that quietly picks the more convenient cousin is not evidence, it is preference with a citation.

What works instead is conceptual modeling using the comparative method. Rather than picking one species as a portrait of the past, you sample many species to establish a relationship between variables, then apply the relationship to the fossil evidence. Sexual dimorphism and mating system is the classic case: measure dimorphism across dozens of primate species, establish the correlation, then use fossil size ranges to infer something about hominin mating systems, while being explicit about the error bars. A second legitimate move is phylogenetic bracketing: a trait shared by chimpanzees, bonobos, gorillas, and humans is more parsimoniously reconstructed as ancestral than as four independent inventions.

Key idea: Living primates are evidence about relationships between variables, not portraits of our ancestors, and any argument that selects a single convenient species as a model of early humans should be treated with suspicion.

The conservation emergency

A large collaborative assessment led by Alejandro Estrada and published in Science Advances in 2017 found that roughly sixty percent of primate species were threatened with extinction and about three-quarters had declining populations. Madagascar is the worst case: recent IUCN assessments have found nearly all lemur species threatened, an entire endemic radiation at risk simultaneously.

The drivers are mostly about land. Agricultural expansion for oil palm, soy, cattle, and rubber removes forest at scale; logging and mining fragment what remains; roads open interiors to hunting. Direct hunting for meat and the live capture of infants for the pet trade take animals directly, and both are worsened by fragmentation. Disease is a serious and underappreciated cause: Ebola outbreaks in Central Africa in the 2000s are estimated to have killed a very large fraction of gorillas in affected areas, and human respiratory viruses kill habituated apes.

Two honest complications. First, most primate range countries are also places with high rates of rural poverty, and conservation that treats local people as the problem tends to fail; approaches built on secure land tenure, community management, and genuine benefit sharing do better. Second, consumption in wealthy countries drives much of the land conversion, which means the leverage is not only local. There is at least one clear success worth holding onto: intensive protection, veterinary care, and carefully regulated tourism raised the mountain gorilla population from a few hundred in the 1980s to more than a thousand, and in 2018 the subspecies was downlisted from Critically Endangered to Endangered. It is the exception, and it shows what sustained investment can do.

Common misconceptions

• Only humans make tools. Chimpanzees, capuchins, macaques, and orangutans all manufacture and use tools, with traditions that differ between neighboring communities.
• Apes have been taught language. Apes acquire symbol vocabularies and considerable comprehension; none has shown the productive recursive syntax human children generate spontaneously.
• Chimpanzees show what early humans were like. Chimpanzees have evolved as long as we have, and the choice between chimpanzee and bonobo models determines the answer before the evidence is consulted.
• Dominance means the biggest male controls everything. Rank is often inherited maternally, coalitions frequently outweigh individual strength, and high rank carries costs as well as benefits.
• Conservation is mainly about stopping local hunters. Habitat conversion driven by global commodity demand is the largest single pressure, and excluding local communities generally makes outcomes worse.

Recap

• Focal animal sampling and ethograms made behavioral data comparable; habituation enables the work but risks disease and poaching.
• Primate societies run on hierarchy, grooming, coalitions, and reconciliation, with infanticide and intergroup violence documented and their interpretation partly contested.
• Cognitive evidence supports self-recognition, perspective taking, memory, and symbol use; false belief and syntax remain unsettled.
• Tool traditions differ between communities and are transmitted socially; human culture is distinguished by cumulative complexity from high-fidelity transmission.
• Referential models using one living species as an ancestral portrait mislead; conceptual modeling across many species and phylogenetic bracketing are the sound alternatives.
• About sixty percent of primate species are threatened, driven mainly by habitat conversion, hunting, and disease.

Sources

1. Whiten, A., Goodall, J., McGrew, W. C., Nishida, T., Reynolds, V., Sugiyama, Y., Tutin, C. E. G., Wrangham, R. W., & Boesch, C. (1999). Cultures in chimpanzees. Nature, 399, 682-685. doi.org
2. Estrada, A., et al. (2017). Impending extinction crisis of the world's primates. Science Advances, 3(1), e1600946. doi.org
3. IUCN. (n.d.). The IUCN Red List of Threatened Species. iucnredlist.org
4. the Jane Goodall Institute. (n.d.). Chimpanzee behavior and research at Gombe. janegoodall.org
5. Britannica. (2024). Primate social behavior. britannica.com

The big picture

At Koobi Fora, on the eastern shore of Lake Turkana, a survey line is six people spaced ten meters apart, walking slowly across grey and buff badlands, staring at the ground. Nobody digs. Most of what you stoop for is rock. Then a fragment catches your eye because its texture is wrong for the sediment around it, and you turn it over, and it is a molar: enamel, cusps, roots. You do not pick it up. You flag it, and the next several days go to establishing which layer it eroded from, because a tooth without a layer is a curiosity and a tooth with a layer is data.

This lesson is about how that fragment becomes evidence. Three separate problems have to be solved: how it survived at all, when it was alive, and what its anatomy means. Each has an established method and each has a characteristic way of going wrong, and knowing both is the difference between reading the literature and being pushed around by it.

Taphonomy: everything that happens between death and your hand

Taphonomy is the study of what happens to organic remains between death and recovery. The default outcome is total destruction. An animal dies, scavengers disarticulate and gnaw it, insects and bacteria consume the soft tissue, sun and rain crack and exfoliate the bone surface over a few years, and within a decade or two on an open surface there is nothing left. Fossilization requires that this sequence be interrupted by rapid burial: a flood depositing silt, a lake margin, a volcanic ash fall, a collapse into a cave. Once buried, groundwater slowly replaces or infills the bone's mineral structure, and over enough time the specimen becomes rock. Then it must survive tectonic deformation, and finally erosion must expose it in the brief window during which a person happens to walk past. The number of individuals that complete this obstacle course is a vanishingly small fraction of those who lived.

The consequence is that the record is not a random sample. It is biased in ways you can predict, which means you can partly correct for it.

• Environment. Depositional settings preserve; erosional ones do not. The East African record is rich because the Rift Valley combines active sedimentation with volcanic ash and continuing uplift and erosion. Tropical rainforests, with acidic, biologically active soils, preserve almost nothing, which is why the fossil record of chimpanzees, our closest relatives, amounts to a handful of teeth from a single Kenyan locality around half a million years old.
• Body part. Dense, mineral-rich tissue survives best, so teeth are by a wide margin the most common hominin fossils. This is why so much of paleoanthropology is dental anatomy: it is not that teeth are the most interesting, it is that teeth are what there is.
• Body size and age. Large robust bones survive better than small delicate ones, so infants and juveniles are badly under-represented, which distorts everything we can say about growth and mortality.
• Accumulating agent. Many South African cave assemblages were collected by predators rather than by geology. The Taung child's skull carries damage around the eye sockets consistent with a large raptor, and much of the Swartkrans material appears to be leopard leavings dropped into a shaft. The dead in those caves are a sample of what a predator could catch.

Key idea: The fossil record over-represents durable tissues, large bodies, and depositional environments, so a first appearance date is a statement about preservation and discovery before it is a statement about biology.

Dating, without re-deriving it

If you have worked through an introductory archaeology course, you already have the physics: superposition, radiocarbon and its calibration, and the logic of radiometric decay. This lesson will not repeat that derivation. What matters here is which method applies to hominin material and why the answer differs sharply between East and South Africa.

Start with the limitation that shapes the whole field. Radiocarbon dating is only useful back to roughly fifty thousand years, because after about ten half-lives there is too little carbon-14 left to measure reliably. That covers Neanderthals, the late Denisovans, and the spread of our own species, and it covers essentially nothing else in this course. For everything older you need other clocks.

Method	What is dated	Useful range	Where it matters here
Biostratigraphy	Associated fauna, especially pigs and elephants	Relative, calibrated by other methods	Cross-correlating East African sites
Potassium-argon and argon-argon	Volcanic ash and lava	About 100,000 years to billions	The backbone of East African hominin chronology
Paleomagnetism	Iron minerals recording field reversals	Millions of years, as a pattern match	Bracketing at the Brunhes-Matuyama boundary near 780,000 years
Uranium series	Carbonate flowstones, some teeth	To about 500,000 years	South African caves; contested cave art dates
Electron spin resonance	Tooth enamel	Tens of thousands to about two million years	Sites without volcanic material
Optically stimulated luminescence	Last exposure of quartz grains to sunlight	To a few hundred thousand years	Sediment burial at open-air sites in Australia and Africa
Cosmogenic nuclide burial dating	Aluminium and beryllium isotopes in buried quartz	Hundreds of thousands to several million years	South African cave infills such as Sterkfontein

Now the geography. The East African Rift is a paleoanthropologist's ideal: volcanoes erupt periodically, dropping datable ash layers called tuffs across a landscape where lake and river sediments are accumulating and burying bone. A fossil found between two dated tuffs is bracketed: older than the one above, younger than the one below. This is why the Hadar and Turkana specimens carry confident dates to within tens of thousands of years.

The South African cave sites have no volcanoes. Their deposits are breccia, a concrete-like mix of sediment, bone, and calcite that accumulated in complex sequences and was repeatedly reworked. Dating them requires uranium-series work on flowstones that may or may not have formed at the same time as the fossils, or cosmogenic burial dating of the quartz grains in the infill. The results have been fought over for decades. The nearly complete Sterkfontein skeleton known as Little Foot has been assigned ages ranging from about 3.7 million to considerably younger, with the disagreement turning on whether the dated material is genuinely contemporaneous with the bones. When you read that a South African fossil is such-and-such an age, look for who dated what.

Key idea: Radiocarbon reaches back only about fifty thousand years, so most hominin chronology depends on volcanic ash in East Africa and on harder, more contested methods in the South African caves.

Homology, homoplasy, and cladistic reasoning

To use anatomy as evidence of relationship you must first separate two kinds of similarity. Homology is similarity inherited from a common ancestor: the same bones in the same order in a human arm, a bat wing, and a whale flipper. Homoplasy, which includes convergence and parallelism, is similarity acquired independently: the wings of a bat and of a bird, or the streamlined shape of a dolphin and of a shark. Only homology carries information about ancestry, and telling the two apart is a large part of the practical work.

Cladistics formalizes this. A trait present in the ancestor and retained by descendants is primitive, or plesiomorphic; a trait that is new in a lineage is derived, or apomorphic. Only shared derived traits, called synapomorphies, are evidence that two groups are closely related. Shared primitive traits are not: chimpanzees and gorillas both have body hair, but so did the ancestor of all apes, so hairiness tells you nothing about whether they are each other's closest relatives. Analyses combine many characters and prefer the tree requiring the fewest independent changes, a criterion called parsimony.

Two cautions. First, a cladogram shows branching relationships, not ancestry: it says two forms share a more recent common ancestor with each other than with a third, and does not claim either is the ancestor of the other. Second, homoplasy is common in hominin evolution, particularly in features related to chewing hard foods and to bipedal posture, which is exactly why morphological trees of early hominins keep changing and why researchers argue.

Functional morphology: what a bone can tell you about a life

Bone is living tissue that remodels in response to loading, a relationship summarized as Wolff's law, so shape carries a record of use as well as of ancestry. Functional morphology is the discipline of reading that record, and it works best when a specific mechanical hypothesis is tested against living species where behavior is known.

Question	Feature examined	Reasoning
What did it eat?	Molar area, enamel thickness, mandibular robusticity, sagittal crest, zygomatic flaring	Large flat molars with thick enamel and big chewing muscles indicate hard or abrasive foods
What did it eat recently?	Dental microwear: pits versus parallel scratches	Pitting reflects hard brittle items; scratching reflects tough leaves and grasses; wear reflects the last days of life
What was in the diet over years?	Stable carbon isotopes in enamel	Distinguishes C3 plants of trees and shrubs from C4 grasses and sedges, recorded while enamel formed
How did it move?	Foramen magnum position, pelvic shape, femoral bicondylar angle, foot arch, curvature of finger bones	Bipedal posture requires the head balanced above the spine and the knee brought under the body's midline
How big was it?	Joint surface areas, especially the femoral head	Weight-bearing joints scale predictably with body mass across primates
How big was the brain?	Endocranial volume from an endocast	The interior of the braincase preserves volume and sometimes surface impressions of vessels and sulci

A worked example is coming in Module 4 that shows both the power and the danger. Paranthropus boisei has the largest cheek teeth and the most massive chewing apparatus of any hominin, which for decades earned it the nickname Nutcracker Man. Then microwear analysis found the wear pattern of a hard-object feeder was absent, and carbon isotopes showed a diet dominated by C4 plants such as sedges and grasses. Anatomy said it could crack hard things; wear and chemistry said it usually did not. The lesson is that morphology reveals capacity, and capacity is not the same as behavior. Whenever you can, get an independent line of evidence.

Key idea: Anatomy reveals what an animal was capable of, while wear and chemistry reveal what it actually did, and the two can disagree, as they did for Paranthropus boisei.

The honest limits

Most hominin fossils are fragments. Many are distorted by the weight of sediment. Species-level claims often rest on a handful of specimens, sometimes on one, and the field has repeatedly named new species that later turned out to fall inside the variation of an existing one. There are structural safeguards: publishing measurements so others can reanalyze, distributing high-quality casts, and increasingly releasing three-dimensional surface scans so that anyone can measure a specimen without traveling to it. There is also an ongoing controversy about access, since some celebrated fossils have been kept from independent study for long periods, which is the opposite of how a science that argues from small samples should operate.

Common misconceptions

• Fossils are just old bones. Fossilization involves burial and mineral replacement, and the vast majority of individuals leave no trace at all.
• Radiocarbon dates the whole human fossil record. It reaches only about fifty thousand years, covering the last chapter of this course and almost none of the rest.
• A cladogram tells you which fossil is the ancestor. It shows relative branching order, not ancestry; naming an actual ancestor requires far stronger evidence.
• Similar anatomy always means close relationship. Homoplasy is common, especially in chewing and posture, and shared primitive traits carry no information about relatedness.
• Big teeth prove a hard-object diet. Anatomy shows capacity; microwear and isotopes showed that Paranthropus boisei mostly ate soft C4 vegetation.

Recap

• Taphonomy determines what survives, biasing the record toward teeth, large bodies, and depositional environments, and leaving tropical forests nearly blank.
• Dating hominin sites depends on volcanic ash in East Africa and on uranium series and cosmogenic methods in the South African caves, where dates are more contested.
• Only shared derived traits support relationship claims; homoplasy and shared primitive traits mislead.
• Functional morphology reads diet, locomotion, body mass, and brain size from specific features, with living comparisons as the calibration.
• Capacity is not behavior: independent lines such as microwear and isotopes are needed to say what an animal actually did.

Sources

1. Smithsonian National Museum of Natural History. (n.d.). Dating and the fossil record. Human Origins Program. humanorigins.si.edu
2. Ungar, P. S., & Sponheimer, M. (2011). The diets of early hominins. Science, 334(6053), 190-193. doi.org
3. U.S. Geological Survey. (n.d.). Geologic time and radiometric dating. usgs.gov
4. Wikipedia contributors. (2025). Taphonomy. en.wikipedia.org

The big picture

In December 1967 Vincent Sarich and Allan Wilson published a paper in Science that ran barely three pages. They had not looked at a single fossil. They had measured how strongly antibodies raised against the blood protein albumin of one primate reacted with the albumin of another, treating the strength of that reaction as a proxy for how different the two proteins were, and calibrating the scale against divergence times that fossils already fixed. Their conclusion was that humans and African apes had separated roughly five million years ago.

The paleoanthropology of the day put the split at fifteen to thirty million years, on the strength of a Miocene ape called Ramapithecus whose jaw fragments had been read as hominin. Sarich was unsubtle about the implication, arguing in effect that no fossil older than about eight million years could be a hominin no matter what its anatomy looked like. He turned out to be right. Through the 1970s and 1980s, new Ramapithecus material from Pakistan, especially a face found by David Pilbeam's team, showed the animal was a relative of orangutans. The molecules had beaten the bones.

That episode is the beginning of molecular anthropology as a discipline that can overturn conclusions rather than decorate them. This lesson covers how molecular clocks work and why their numbers keep moving, and then the ancient DNA revolution, which within about fifteen years produced genomes from people who died forty thousand years ago and rewrote Module 5 of this course.

How a molecular clock works, and why it needs calibrating

The idea rests on Motoo Kimura's neutral theory, published in 1968. Most substitutions in DNA have no effect on fitness, so they are not filtered by selection; they simply drift to fixation at a rate governed by the mutation rate. Over long spans, neutral differences between two lineages should therefore accumulate roughly in proportion to the time since they split.

The catch is in the word roughly, and in the units. A clock with an unknown rate is a stopwatch with no markings on the dial. To convert genetic distance into years you need a calibration, and there are two ways to get one, which disagree.

• Phylogenetic calibration. Take a split whose date the fossil record fixes reasonably well, such as the human-orangutan divergence at roughly thirteen million years, count the substitutions accumulated since, and derive a per-year rate. This gives about one substitution per billion base pairs per year.
• Pedigree calibration. Sequence both parents and a child and count directly how many new mutations the child carries. Studies from the 2010s converged on roughly 1.2 times ten to the minus eight per base pair per generation, which works out to about half the phylogenetic rate.

Halving the rate roughly doubles every estimated date. Under the older phylogenetic rate the human and chimpanzee lineages separated four to six million years ago; under the pedigree rate the same data give seven to thirteen million. Nobody has fully resolved this. Proposed explanations include changes in generation length over time, a slowdown in mutation rate in the great ape lineage, and undetected mutations missed by trio sequencing. The honest current statement is that the human-chimpanzee split lies somewhere in the range of six to nine million years with real uncertainty on both sides, and this matters directly, because Sahelanthropus at about seven million years sits right on the contested boundary.

Key idea: Molecular clocks measure genetic distance precisely and convert it to years only through a calibration, and the shift from fossil-calibrated to pedigree-calibrated mutation rates roughly doubled estimated divergence dates.

Genes split before populations do

Here is a distinction that resolves an enormous amount of confusion in the literature. When you compare two people's DNA at a given locus and find their common ancestor lived 800,000 years ago, that is a statement about that gene, not about their populations. Genetic variation existing in an ancestral population gets inherited by both descendant populations, so lineages at any locus coalesce somewhere back inside the ancestral population, before the populations themselves parted. Gene divergence is always older than population divergence, sometimes by a lot.

A consequence is incomplete lineage sorting. Because the ancestral population carried multiple variants at every locus and each one sorted independently, a substantial fraction of the human genome, on the order of a fifth to a third depending on the analysis, is more closely related to the gorilla sequence than to the chimpanzee sequence at that particular position. This does not mean gorillas are sometimes our closest relatives. It means single loci are noisy and only the genome-wide majority signal reflects the species tree. When you read that Neanderthals and modern humans split between about 550,000 and 765,000 years ago, that figure is a population split estimated from genome-wide data, and it is deliberately given as a range.

The ancient DNA revolution, including its embarrassing decade

In 1984 Russell Higuchi, working in Allan Wilson's laboratory, recovered short mitochondrial sequences from a 140-year-old museum skin of a quagga, an extinct South African zebra. The following year Svante Paabo published DNA from an Egyptian mummy, a result later shown to be modern contamination. Then the polymerase chain reaction arrived, and PCR will amplify anything, including a single stray molecule from the person holding the pipette.

The early 1990s produced claims of DNA from dinosaur bone, from insects in amber, and from Miocene magnolia leaves. None replicated. The dinosaur sequence turned out to be human. By 2000 the field had adopted strict authentication criteria, and the reforms are what made everything since possible: dedicated clean rooms physically separate from post-PCR labs, bleach and ultraviolet decontamination, negative controls at every step, and independent replication in a second laboratory.

The deeper solution came from understanding how ancient DNA breaks. After death, DNA fragments rapidly, so ancient molecules are short, with median lengths often between about 50 and 70 base pairs. Depurination cleaves the backbone. And cytosine deaminates to uracil, which sequencing machines read as thymine, producing a characteristic excess of cytosine-to-thymine misreads concentrated at the ends of fragments, where the DNA is single-stranded. That damage pattern cannot be faked by modern contamination, so it became the signature that authenticates a sequence as genuinely old. What was once a source of error is now the quality check.

Key idea: Cytosine deamination at fragment ends and short fragment lengths distinguish genuinely ancient DNA from modern contamination, turning the chemistry of degradation into the field's authentication tool.

What high-throughput sequencing made possible

Next-generation sequencing, which reads millions of short fragments at once, is exactly suited to material that survives only as short fragments. Combined with hybridization capture, which uses probes to fish target sequences out of a sample dominated by soil bacteria, it transformed what was recoverable. One further trick was anatomical: in 2015 Ron Pinhasi's group showed that the petrous portion of the temporal bone, the dense capsule around the inner ear, yields far more endogenous DNA than other skeletal elements, in some cases by a factor approaching a hundred. That single finding reoriented sampling practice, and it also sharpened an ethical problem, because obtaining it destroys part of the ear region of a skull.

Year	Result	Why it mattered
2010	Draft Neanderthal genome, Green and colleagues	Showed one to four percent Neanderthal ancestry in non-Africans
2010	Denisova Cave finger bone genome	A previously unknown population identified from DNA before any diagnostic skull was known
2014	High-coverage Altai Neanderthal genome, Pruefer and colleagues	Population split estimates and evidence of close inbreeding in that individual's ancestry
2016	Nuclear DNA from Sima de los Huesos, Spain, about 430,000 years old	Oldest hominin nuclear DNA recovered; placed the Sima people on the Neanderthal line
2017	Neanderthal and Denisovan DNA from cave sediments, Slon and colleagues	Occupation detectable in caves containing no hominin bones at all
2019-2020	Enamel proteomes of Gigantopithecus at 1.9 million years and Homo antecessor at about 800,000	Proteins survive far longer than DNA and can resolve relationships
2021-2022	Mammoth DNA past one million years; environmental DNA at about two million years in Greenland	Established that the ceiling is set by temperature, not by time alone

The limits, which are severe and geographically unfair

Survival is governed by thermal age: the integrated effect of temperature over time. Permafrost preserves; a hot cave does not. The consequence for this course is stark. The continent where nearly all of human evolution happened is the continent from which ancient DNA is hardest to recover. The oldest African human genomes come from Taforalt in Morocco at roughly 15,000 years, from Mota Cave in Ethiopia at about 4,500 years, and from Shum Laka in Cameroon at roughly 8,000 and 3,000 years. Set that against 430,000-year-old nuclear DNA from a cool cave in northern Spain. There will almost certainly never be DNA from Lucy, and no amount of funding changes that.

Ancient proteins partly compensate, since enamel peptides survive an order of magnitude longer, and amelogenin peptides can even establish biological sex from a single tooth. But proteins carry far less information than a genome, and the mandible from Baishiya Karst Cave in Tibet was assigned to Denisovans on the strength of collagen peptide differences, which is a thinner thread than a sequenced genome.

The ethics are not an appendix to this. Sampling is destructive and irreversible, and the remains are people. The Ancient One, also called Kennewick Man, was found in Washington State in 1996, litigated over for years, sequenced in 2015 with a result showing closest affinity to living Native Americans, and repatriated for reburial in 2017. Indigenous scholars including Kim TallBear have argued that consent, benefit, and control over research questions belong to descendant communities rather than to whoever holds the bone. A large international group of practitioners published global guidelines in 2021 covering permits, community engagement, minimal destructive sampling, and data sharing. Compliance is uneven, and this is a live argument rather than a settled protocol.

Key idea: Ancient DNA recovery is governed by thermal history, so the record is richest exactly where human evolution was not happening, and every sample destroyed is part of a person whose descendants may have a claim on the decision.

Common misconceptions

• Molecular clocks give exact dates. They give distances, which become dates only through a calibration whose value is disputed, which is why estimates have roughly doubled.
• Dinosaur DNA has been recovered. The 1990s claims were contamination and none replicated; the oldest authenticated DNA is roughly one to two million years old and comes from permafrost.
• If two genes give different trees, one study is wrong. Incomplete lineage sorting guarantees that individual loci disagree; only genome-wide signal reflects the species tree.
• Ancient DNA can be recovered from any old bone. Survival depends on temperature history, which is why Siberian and northern European samples dominate and tropical African samples barely exist.
• Sequencing ancient remains is non-invasive. It consumes bone or tooth permanently, usually from the petrous bone or a molar root.

Recap

• Sarich and Wilson's 1967 albumin study put the human-African ape split near five million years and forced the reinterpretation of Ramapithecus.
• Clocks need calibration; the switch from fossil-based to pedigree-based mutation rates roughly doubled divergence estimates and the discrepancy is unresolved.
• Gene divergence predates population divergence, and incomplete lineage sorting makes a large minority of loci disagree with the species tree.
• Ancient DNA is short and carries cytosine-to-thymine damage at fragment ends, a signature now used for authentication after the contamination failures of the 1990s.
• Petrous bone sampling, hybridization capture, and high-throughput sequencing produced Neanderthal, Denisovan, and 430,000-year-old Sima genomes.
• Thermal history limits recovery, leaving Africa's deep past inaccessible, and destructive sampling raises consent questions now covered by international guidelines.

Sources

1. Sarich, V. M., & Wilson, A. C. (1967). Immunological time scale for hominid evolution. Science, 158(3805), 1200-1203. doi.org
2. Green, R. E., et al. (2010). A draft sequence of the Neandertal genome. Science, 328(5979), 710-722. doi.org
3. Meyer, M., et al. (2016). Nuclear DNA sequences from the Middle Pleistocene Sima de los Huesos hominins. Nature, 531, 504-507. doi.org
4. Max Planck Institute for Evolutionary Anthropology. (n.d.). Department of Evolutionary Genetics. eva.mpg.de
5. Wikipedia contributors. (2025). Ancient DNA. en.wikipedia.org

The big picture

In 1976 a group of paleontologists working at Laetoli in northern Tanzania got into a fight with elephant dung. Andrew Hill, ducking a lump thrown by a colleague, fell onto a dry streambed and found himself looking at a hardened ash surface pocked with animal tracks. Two years later Paul Abell spotted something else in the same tuff: a heel print that did not belong to any animal in the fauna. Excavation exposed a trackway of about seventy hominin footprints running some twenty-seven meters, made by at least two individuals walking side by side across wet volcanic ash from the Sadiman volcano roughly 3.66 million years ago, before rain and sun set the ash like cement.

Read the prints themselves. There is a deep heel strike, a transfer of weight along a raised arch, and a push-off from a big toe held in line with the others, not sticking out sideways like a thumb. There is no knuckle impression anywhere. The stride length indicates a walk, not a run. Whoever made these prints, roughly 700,000 years before the first stone tool anyone has found, was already walking the way you walked to the room you are reading this in.

That is the subject of this lesson: the anatomy of habitual bipedalism, the handful of contested fossils that push its origin back toward seven million years, the price the human body still pays for it, and the competing accounts of why it happened at all.

What makes a fossil a hominin

Two features are conventionally used to identify a member of the human lineage after its split from the chimpanzee lineage. The first is habitual bipedalism. The second is a change in the front teeth: apes have large, projecting, blade-like upper canines that are sharpened against the lower third premolar in what is called a honing complex. Hominins lose this. The canine shrinks, becomes incisor-shaped, and the honing wear disappears, which most researchers read as a reduction in male-male canine display and fighting.

Neither criterion is airtight. Some Miocene apes show partial reductions, and bipedal posture is not an all-or-nothing condition. But when a fossil shows both, the case is strong.

Reading bipedalism off a skeleton

Walking upright is a whole-body reorganization, and each change leaves a signature that survives in fossil bone. This table is worth learning properly, because Module 4 and Module 5 both assume it.

Region	Human condition	Ape condition	Why
Foramen magnum	Positioned forward, under the skull	Positioned toward the rear	Balances the head on a vertical spine instead of hanging it from neck muscles
Spine	S-shaped, with a lumbar curve and typically five lumbar vertebrae	C-shaped, with three or four short lumbar vertebrae	Places the trunk's center of mass over the hips
Pelvis	Short, broad, basin-shaped, with iliac blades rotated to face sideways	Tall, flat, blade-like ilia facing backward	Repositions gluteus medius and minimus as abductors that stop the pelvis dropping during single-leg support
Femur	Angled inward from hip to knee, producing a valgus or bicondylar angle	Nearly vertical shaft	Brings the knee and foot under the body's midline so you do not lurch side to side
Knee	Deep patellar groove with a high lateral lip; expanded condyles	Shallow groove	Keeps the kneecap tracking under full extension and body weight
Foot	Longitudinal and transverse arches, robust calcaneus, adducted big toe, short straight toes	Flat, flexible foot with a divergent grasping big toe and long curved toes	Converts the foot from a grasping organ into a shock absorber and lever

The pelvic point repays a moment. Stand up and lift one foot off the floor. Your pelvis stays level. It does that because the gluteal muscles on the standing side contract to hold it, which they can only do because your iliac blades face sideways. A chimpanzee cannot do this, which is why a chimpanzee walking bipedally rolls heavily from side to side. Clinicians still test for weakness in exactly these muscles by asking a patient to stand on one leg and watching whether the opposite hip drops.

Key idea: Bipedalism is legible in the fossil record because it required specific structural changes at the skull base, spine, pelvis, knee, and foot, each of which is recognizable in a single bone.

The earliest candidates, and the arguments about them

Three sets of fossils push hominin origins back past four million years. All three are contested, and you should know why.

Sahelanthropus tchadensis was recovered in 2001 at Toros-Menalla in the Djurab desert of Chad by a team led by Michel Brunet, and dated by associated fauna to roughly seven to six million years. The cranium, nicknamed Toumai, was badly crushed and had to be digitally reconstructed. Its case for hominin status rests on a small, non-honing canine, an enormous continuous brow ridge combined with a small brain of about 360 cubic centimeters, and a foramen magnum that the describers argue is positioned forward. In 2022 the team published a femur and two ulnae from the site, arguing they show bipedalism. Roberto Macchiarelli, Bernard Wood, and colleagues have argued the opposite from the same femur, contending its features fall within the ape range and that the specimen may not even belong to the same individual or taxon. This is unresolved.

Orrorin tugenensis, from the Tugen Hills of Kenya at about six million years, was announced in 2000 by Brigitte Senut and Martin Pickford. The strongest evidence is a femoral neck whose internal cortical bone is thicker on the underside than the top, a distribution produced by the loading pattern of bipedal walking. The same individual's finger bone is curved, in the way of an animal that still climbed.

Ardipithecus ramidus is the richest of the three. Tim White's team found the partial skeleton catalogued ARA-VP-6/500, nicknamed Ardi, at Aramis in Ethiopia's Middle Awash in 1994, dated to 4.4 million years, and spent fifteen years extracting bone so soft it had to be stabilized in the ground before removal. Publication came in 2009. Ardi stood about 1.2 meters, weighed perhaps 50 kilograms, had a brain of 300 to 350 cubic centimeters, and combined a short broad pelvis suited to upright walking with a fully divergent, grasping big toe. She was a biped on the ground and a careful climber in trees, and she is a genuine mosaic rather than a halfway house. Her canines are small in both sexes. Some researchers, including Esteban Sarmiento, have questioned whether Ardipithecus is a hominin at all rather than an ape near the divergence.

One finding from Aramis outlived every argument about Ardi's status. The associated fauna and plant remains, including forest-floor snails, colobus monkeys, and seeds of woodland trees, describe woodland, not open grassland. That evidence killed the simple savanna hypothesis: whatever bipedalism was for, it did not begin on a treeless plain.

Key idea: The earliest hominin candidates are each known from few specimens and every one of them is disputed, but the Aramis environmental evidence establishes that bipedalism began in woodland rather than open savanna.

What bipedalism cost

Natural selection does not optimize; it trades. Reorganizing a quadruped for upright walking created a set of problems humans still live with. The vertebral column carries compressive loads it was not built for, which is why intervertebral discs herniate and why low back pain is among the leading causes of disability worldwide. Abdominal contents press downward onto a pelvic floor and an inguinal region evolved for horizontal loading, producing hernias and prolapse. Blood returning from the legs must climb, producing varicose veins. Weight passing through two joints rather than four contributes to knee and hip osteoarthritis. And humans are slow: a world-class sprinter is outrun by an ordinary dog.

The most discussed cost is childbirth. The classic obstetrical dilemma holds that the pelvis was narrowed for efficient walking while the fetal head enlarged, leaving human birth uniquely difficult and requiring the infant to rotate through a twisted birth canal. Human infants are born neurologically immature as a result, a condition called secondary altriciality.

That account is under real challenge. Holly Dunsworth and colleagues argued in 2012 that gestation length is set not by pelvic width but by the mother's metabolic ceiling: at about nine months, the energy demand of the fetus approaches roughly twice the mother's baseline metabolic rate, which is about the sustainable limit. Anna Warrener and colleagues then measured the locomotor cost of a wider pelvis and found no significant energetic penalty, removing the assumed pressure keeping the birth canal narrow. Defenders of the dilemma answer that obstructed labor is still a leading cause of maternal death and that the fit really is tight. Both camps agree the fit is tight; they disagree on why.

Key idea: Bipedalism imposed back, hernia, circulatory, and joint problems, and the traditional explanation for difficult human birth is now contested by evidence that gestation length is capped by maternal metabolism rather than by pelvic width.

Why did it happen?

There is no accepted answer, and there are several serious proposals with different kinds of support.

• Energetic efficiency. The best-supported. Peter Rodman and Henry McHenry argued in 1980 that upright walking is cheap over distance. Michael Sockol, David Raichlen, and Herman Pontzer tested it directly in 2007 by measuring oxygen consumption in chimpanzees trained to walk on a treadmill both quadrupedally and bipedally, alongside humans. Human walking cost roughly a quarter of chimpanzee locomotion per unit distance. If food patches became more scattered, cheap travel is worth a great deal.
• Carrying. Owen Lovejoy's provisioning model links upright posture to freed hands for carrying food, and ties it to reduced canines and a shift in mating strategy. Its weakness is that it is hard to test on fossils.
• Thermoregulation. Peter Wheeler calculated that an upright body intercepts far less solar radiation at midday and sits higher in cooler moving air. The Aramis woodland evidence undercuts it as an origin explanation, though it may have mattered later.
• Postural feeding. Kevin Hunt observed that chimpanzees stand bipedally most often while feeding from small trees, suggesting the posture began as a feeding stance and was later co-opted for travel.

These are not mutually exclusive, and the fossils are consistent with a long period of facultative, mixed locomotion before anything like the Laetoli walk. What the record does not support is a single dramatic moment when an ape stood up on a savanna.

Common misconceptions

• Bipedalism evolved on open savanna. The Aramis fauna and flora place Ardipithecus in woodland at 4.4 million years, before grasslands expanded.
• Bipedalism came after big brains. The order is the reverse: Ardi and Lucy walked upright with brains around a third to a quarter the size of ours.
• Early hominins were either bipeds or climbers. Ardipithecus combined a bipedal pelvis with a grasping big toe; mixed locomotion persisted for millions of years.
• The earliest hominins are securely identified. Sahelanthropus, Orrorin, and Ardipithecus are each argued about, including whether they are hominins at all.
• Human birth is difficult purely because of pelvic width. The metabolic ceiling argument challenges that account, and the locomotor cost of a wider pelvis appears to be negligible.

Recap

• The Laetoli trackway at 3.66 million years records modern-style walking with an adducted big toe and a raised arch.
• Hominin status is conventionally judged by habitual bipedalism plus canine reduction and loss of the honing complex.
• Bipedalism restructured the foramen magnum, spine, pelvis, femur, knee, and foot, each recognizable in isolated bones.
• Sahelanthropus, Orrorin, and Ardipithecus push the record toward seven million years and are all contested.
• The costs include back pain, hernias, varicose veins, joint wear, slow speed, and difficult birth, whose classic explanation is now disputed.
• Energetic efficiency has the strongest experimental support among the explanations for why bipedalism evolved, and no single hypothesis is established.

Sources

1. Smithsonian National Museum of Natural History. (n.d.). Walking upright. Human Origins Program. humanorigins.si.edu
2. Smithsonian National Museum of Natural History. (n.d.). Sahelanthropus tchadensis. Human Origins Program. humanorigins.si.edu
3. Sockol, M. D., Raichlen, D. A., & Pontzer, H. (2007). Chimpanzee locomotor energetics and the origin of human bipedalism. PNAS, 104(30), 12265-12269. doi.org
4. Britannica. (2024). Human evolution. britannica.com
5. Wikipedia contributors. (2025). Laetoli. en.wikipedia.org

The big picture

On the morning of 24 November 1974, Donald Johanson and a graduate student named Tom Gray were driving back from a survey at Hadar, in the Afar region of Ethiopia, when Johanson decided to check one more gully. In the sediment he saw a fragment of arm bone. Then a piece of skull, a femur, ribs, vertebrae, a lower jaw. Over the following weeks the team recovered several hundred fragments representing forty-seven bones from a single individual, roughly forty percent of a skeleton, catalogue number AL 288-1, dated to 3.2 million years. That night a tape of Lucy in the Sky with Diamonds was playing at camp, and the specimen has been Lucy ever since.

She stood about 1.05 meters and weighed perhaps 29 kilograms. Her brain was around 400 cubic centimeters, roughly a chimpanzee's. Her pelvis is short and broad, her femur angles inward at the knee, and there is no argument at all that she walked upright. Before Lucy, nobody had a hominin skeleton complete enough to see how the parts worked together, and the answer they gave was that bipedalism came first and the brain came later, by millions of years.

This lesson covers the australopiths, the genus that occupied Africa for over two million years, the robust forms whose diet turned out to be the opposite of their nickname, and the first stone tools, whose makers we cannot confidently name.

Australopithecus afarensis and the argument about trees

Lucy is not alone. In 1975 the Hadar team excavated locality AL 333, which produced remains of at least thirteen individuals of all ages, called the First Family. In 2000 Zeresenay Alemseged found a nearly complete three-year-old at Dikika, published in 2006 and nicknamed Selam, and older by about 100,000 years than Lucy. Add the Laetoli footprints and afarensis becomes the best-documented early hominin, spanning roughly 3.9 to 3.0 million years.

Below the neck the species is a committed biped. Above and around that, the picture is argued about. The finger and toe bones are curved. The shoulder joint faces more upward than a human's, as in apes that hang. Selam's scapula is closest in shape to a gorilla's. The arms are long relative to the legs. One camp reads these as functional signals that afarensis climbed regularly, probably to sleep and to forage. The other reads them as primitive retentions, features inherited from an arboreal ancestor and not yet selected away, since selection removes traits only when they cost something. Both readings explain the same bones. What would settle it is evidence of use rather than shape, which is why studies of internal bone structure, which remodels with actual loading, have carried the argument forward without ending it.

The First Family also created a long-running dispute about body size. Adult afarensis specimens range widely, and how you read that range depends on whether it reflects two sexes of one species or something else. Philip Reno and colleagues argued in 2003 that dimorphism was modest, close to the human condition. Michael Plavcan, Adam Gordon, and others have argued it was substantially greater, closer to gorillas, which would imply intense male competition. Since mating system inferences ride on this number, the disagreement matters, and it is a clean example of a question that more fossils, not more argument, will eventually settle.

Key idea: Australopithecus afarensis was fully bipedal with a chimpanzee-sized brain, and its curved fingers and upward-facing shoulder joint are read either as evidence of continued climbing or as unselected retentions, with the evidence supporting both readings.

The rest of the genus, and one important complication

Australopithecus anamensis, from Kanapoi and Allia Bay in Kenya at 4.2 to 3.8 million years, was long treated as afarensis's direct ancestor in a straight line, one species transforming into the next. In 2019 Yohannes Haile-Selassie published a nearly complete anamensis cranium, MRD-VP-1/1, from Woranso-Mille in Ethiopia, dated to 3.8 million years. That date overlaps with a frontal bone from Belohdelie attributed to afarensis at about 3.9 million. If both attributions hold, the two species coexisted, and the tidy ancestor-descendant line becomes a branching bush.

Australopithecus africanus is South Africa's species, known from Taung, Sterkfontein, and Makapansgat, roughly 3.3 to 2.1 million years old. Australopithecus sediba, found at Malapa in 2008 when Lee Berger's nine-year-old son Matthew picked up a clavicle-bearing block, dates to about 1.98 million years and mixes australopith and Homo-like features; whether it is close to Homo's ancestry is disputed. Kenyanthropus platyops, announced in 2001 from a flat-faced 3.5-million-year-old cranium, has been argued by Tim White to be a distorted afarensis, since the specimen is broken into more than a thousand pieces. Australopithecus deyiremeda, named in 2015, faces similar skepticism. This is not a field that agrees about how many species it has.

Taung, and what the reception says about the reception

In 1924 a quarry at Taung in South Africa sent Raymond Dart a box of fossil-bearing rock containing a natural endocast of a small brain, which fitted a face and mandible he then spent weeks freeing with his wife's knitting needles. The child was about three years old, had milk teeth and a first molar, a brain of roughly 400 cubic centimeters, and a foramen magnum positioned forward. Dart published in Nature in February 1925 as Australopithecus africanus and argued it was a human ancestor.

The reaction was dismissal, and it lasted more than twenty years. Three things drove it. First, the Piltdown remains found in England from 1912 had a large braincase with an ape-like jaw, and had convinced most of the establishment that the brain enlarged first; Taung had exactly the opposite combination. Second, Taung was a juvenile, and juvenile apes resemble hominins more than adults do, so critics argued Dart had been fooled by an infant chimpanzee. Third, almost nobody in European science expected humanity to have originated in Africa. Dart was vindicated by Robert Broom's adult specimens from Sterkfontein, including the 1947 cranium nicknamed Mrs Ples, and finally in 1953 when fluorine testing and microscopic examination exposed Piltdown as a deliberate forgery: a modern human cranium with a filed orangutan jaw, stained to match.

One postscript. In 1995 Lee Berger and Ron Clarke reexamined the damage around the Taung child's eye sockets and the puncture marks on the skull and argued it had been killed by a large eagle, based on comparison with the remains of monkeys taken by crowned eagles today. The most famous hominin fossil in South Africa is a raptor's leftovers.

Key idea: The twenty-year rejection of the Taung Child was driven by the Piltdown forgery, by the specimen's juvenile status, and by an expectation that human origins lay outside Africa, and none of those objections concerned the fossil's actual anatomy.

The robust australopiths and the nickname that was wrong

Between about 2.7 and 1.0 million years ago a distinctive group occupied eastern and southern Africa, usually placed in their own genus, Paranthropus. They are built around chewing.

Species	Key specimen	Age and region	Features
Paranthropus aethiopicus	KNM-WT 17000, the Black Skull	About 2.5 million years, West Turkana	Tiny braincase near 410 cubic centimeters with an enormous sagittal crest and extreme facial projection
Paranthropus boisei	OH 5, found by Mary Leakey in 1959 at Olduvai	About 2.3 to 1.3 million years, East Africa	Molars roughly four times the occlusal area of ours; dish-shaped face; flaring cheekbones
Paranthropus robustus	Swartkrans and Kromdraai material, from 1938	About 2.0 to 1.0 million years, South Africa	Similar but less extreme; evidence of dietary variability

The sagittal crest is a ridge of bone providing attachment area for temporalis muscles too large to fit on the braincase, and the flared zygomatic arches make room for masseter muscles. Louis Leakey nicknamed OH 5 Nutcracker Man, and for fifty years textbooks explained the anatomy as an adaptation to cracking hard nuts and seeds.

Then two independent lines of evidence arrived. Peter Ungar's microwear work found that boisei enamel lacked the heavy pitting that hard-object feeders like modern mangabeys show; the surfaces were scratched, in the pattern of tough, abrasive vegetation. And in 2011 Thure Cerling and colleagues published carbon isotope data from boisei enamel showing that around three quarters of its diet came from C4 plants, meaning grasses and sedges rather than trees and shrubs. Nutcracker Man ate sedges. The massive apparatus was for processing large volumes of low-quality abrasive food, not for concentrating force on hard objects. Interestingly, South African Paranthropus robustus does show some hard-object wear, so the two species diverged in diet as well as geography.

Key idea: Microwear and carbon isotopes overturned the hard-object interpretation of Paranthropus boisei, showing a diet dominated by C4 grasses and sedges and demonstrating that chewing anatomy indicates capacity rather than habitual behavior.

The first stone tools, and who is not proven to have made them

For decades the anchor date was Gona in Ethiopia, where Sileshi Semaw's team documented flaked stone at 2.6 million years, contemporaneous with a similar assemblage at Ledi-Geraru. This is the Oldowan, named for Olduvai Gorge: cores struck with a hammerstone to detach flakes with edges sharp enough to slice hide and muscle. The technique requires understanding conchoidal fracture and choosing a striking platform at an angle below ninety degrees, which is harder than it sounds; the flakes, not the cores, are usually the point.

Then the dates started moving. In 2015 Sonia Harmand's team published 149 artifacts from Lomekwi 3 on the western side of Lake Turkana, dated to 3.3 million years, made by resting a core on an anvil or by bipolar percussion rather than freehand flaking. If correct, stone tool use predates the genus Homo by half a million years. Some researchers have questioned the stratigraphic context and whether the breakage is genuinely deliberate. Earlier, in 2010, Shannon McPherron's team reported cut-marked bones from Dikika at 3.39 million years, and Manuel Dominguez-Rodrigo and colleagues countered that the marks are consistent with trampling by animals on sediment. Both disputes remain open in the literature.

Independent of those cases, there is good evidence that australopiths could have made tools. The afarensis hand includes a third metacarpal with a styloid process, part of the human apparatus for a forceful precision grip, and trabecular bone patterns in africanus hands indicate human-like loading. At Bouri in Ethiopia, cut-marked and hammerstone-fractured antelope bones at 2.5 million years are associated with Australopithecus garhi and no Homo. And Tiago Proffitt's 2016 finding that wild capuchins in Brazil produce sharp-edged flakes as an unintended byproduct of stone pounding is a caution about inferring intent from flakes alone.

Homo itself first appears, on current evidence, as a single mandible: LD 350-1 from Ledi-Geraru, published by Brian Villmoare's team in 2015 and dated to about 2.8 million years, with a primitive jaw shape and more derived tooth proportions. It is one specimen, and its attribution is argued about, which is where this part of the record stands.

Key idea: The earliest stone tools may predate the genus Homo, australopith hands were capable of the required grip, and no fossil maker has been securely associated with the oldest claimed assemblages.

Common misconceptions

• Lucy was an ape that had not finished becoming bipedal. Her pelvis and knee are unambiguously those of a habitual biped; what is argued about is how much she also climbed.
• Australopithecus afarensis evolved directly into Australopithecus africanus and then into Homo. Overlapping dates and multiple contemporaneous species describe a branching bush, not a ladder.
• Nutcracker Man cracked nuts. Microwear and carbon isotopes indicate a diet of C4 grasses and sedges, and the nickname has outlived the evidence for it.
• The Taung Child was rejected because the evidence was weak. It was rejected because of the Piltdown forgery, its juvenile status, and an expectation that human origins were European.
• Only Homo made stone tools. Oldowan-style flaking may go back to 3.3 million years, before Homo appears in the record, and australopith hands were capable of it.

Recap

• Lucy, AL 288-1 at 3.2 million years, established that bipedalism preceded brain expansion by millions of years.
• Curved digits and an upward-facing shoulder joint fuel an unresolved argument about how much afarensis climbed.
• The 3.8-million-year anamensis cranium published in 2019 overlaps with early afarensis, favoring branching over a single lineage.
• Taung was dismissed for twenty years for reasons that had nothing to do with its anatomy, and Piltdown's exposure in 1953 ended the objection.
• Paranthropus built an extreme chewing apparatus, and isotopes show boisei used it on abrasive C4 vegetation rather than hard nuts.
• Oldowan tools are securely dated to 2.6 million years, Lomekwi claims 3.3 million and is contested, and the identity of the first toolmakers is not established.

Sources

1. Smithsonian National Museum of Natural History. (n.d.). Australopithecus afarensis. Human Origins Program. humanorigins.si.edu
2. Cerling, T. E., et al. (2011). Diet of Paranthropus boisei in the early Pleistocene of East Africa. PNAS, 108(23), 9337-9341. doi.org
3. Harmand, S., et al. (2015). 3.3-million-year-old stone tools from Lomekwi 3, West Turkana, Kenya. Nature, 521, 310-315. doi.org
4. Villmoare, B., et al. (2015). Early Homo at 2.8 Ma from Ledi-Geraru, Afar, Ethiopia. Science, 347(6228), 1352-1355. doi.org
5. Britannica. (2024). Australopithecus. britannica.com

The big picture

In August 1984, Kamoya Kimeu was walking a slope of black lava pebbles beside the Nariokotome sand river, west of Lake Turkana, when he noticed a piece of skull about the size of a matchbox. Five field seasons of excavation followed. What came out, catalogued KNM-WT 15000, is the most complete early hominin skeleton ever found: a boy who died between roughly eight and eleven years of age, about 1.5 to 1.6 million years ago, on the muddy margin of a swamp.

Stand him up and the surprise is how ordinary he looks from the neck down. He was around 1.6 meters tall as a child. His legs are long relative to his arms. His pelvis is narrow, his rib cage is barrel-shaped rather than funnel-shaped, and there is nothing about his hands or feet that suggests climbing. From the neck up he is not ordinary at all: a long low braincase of roughly 880 cubic centimeters, thick cranial bone, a shelf of bone over the eyes, and no chin.

This lesson is about that body and the million-year career it had: the first hominin to leave Africa, the first to carry a brain that cost real metabolic money, and the one at the center of the argument about when our ancestors learned to cook.

The muddle before erectus

Between 2.4 and 1.6 million years ago the record contains several kinds of early Homo, and sorting them is unfinished business. In 1960 Jonathan Leakey found a juvenile mandible and parietal bones at Olduvai Gorge, catalogued OH 7. Louis Leakey, Phillip Tobias, and John Napier named it Homo habilis in 1964, and the naming was controversial because its brain, around 600 cubic centimeters, fell below the 750-cubic-centimeter threshold that some anatomists had proposed as the boundary of the genus. The Leakeys argued the threshold should move. It did.

KNM-ER 1470, found near Lake Turkana in 1972 and dated to about 1.9 million years, has a larger braincase near 750 cubic centimeters and a broad flat face, and is usually placed in a separate species, Homo rudolfensis. Whether habilis and rudolfensis are two species, one variable species, or in habilis's case not really Homo at all is unsettled. Bernard Wood and Mark Collard argued in 1999 that habilis fits the australopith adaptive pattern better than the Homo one in body proportions, growth, and chewing, and that it should be reassigned. The argument has not been resolved, and you will see both usages in current writing.

What changed with Homo erectus

Homo erectus appears in Africa by about 1.9 million years ago and persists in Southeast Asia until roughly 110,000 years ago, which makes it the longest-lived species of Homo by a wide margin. Some researchers separate the earlier African material as Homo ergaster; others treat the whole set as one geographically variable species. Both usages are current.

Feature	Australopithecus	Homo erectus
Cranial capacity	About 400 to 550 cubic centimeters	About 600 to 1,100, increasing through time
Body height	Roughly 1.0 to 1.4 meters	Roughly 1.5 to 1.8 meters
Limb proportions	Relatively long arms, short legs	Long legs, short arms, essentially modern
Rib cage and pelvis	Funnel-shaped thorax, wide flaring pelvis	Barrel-shaped thorax, narrow pelvis
Skull	Small braincase, prognathic face	Long low vault, thick bone, projecting brow ridge, occipital torus
Range	Africa only	Africa, the Caucasus, East and Southeast Asia

The postcranial changes point at one behavior: long-distance walking in open, hot country, with no remaining commitment to trees. Narrow hips and long legs make an efficient stride. A barrel chest without a flaring gut suggests a smaller digestive tract, which suggests better food.

Key idea: Homo erectus is the first hominin with an essentially modern body plan, built for efficient long-distance terrestrial travel, with a brain roughly twice the australopith size.

Out of Africa, earlier than anyone expected

In 1991, excavators working beneath the ruins of the medieval town of Dmanisi in Georgia, in a grain storage pit cut into older sediments, found a hominin mandible. The site has since produced five crania and postcranial remains, dated to between about 1.85 and 1.77 million years. That is essentially as old as the earliest African Homo erectus, which means the dispersal happened almost immediately, and it happened with Oldowan technology rather than with handaxes.

Dmanisi also produced one of the field's sharpest arguments. Skull 5, catalogued D4500, has a braincase of only 546 cubic centimeters attached to a very large, projecting face, a combination unlike anything else. David Lordkipanidze and colleagues argued in 2013 that the five Dmanisi crania, despite spanning as much variation as researchers use to separate African species, are one population and therefore one species. Their further inference was radical: if that much variation fits inside one species at one site, then Homo habilis, Homo rudolfensis, and Homo ergaster in Africa might also be a single variable species. Many specialists rejected the extension, arguing that variation at one site over a short interval is not comparable to variation across a continent and a million years. The Dmanisi sample is genuinely remarkable; what it implies for African taxonomy is not agreed.

Elsewhere in Asia the record is long. Eugene Dubois found a skullcap and femur at Trinil in Java in 1891 and 1892, naming them Pithecanthropus erectus, the first deliberate search for a human ancestor that actually succeeded. Sangiran in Java has yielded material around 1.5 million years and possibly older. Zhoukoudian near Beijing produced a large collection excavated between 1921 and 1937, dating to roughly 780,000 to 400,000 years; the original fossils were lost in 1941 while being evacuated during the Japanese invasion, and only casts and records survive. Stone tools at Shangchen in Shaanxi have been dated to 2.12 million years, which if correct would push the dispersal earlier still, and which is contested. At the young end, the Ngandong terrace in Java has been dated to about 117,000 to 108,000 years, meaning Homo erectus was alive while our own species was spreading.

Around 1.76 million years ago, at Kokiselei 4 in West Turkana, a new technology appears: the Acheulean handaxe. Unlike an Oldowan flake, a handaxe has imposed form, worked bifacially to a symmetrical teardrop that exists in the maker's head before it exists in the stone. What is strange is what happens next, which is almost nothing. The basic design persists with modest change for well over a million years across three continents, which is a stability no later human technology comes close to matching, and nobody has a fully satisfying explanation for it.

Key idea: Hominins left Africa by 1.85 million years ago carrying simple Oldowan tools, and the Dmanisi sample's extreme variation has been used to argue that early Homo comprised fewer species than commonly recognized, an extension most specialists reject.

What a brain costs

Brain tissue is metabolically expensive at rest, far more so than muscle. In an adult human the brain is about two percent of body mass and consumes roughly twenty to twenty-five percent of resting metabolic rate. In a newborn the figure is around sixty percent. Christopher Kuzawa's team showed in 2014 that the peak comes in mid-childhood, at about five years old, when the brain takes something like two thirds of resting metabolism, and that childhood body growth slows almost exactly when brain energy use peaks. The bill has to be paid from somewhere.

Leslie Aiello and Peter Wheeler proposed the expensive tissue hypothesis in 1995: total metabolic rate in primates is roughly what body size predicts, so a larger brain must be offset by a reduction in some other expensive organ, and the candidate is the gut. Human guts are about sixty percent the size predicted for a primate of our mass, with a small colon and a long small intestine, exactly what a high-quality, easily digested diet permits. The hypothesis elegantly links meat eating, tool use, gut reduction, and brain expansion.

Then Ana Navarrete, Carel van Schaik, and Karin Isler tested it properly. Publishing in Nature in 2011 with data on brain and organ masses from more than a hundred mammal species, they found no negative correlation between brain size and gut size. What did correlate negatively with brain size was fat storage, suggesting a trade-off between building a brain and banking energy against shortfall. Herman Pontzer's comparative energetics work added another piece: humans have evolved a higher total daily energy expenditure than other apes, so we did not only reallocate a fixed budget, we enlarged it. The current picture is that better diet, cooking, cooperative childcare, and an accelerated metabolism together paid for the brain, and that gut reduction is a consequence rather than the mechanism.

A parallel argument concerns what the brain was for. Robin Dunbar's social brain hypothesis reported that neocortex ratio predicts group size across primates and extrapolated a characteristic human group of about 150. Later analyses with larger samples and better phylogenetic controls have weakened it: Alex DeCasien and colleagues found in 2017 that diet, specifically frugivory, predicts primate brain size better than any measure of social complexity. Dunbar's number remains widely quoted and is not well supported.

Key idea: Brain tissue consumes twenty to twenty-five percent of adult resting metabolism, and the expensive tissue hypothesis that gut reduction paid for it failed a broad comparative test, leaving diet quality, cooking, cooperative care, and raised total expenditure as the working explanation.

Fire and cooking: a hypothesis at odds with its archaeology

Richard Wrangham's cooking hypothesis is the boldest version of the dietary argument. He proposes that control of fire and habitual cooking began with Homo erectus around 1.8 to 1.9 million years ago and explains the whole erectus package at once: smaller teeth, smaller gut, larger brain, and a species that sleeps on the ground rather than in trees. The biological logic is good. Cooking gelatinizes starch and denatures protein, raising the net energy actually extracted; Rachel Carmody's feeding experiments showed measurable weight differences in animals fed cooked versus raw versions of the same food. Cooking detoxifies, kills pathogens, and softens. Chimpanzees spend roughly four to seven hours a day chewing. Humans spend under one.

The archaeology does not cooperate. The earliest widely accepted evidence of controlled fire is at Wonderwerk Cave in South Africa, where Francesco Berna's team in 2012 identified burned bone and ashed plant material about thirty meters inside the cave at roughly one million years, using micromorphology and infrared spectroscopy to rule out natural burning. Burned patches at Koobi Fora and Chesowanja around 1.5 million years are ambiguous and could be lightning or brush fire. Gesher Benot Ya'aqov in Israel, at about 790,000 years, has repeated hearth locations across levels, which is a strong signal of intent. And Wil Roebroeks and Paola Villa's 2011 survey of European sites concluded that habitual fire use only becomes common after about 400,000 to 300,000 years ago.

So there is a gap of roughly a million years between the hypothesis and its evidence. Wrangham's answer is that fire use at open-air sites would be nearly invisible after that much time, and that the biological evidence should be weighted. His critics answer that arguing from an absence you have explained away is not a test. This one is genuinely unresolved, and it is worth watching, because it is the clearest case in the course of a well-motivated biological argument running ahead of what the ground has produced.

Common misconceptions

• Homo erectus was a stooping brute. Its body proportions are essentially modern, built for efficient long-distance walking; only the skull is markedly different.
• Hominins left Africa once they had large brains and handaxes. The Dmanisi hominins left by 1.85 million years with small brains and Oldowan flakes.
• Peking Man's fossils are in a museum. The originals were lost in 1941 during evacuation; research depends on casts and the excavation records.
• Larger brains were paid for by smaller guts. The comparative test found no brain-gut trade-off across mammals; fat storage, not gut size, correlated negatively with brain size.
• Cooking is established at 1.9 million years. The biological argument is strong but the earliest solid archaeological fire is about one million years old, and habitual use appears far later.

Recap

• KNM-WT 15000 shows a modern body plan with long legs, a narrow pelvis, and a brain near 880 cubic centimeters at about 1.5 million years.
• Early Homo taxonomy remains unsettled, with arguments that habilis belongs in Australopithecus.
• Dmanisi at 1.85 to 1.77 million years documents an immediate dispersal with Oldowan tools, and its variation fuels a contested single-species argument.
• Acheulean handaxes appear at 1.76 million years and then change remarkably little for over a million years.
• The adult brain takes twenty to twenty-five percent of resting metabolism and about two thirds in a five-year-old, and the expensive tissue hypothesis did not survive comparative testing.
• Cooking would explain the erectus package, but controlled fire is only solid from about one million years and habitual from about 400,000.

Sources

1. Smithsonian National Museum of Natural History. (n.d.). Homo erectus. Human Origins Program. humanorigins.si.edu
2. Lordkipanidze, D., et al. (2013). A complete skull from Dmanisi, Georgia, and the evolutionary biology of early Homo. Science, 342(6156), 326-331. doi.org
3. Aiello, L. C., & Wheeler, P. (1995). The expensive-tissue hypothesis. Current Anthropology, 36(2), 199-221. doi.org
4. Berna, F., et al. (2012). Microstratigraphic evidence of in situ fire in the Acheulean strata of Wonderwerk Cave, South Africa. PNAS, 109(20), E1215-E1220. doi.org
5. Roebroeks, W., & Villa, P. (2011). On the earliest evidence for habitual use of fire in Europe. PNAS, 108(13), 5209-5214. doi.org

The big picture

In August 1856, quarrymen clearing a small limestone cave in the Neander valley near Dusseldorf shoveled out about sixteen bones and threw them onto the spoil heap. A local schoolteacher and naturalist named Johann Carl Fuhlrott collected what survived: a thick skullcap with a heavy continuous brow, some limb bones, part of a pelvis. He was convinced they were ancient and human, and in 1857 the anatomist Hermann Schaaffhausen published them with him. In 1864 the Irish geologist William King gave them a species name, Homo neanderthalensis, the first extinct human species ever named. Rudolf Virchow, the most powerful pathologist in Europe, pronounced them a modern man deformed by rickets and old injuries, and much of the profession agreed.

Then in 1908 a nearly complete skeleton came out of a small cave at La Chapelle-aux-Saints in France, and Marcellin Boule spent five years describing it. His reconstruction gave the world its Neanderthal: knees permanently bent, head thrust forward on a slouched neck, a divergent grasping big toe, a shambling creature clearly off the line to us. The individual Boule worked from was old and had severe osteoarthritis in his spine, hip, and jaw, and had lost most of his teeth years before death. In 1957 William Straus and Alec Cave reexamined the skeleton and concluded that a healthy Neanderthal, given a shave and a suit, would draw no attention on a New York subway.

This lesson is about what a century of correcting Boule produced, and about a second archaic population that nobody suspected existed until a genome was sequenced from a fragment of finger bone.

What a Neanderthal was

Neanderthals occupied Europe and western Asia from at least 430,000 years ago, if the Sima de los Huesos people in northern Spain are counted as their ancestors, until roughly 40,000 years ago. Their bodies were built for cold and for high energy expenditure.

Feature	Neanderthal condition	Interpretation
Cranial capacity	About 1,200 to 1,750 cubic centimeters, mean near 1,450	At or above the modern human mean of roughly 1,350, in a larger, heavier body
Skull shape	Long and low with an occipital bun at the rear	Contrasts with the globular modern braincase
Face	Midfacial projection, huge nasal aperture, no chin	Possibly related to conditioning cold dry air and to heavy anterior tooth use
Brow	Double-arched continuous supraorbital torus	A diagnostic feature distinguishing them from modern humans
Jaw	Retromolar space behind the last molar	Reflects a long face and forward-set dentition
Body	Wide trunk, short forearms and lower legs, thick cortical bone	Consistent with Bergmann's and Allen's rules for cold climates

Estimates of their daily energy requirement run from about 3,300 to 5,000 kilocalories, which is athlete territory sustained for a lifetime. That number is worth holding, because any explanation of their disappearance has to account for a population that needed a great deal of food every single day.

Key idea: Neanderthal brains averaged at or above the modern human mean, and their bodies show cold-climate proportions and extremely high energy requirements.

What they made and ate

Their toolkit, the Mousterian, is built around the Levallois technique, and it is worth understanding because it is a direct window onto planning. In freehand flaking you strike a core and take what comes off. In Levallois you first shape the core over many preparatory removals, trimming its edges and convexity, so that a single final blow detaches a flake whose size and shape were determined before it existed. The knapper is working several steps ahead toward an object that is not yet there. Neanderthals also hafted stone points to wooden shafts using birch bark tar, which requires heating bark in a low-oxygen environment; lumps of the tar survive from Konigsaue in Germany. How difficult that production was is argued about, and a 2019 set of experiments showed usable tar can be made by a fairly simple rolling method, which lowers the inference about cognitive complexity somewhat.

They hunted large game at close range. Skeletal trauma is common, and a 1995 study comparing injury patterns to those of rodeo riders became famous; a 2018 reanalysis by Judith Beier's team, using a much larger sample and controlling for preservation, found Neanderthal cranial trauma rates statistically indistinguishable from those of contemporary modern humans, which deflates the popular image of uniquely violent lives. Dental calculus has yielded starch grains from cooked plants at Shanidar and Spy, and at El Sidron in Spain calculus contained yarrow and camomile, bitter plants with little nutritional value, which the excavators suggested might be medicinal and which others read as incidental.

The symbolic question, case by case

Whether Neanderthals thought symbolically is the field's longest-running argument, and it is best approached one piece of evidence at a time rather than as a yes-or-no verdict.

• Bruniquel Cave, France. The strongest case. More than 400 broken stalagmite segments arranged into two rings and several heaps, 336 meters inside a cave, with traces of fire on them, dated by uranium series to about 176,500 years ago. Nobody argues these assembled themselves, and no modern humans were in Europe then.
• Burial. Deliberate interment is accepted at several sites on the basis of body position and grave cuts. The Shanidar 4 flower burial, based on pollen clumps around a body, was challenged in 1999 when Jeffrey Sommer argued that burrowing rodents cache flower heads; excavation of a further individual, Shanidar Z, from 2018 onward has revived the argument for intentional deposition at that spot without settling the flowers.
• Pigment and ornament. Perforated and pigment-stained marine shells from Cueva de los Aviones in Spain, around 115,000 to 120,000 years old, and eagle talons cut and polished at Krapina in Croatia around 130,000 years, both predate modern humans in those regions.
• Cave art. In 2018 Dirk Hoffmann's team published uranium-thorium dates on carbonate crusts overlying red marks at three Spanish caves, giving minimum ages above 64,800 years, before modern humans arrived. Ludovic Slimak and others attacked the method, arguing that carbonate systems are open and that the samples may not date what they appear to date. This case is genuinely unresolved, and it is the one most often reported as settled.

Key idea: The evidence for Neanderthal symbolic and complex behavior ranges from essentially unassailable, as at Bruniquel, to actively disputed, as with the Spanish cave art dates, and the honest position tracks each case separately.

Denisovans: a population found in a genome

In 2008, excavators at Denisova Cave in the Altai Mountains of southern Siberia recovered the tip of a finger bone from a young girl. It was unremarkable to look at. In 2010 Johannes Krause and colleagues sequenced its mitochondrial genome and found a lineage that diverged from both modern humans and Neanderthals far earlier than they diverged from each other, and later that year David Reich's team published the nuclear genome. A previously unknown human population had been identified, named, and characterized without a single diagnostic skull. Nothing like that had ever happened before.

The cave kept producing. In 2018 Viviane Slon's team published Denisova 11, a bone fragment from a girl of about thirteen whose mother was a Neanderthal and whose father was a Denisovan: not distant admixture, but a first-generation offspring of two populations. Finding one such individual among the small number of archaic genomes sequenced implies that these encounters were not rare.

Morphology followed slowly. In 2019 a mandible from Baishiya Karst Cave on the Tibetan Plateau, at 160,000 years, was assigned to Denisovans on the basis of a single distinguishing position in its collagen peptides, later supported by mitochondrial DNA recovered from the cave sediments. In 2025, protein and mitochondrial evidence linked the large Harbin cranium from northeastern China, described in 2021 as Homo longi, to the Denisovan population, which if it holds finally gives the group a face. That work is recent and the field is still absorbing it.

Their genetic legacy is large in one part of the world. Papuans and Aboriginal Australians carry roughly three to five percent Denisovan ancestry, and analyses indicate more than one distinct introgression event from populations related to but not identical with the Altai Denisovans. The best-characterized functional case is EPAS1, a gene affecting the response to low oxygen, whose high-altitude variant in Tibetans was shown in 2014 to have been inherited from Denisovans. A population that vanished tens of thousands of years ago supplied the allele that makes life at 4,000 meters tolerable.

Key idea: Denisovans were defined from a genome rather than from anatomy, they interbred with Neanderthals directly, and their DNA supplied the high-altitude EPAS1 variant carried by Tibetans today.

Species, or not

Since Neanderthals, Denisovans, and modern humans all produced fertile offspring together, a strict biological species concept would treat them as one species with three regional populations, and some researchers use Homo sapiens neanderthalensis for exactly that reason. Others keep the separate species names, noting that the introgressed archaic DNA in living genomes shows signs of having been selected against, especially on the X chromosome and near genes expressed in testes, which is the pattern expected when hybrids have reduced fertility. Both readings are defensible from the same data, and the argument is really about how much reproductive leakage a species boundary tolerates.

Why they disappeared

Thomas Higham's dating program, published in 2014 and based on ultrafiltered collagen from forty European sites, placed the last Neanderthals between roughly 41,000 and 39,000 years ago, with an overlap with modern humans in Europe of up to about five thousand years. Several factors are on the table, and the evidence favors a combination rather than a single cause.

• Small populations and inbreeding. The high-coverage Altai genome showed that individual's parents were as closely related as half-siblings, and long runs of homozygosity appear in other Neanderthal genomes. Effective population sizes are estimated in the low thousands across their entire range.
• Climate instability. The period includes abrupt Dansgaard-Oeschger oscillations and Heinrich events that fragmented habitats, which small dispersed populations absorb poorly.
• Competition and demographic swamping. Modern human groups appear to have sustained higher population densities and wider social networks, so even a modest, sustained difference in birth or death rates removes a small population over centuries.
• Assimilation. Interbreeding absorbed some Neanderthal ancestry into the incoming population, which is why one to two percent of the genome of every person of non-African descent is Neanderthal in origin.

The last point deserves the last word. Whatever happened, they did not vanish entirely. Their DNA is being read right now by the machinery of your own cells.

Common misconceptions

• Neanderthals were stooped and dim. Boule's reconstruction was based on an arthritic individual and was corrected in 1957; their brains averaged at or above the modern mean.
• Neanderthals were our ancestors. They are a sister lineage; modern humans descend from African populations, with a small archaic contribution from interbreeding.
• Neanderthals had no symbolic behavior. Bruniquel's 176,000-year-old constructions, pigment-stained shells, and worked eagle talons all precede modern humans in Europe.
• Neanderthals definitely painted cave walls. The 2018 uranium-thorium dates from Spain are contested on methodological grounds and are not established.
• Denisovans are known from many skeletons. They were defined from a genome, and the morphological record remains a mandible, fragments, and the recently linked Harbin cranium.

Recap

• The 1856 Feldhofer bones were dismissed as pathological, and Boule's 1911 reconstruction gave the world a stooping Neanderthal that took until 1957 to correct.
• Neanderthal brains averaged near 1,450 cubic centimeters, their bodies were cold-adapted, and their daily energy needs ran to several thousand kilocalories.
• Levallois flaking required planning several removals ahead toward a predetermined flake, and hafting with birch bark tar is documented.
• Symbolic evidence must be judged case by case: Bruniquel is secure, the Spanish cave art dates are contested.
• Denisovans were identified from a genome in 2010, produced a first-generation hybrid with a Neanderthal in Denisova 11, and gave Tibetans their EPAS1 variant.
• Neanderthals disappeared between about 41,000 and 39,000 years ago, most plausibly through a combination of small population size, climate instability, competition, and assimilation.

Sources

1. Smithsonian National Museum of Natural History. (n.d.). Homo neanderthalensis. Human Origins Program. humanorigins.si.edu
2. Higham, T., et al. (2014). The timing and spatiotemporal patterning of Neanderthal disappearance. Nature, 512, 306-309. doi.org
3. Jaubert, J., et al. (2016). Early Neanderthal constructions deep in Bruniquel Cave in southwestern France. Nature, 534, 111-114. doi.org
4. Slon, V., et al. (2018). The genome of the offspring of a Neanderthal mother and a Denisovan father. Nature, 561, 113-116. doi.org
5. Huerta-Sanchez, E., et al. (2014). Altitude adaptation in Tibetans caused by introgression of Denisovan-like DNA. Nature, 512, 194-197. doi.org

The big picture

In 1961, miners cutting barite out of a hillside at Jebel Irhoud in Morocco found a human skull. It went into the literature as a Neanderthal, then as an African variant, and was guessed at around 40,000 years old. Jean-Jacques Hublin reopened the site in 2004, and his team spent years removing the mine spoil and excavating what was left in place. They recovered remains of at least five individuals and, crucially, flint tools that had been heated in a fire, which meant the burnt flints could be dated by thermoluminescence. The answer, published in 2017, was 315,000 years, plus or minus 34,000.

Look at the face of the Irhoud skull and it is essentially ours: small, tucked under the braincase, with a modest brow and a delicate jaw. Look at the braincase from the side and it is not ours at all, long and low rather than globular. Our species did not arrive assembled. The face came first by a very long way, and the round skull that houses your brain took another quarter of a million years to appear.

This lesson covers where Homo sapiens came from, what interbreeding with archaic populations left in living genomes, when behavior we recognize as ours shows up in the record, and how the species reached every habitable continent.

Two models, and what replaced both

For most of the late twentieth century two accounts competed. Multiregional evolution, argued by Franz Weidenreich and later Milford Wolpoff and Alan Thorne, held that Homo erectus populations across the Old World evolved together into modern humans, connected by enough gene flow to remain one species and showing regional continuity in features. Recent African origin, argued by Chris Stringer and Peter Andrews in 1988, held that modern humans arose in Africa and spread outward, replacing the archaic populations they met.

The 1987 mitochondrial study by Cann, Stoneking, and Wilson pushed hard toward Africa: the deepest mitochondrial lineages were African, and the common ancestor of all living mitochondrial lineages lived roughly 200,000 years ago. Recall from Lesson 4 that this is a statement about one non-recombining locus, not a claim that one woman was alive alone. The nuclear genomes that followed pointed the same direction, and Africa's greater overall genetic diversity is consistent only with an African origin.

What has emerged since is neither original model. Eleanor Scerri and colleagues argued in 2018 that the fossil, archaeological, and genetic evidence fits a pan-African picture: populations subdivided across the continent by deserts, forests, and rivers, semi-isolated for long stretches, exchanging genes intermittently, with modern features assembling in mosaic fashion in different places at different times. The fossils support this. Jebel Irhoud in Morocco at 315,000 years, Florisbad in South Africa at about 259,000, and Omo Kibish in Ethiopia, redated in 2022 to more than 233,000 years, are separated by the length of a continent. In 2023 Aaron Ragsdale's team fitted African genome data and found that a weakly structured stem, with two or more long-separated African populations merging, explained the data better than a single origin. This is the active frontier and it is not settled.

Key idea: Homo sapiens originated in Africa but not from a single small region: current evidence favors structured, semi-isolated African populations that exchanged genes while modern features accumulated piecemeal.

The interbreeding, and what it left in you

The 2010 draft Neanderthal genome delivered a result nobody's model had predicted: people outside Africa carry roughly one to two percent Neanderthal ancestry, distributed differently in each individual so that collectively living humans retain a large fraction of the Neanderthal genome in pieces. Denisovan ancestry runs at three to five percent in Papuans and Aboriginal Australians and at lower levels across East and South Asia. Africans carry a small amount of Neanderthal ancestry too, around a third of a percent, mostly from populations moving back into Africa.

Dating it used the recombination logic from Lesson 4. The 45,000-year-old femur from Ust-Ishim in Siberia, sequenced in 2014, carried Neanderthal segments that were still long, and the length distribution put the admixture roughly 50,000 to 60,000 years before that individual lived. A jaw from Pestera cu Oase in Romania, around 40,000 years old, carried segments so long that the person had a Neanderthal ancestor four to six generations back, though that lineage left no detectable descendants.

Introgressed variant	Source	Effect
EPAS1 haplotype	Denisovan	High-altitude oxygen response in Tibetans
TLR immune gene cluster	Neanderthal	Altered innate immune response to pathogens
Chromosome 3 haplotype	Neanderthal	Raised risk of severe respiratory illness, identified during the COVID-19 pandemic
OAS gene cluster on chromosome 12	Neanderthal	Protective against some viral infections
Skin and hair keratin genes	Neanderthal	Contributions to pigmentation and hair phenotypes

The pattern makes sense: variants dealing with local pathogens, climate, and altitude were the useful inheritance, because archaic populations had been adapting to those places for hundreds of thousands of years while our ancestors had just arrived.

When did behavior become recognizably ours?

The older account located a threshold in Europe: a sudden Upper Paleolithic revolution around 50,000 to 40,000 years ago, with art, ornament, blade technology, bone tools, and long-distance exchange appearing together. Sally McBrearty and Alison Brooks attacked that framing in 2000 in a paper whose title said the revolution had not happened. Their argument was that every element appears earlier and separately in the African Middle Stone Age, and that the European pattern is an artifact of where archaeologists had been digging for a hundred and fifty years.

The African evidence has since become hard to dismiss. At Olorgesailie in Kenya, pigment processing and obsidian transported over long distances are documented by 320,000 years ago. At Pinnacle Point in South Africa, ochre use and the deliberate heat treatment of silcrete to improve its flaking properties go back about 164,000 years, which is pyrotechnology applied to raw material. At Blombos Cave, engraved ochre blocks with cross-hatched patterns date to 100,000 to 73,000 years, perforated shell beads to about 75,000, and in 2018 Christopher Henshilwood's team published a cross-hatched drawing made with an ochre crayon on a silcrete flake at 73,000 years. Diepkloof has engraved ostrich eggshell containers around 60,000 years old.

Figurative art outside Africa is now also older and further east than the European story allowed. The Lion Man carved from mammoth ivory at Hohle Fels in Germany dates to about 40,000 years, alongside bone flutes. Chauvet's paintings run to about 36,000. But cave paintings of animals on Sulawesi and Borneo are at least 45,000 years old, and a narrative scene from Sulawesi published in 2024 was dated to at least 51,200 years, older than anything in Europe.

A caution worth carrying: behavioral modernity is defined by a checklist that European archaeologists wrote from European sites, and every item on it preserves unevenly. Beads survive; songs, kinship systems, and stories do not.

Key idea: The elements of modern behavior appear separately and gradually across Africa from at least 320,000 years ago rather than arriving as a package in Europe, and the apparent European revolution largely reflects where excavation was concentrated.

The spread

Modern humans were in the Levant early. Misliya Cave in Israel yielded a maxilla dated to 194,000 to 177,000 years, and Skhul and Qafzeh have burials around 120,000 to 90,000 years. Those early excursions do not appear to have left descendants outside Africa; the genetic evidence points to a later expansion, roughly 60,000 to 50,000 years ago, from which all non-African populations descend.

• Sahul. Australia and New Guinea were joined at low sea level, but reaching them still required crossing open water. Madjedbebe in northern Australia was dated by Chris Clarkson's team in 2017 to about 65,000 years, a date contested by researchers who prefer around 50,000 and who question whether artifacts moved downward through sandy sediment.
• Eurasia. Bacho Kiro in Bulgaria and Zlaty kun in Czechia give modern humans in Europe around 45,000 years ago. A claim of modern human presence at Grotte Mandrin in France at about 54,000 years, published in 2022, is disputed on the identification of a single deciduous tooth.
• Siberia and Beringia. The Yana site above the Arctic Circle documents occupation at 32,000 years. Genetic analyses place the separation of ancestral Native American populations from Siberian relatives roughly 23,000 to 19,000 years ago, during or just after the glacial maximum.
• The Americas. Monte Verde in Chile, verified at about 14,500 years, broke the old Clovis-first consensus. Footprints at White Sands in New Mexico were published in 2021 with dates of 23,000 to 21,000 years and supported by independent radiocarbon and luminescence work in 2023; the age remains contested.
• Remote Oceania. The last places on Earth to be settled. Lapita seafarers reached the western Pacific about 3,000 years ago, and Aotearoa was settled around 1300 of the common era, which is more recent than the construction of Notre-Dame.

That expansion left a genetic signature you can read today. Each founding group carried a subset of the diversity of the population it left, so genetic diversity declines steadily with distance travelled from Africa along plausible land routes, a pattern documented by Sohini Ramachandran and colleagues and refined since. It is called a serial founder effect, and it explains why African populations retain more genetic variation than all non-African populations combined. Hold that fact. The next module builds an argument on it.

Common misconceptions

• Homo sapiens appeared suddenly in East Africa about 200,000 years ago. Jebel Irhoud at 315,000 years is in Morocco, and current models describe structured populations across the whole continent.
• Modern humans replaced archaic populations without interbreeding. Non-Africans carry one to two percent Neanderthal ancestry, and Oceanian populations carry three to five percent Denisovan ancestry.
• Art and symbolism began in Europe. Ochre engraving and shell beads in Africa are tens of thousands of years older, and Sulawesi's figurative art predates Chauvet.
• The first modern humans out of Africa are the ancestors of living non-Africans. The Levantine populations of 120,000 years ago appear to have left no lasting descendants; the ancestral expansion was later.
• The Americas were settled about 13,000 years ago by the Clovis people. Monte Verde is securely older, and the White Sands footprints may push occupation back to before the glacial maximum.

Recap

• Jebel Irhoud at 315,000 years shows a modern face with an archaic elongated braincase, so our anatomy assembled in mosaic fashion.
• Multiregionalism and simple replacement have both given way to a pan-African structured model with ongoing debate about its details.
• Non-Africans carry one to two percent Neanderthal ancestry; Oceanians add three to five percent Denisovan, including the functional EPAS1 and immune variants.
• Segment lengths in the Ust-Ishim and Oase genomes dated the Neanderthal admixture to roughly 50,000 to 60,000 years ago.
• Modern behavior appears piecemeal in Africa from at least 320,000 years, undercutting the European revolution model.
• Serial founder effects during the expansion left genetic diversity declining with distance from Africa, a fact Module 6 depends on.

Sources

1. Hublin, J.-J., et al. (2017). New fossils from Jebel Irhoud, Morocco and the pan-African origin of Homo sapiens. Nature, 546, 289-292. doi.org
2. Scerri, E. M. L., et al. (2018). Did our species evolve in subdivided populations across Africa, and why does it matter? Trends in Ecology and Evolution, 33(8), 582-594. doi.org
3. Clarkson, C., et al. (2017). Human occupation of northern Australia by 65,000 years ago. Nature, 547, 306-310. doi.org
4. Smithsonian National Museum of Natural History. (n.d.). Homo sapiens. Human Origins Program. humanorigins.si.edu
5. Britannica. (2024). Homo sapiens. britannica.com

The big picture

In 2000 Nina Jablonski and George Chaplin published a study that did something obvious that nobody had done properly. They took skin reflectance measurements from indigenous populations at more than fifty locations, taken with a spectrophotometer on the inner upper arm where sun exposure is least, and plotted them against ultraviolet radiation data collected by NASA's Total Ozone Mapping Spectrometer satellite. The relationship was extraordinarily tight. Skin reflectance tracked annual UV exposure across the globe with a correlation above 0.9.

They also found something smaller and stranger. In nearly every population measured, women were slightly lighter-skinned than men.

Those two results are the entry point for this lesson, which is about how living human populations vary biologically, why they vary, and what shape that variation takes. Skin color is the worked example, and then altitude, cold, and diet each supply a case where the same problem was solved several different ways by populations that never met.

Skin color, worked properly

Skin pigmentation comes from melanin, produced by melanocytes and packaged into organelles called melanosomes. Here is a fact that surprises most people: populations do not differ much in how many melanocytes they have. They differ in the size, number, packaging, and persistence of melanosomes, and in the ratio of brown-black eumelanin to red-yellow pheomelanin.

The selective pressure has two opposing arms, which is why the outcome is a gradient rather than a maximum.

• Too little melanin near the equator. Ultraviolet B radiation penetrating unprotected skin degrades folate circulating in surface blood vessels. Folate deficiency in early pregnancy causes neural tube defects and also impairs spermatogenesis. Both hit reproduction directly, which makes this a powerful selective force.
• Too much melanin at high latitude. Ultraviolet B is also the trigger that converts 7-dehydrocholesterol in the skin into the precursor of vitamin D. In northern Europe in winter there is not enough UVB to make vitamin D through heavily pigmented skin. Deficiency produces rickets, and a rachitic pelvis makes childbirth dangerous, which again bears directly on reproduction.

The Jablonski and Chaplin sex difference fits this model: women need more calcium and vitamin D during pregnancy and lactation, so the vitamin D arm of the trade-off presses slightly harder on them. Skin cancer, incidentally, is a real cost of depigmentation but a weak selective force, because most cases occur after reproduction.

The evolutionary sequence runs like this. Chimpanzees have pale skin under dark hair. When the hominin lineage lost most of its body hair, plausibly between 1.6 and 1.2 million years ago, dark skin became necessary in equatorial Africa. The genetic signature of this survives: the MC1R gene shows essentially no functional variation in African populations, the pattern produced by strong purifying selection, while it is highly variable in Europeans and East Asians, where the constraint relaxed.

Then comes the part that matters most for the next lesson. Depigmentation outside Africa happened more than once, by different genetic routes. In Europe the major contributors include SLC24A5, whose derived variant is near fixation in European populations, along with SLC45A2 and TYR. In East Asia lighter pigmentation involves largely different loci, including distinct variants of OCA2. These are separate solutions to the same problem. Light skin in Sweden and light skin in northern China are not shared inheritance, so they say nothing about relatedness between those populations.

Ancient DNA has sharpened the timing. Iain Mathieson's 2015 analysis of hundreds of ancient European genomes found that the derived pigmentation alleles were not common in Europe until roughly the last eight thousand years, rising as farming spread. The 2018 reconstruction of Cheddar Man, a British individual from about ten thousand years ago, as dark-skinned and blue-eyed followed from this, though pigmentation prediction from ancient genomes carries genuine uncertainty and that reconstruction was reported far more confidently in the press than in the underlying analysis.

One more correction. Pigmentation variation within Africa is greater than between any two other continents. Sarah Tishkoff's group showed in 2017 that variants at MFSD12, DDB1, OCA2, and HERC2 produce a wide range of African skin tones, that some lightening alleles are ancient and were carried out of Africa rather than invented outside it, and that San populations in southern Africa are on average considerably lighter than populations in East Africa. The sentence African people have dark skin is not a description of Africa.

Key idea: Skin color is a gradient produced by opposing selection on folate protection and vitamin D synthesis, depigmentation evolved independently in Europe and East Asia through different genes, and within-Africa pigmentation variation exceeds that between other continents.

Clines, and the failure of concordance

A cline is a gradual geographic gradient in the frequency of a trait or allele. Skin reflectance is clinal along latitude. So is the frequency of the B blood group allele, which grades across Eurasia. So is average nasal aperture width, which correlates with the temperature and humidity of ancestral climates.

Now do something the eye does not do naturally. Draw the map of one cline, then draw the map of another on tracing paper, then lay them over each other. The boundaries do not match. Skin pigmentation gradients run mostly with latitude. Lactase persistence follows dairying history and is high in northern Europe, in East African pastoralist groups, and in parts of Arabia, while low in between. Sickle-cell allele frequency follows the historical distribution of falciparum malaria, so it is high in West Africa, in Greece, in southern India, and in Saudi Arabia. Body proportions follow thermal environment. Each trait has its own map because each responded to its own selective pressure, and the pressures do not have the same geography.

This is called the failure of concordance, and it is the single most important structural fact about human biological variation. Traits do not cluster into packages. Knowing someone's skin color tells you very little about their blood group, their lactase status, or their nasal shape.

Altitude: one problem, three answers

Above about 2,500 meters, the partial pressure of oxygen falls far enough to reduce arterial oxygen saturation, impair work capacity, and, historically, reduce birth weight and infant survival. Three human populations have lived at altitude long enough to adapt, and all three did it differently.

Population	Approximate time at altitude	Physiological pattern	Genetic basis
Tibetan	Tens of thousands of years, with occupation of the plateau documented by 30,000 years	Normal or low hemoglobin, elevated resting ventilation, high exhaled nitric oxide, wide blood vessels	EPAS1, inherited from Denisovans, plus EGLN1 variants
Andean	Roughly 11,000 years	Elevated hemoglobin concentration, large lung volumes and chest dimensions	Different EGLN1 variants and other loci
Ethiopian highland	Several thousand years at minimum	Normal hemoglobin with normal oxygen saturation, mechanism still being characterized	Loci largely distinct from both the Tibetan and Andean sets

The Tibetan and Andean solutions are close to opposite. Tibetans do not thicken their blood; Andeans do. Thick blood carries more oxygen and also raises the risk of chronic mountain sickness, which is common in the Andes and rare in Tibet. Two populations, one stressor, two strategies, and a third in Ethiopia that resembles neither. If human adaptation followed continental groupings, this could not happen.

Key idea: Tibetan, Andean, and Ethiopian highlanders solved the same oxygen problem through different genes and opposite physiologies, which is direct evidence that adaptation is local and population-specific rather than continental.

Cold, and the costs that come with adaptation

Two ecological rules that apply across mammals apply to humans. Bergmann's rule holds that body mass relative to surface area increases in colder climates, conserving heat; Allen's rule holds that limb and appendage length decreases. Compare the stocky, short-limbed body proportions typical of Arctic populations with the tall, long-limbed proportions typical of Nilotic populations in South Sudan, and you are looking at these rules expressed in human bodies.

Some of the genetics is now identified, and one case is a warning against romanticizing adaptation. The CPT1A gene has an Arctic variant that reaches near fixation in several northern populations and alters how fat is metabolized. It appears advantageous on a very high fat, low carbohydrate diet. It also raises the risk of hypoglycemia in infants during illness or fasting, and it is associated with elevated infant mortality in Arctic communities today. Selection does not produce good outcomes. It produces outcomes with a net reproductive advantage in the environment where they arose, and the costs are real people. Separately, a region containing TBX15 and WARS2, associated with body fat distribution, appears in Inuit populations as introgressed archaic DNA of probable Denisovan origin.

Diet, and lactase persistence as recent evolution

Almost all mammals stop producing lactase after weaning. In some human populations the enzyme keeps being produced into adulthood, a trait called lactase persistence. The gene is LCT, but the causal variants are regulatory, sitting in an enhancer inside a neighboring gene, MCM6, and switching lactase expression back on.

Here is the part that matters. There is not one such variant. Europe has one, usually written as the minus 13910 T allele. Sarah Tishkoff's team documented in 2007 that East African pastoralists carry a different one at minus 14010, and populations in Sudan and Arabia carry others at minus 13907 and minus 13915. At least four independent mutations, in populations that kept cattle, camels, or goats, all producing the same phenotype. This is convergent evolution caught in the act, in humans, in the last several thousand years, and the selection coefficients estimated for these alleles are among the strongest measured anywhere in the human genome.

The timing contains a genuine puzzle. Milk residues on pottery show dairying in Anatolia and Europe by around eight thousand years ago, but ancient DNA shows the persistence allele was still rare in Bronze Age Europe, meaning people processed and consumed milk for thousands of years before most of them could digest lactose as adults. A large analysis published in 2022 argued that the advantage was not routine nutrition, since fermenting milk into cheese or yoghurt removes most lactose anyway, but survival during famine and disease, when a lactose-intolerant person drinking milk suffers diarrhea and dehydration that a healthy person merely finds unpleasant. The selection was episodic, driven by crises.

Other dietary adaptations follow the same pattern of local, recent, specific change. Greenlandic Inuit carry FADS variants affecting fatty acid processing that also influence height and weight, documented by Matteo Fumagalli's team in 2015. Bajau divers in Indonesia have spleens roughly fifty percent larger than neighboring populations, associated with a PDE10A variant, supporting long breath-hold dives. Populations in the Atacama carry AS3MT variants improving arsenic metabolism. And falciparum malaria alone has driven at least four separate protective adaptations: the sickle allele, several thalassemias, glucose-6-phosphate dehydrogenase deficiency, and, against vivax malaria, the Duffy-null variant.

Four kinds of adjustment, which get confused constantly

When a body copes with an environment, there are four distinct mechanisms, operating on four different timescales. Confusing them produces most of the bad reasoning in this area.

Mechanism	Timescale	Reversible?	Example at altitude
Genetic adaptation	Generations	Only by further evolution	The Tibetan EPAS1 variant
Developmental adjustment	During growth	No, once growth is complete	Larger chest dimensions in someone raised at altitude from birth
Acclimatization	Hours to weeks	Yes	Increased breathing rate and red cell production after arriving
Cultural adaptation	Immediate	Yes	Coca leaf, supplemental oxygen, insulated housing, staged ascent

A lowlander who moves to La Paz will acclimatize within weeks and will never develop the chest capacity of someone who grew up there, let alone the Tibetan genetic pattern. Their children, raised at altitude, will develop that chest capacity without inheriting a single altitude allele. The most dramatic demonstration of developmental plasticity in humans is height: average adult height in the Netherlands rose by roughly twenty centimeters in about a century and a half, driven by nutrition and disease reduction rather than by any change in allele frequencies. Franz Boas's 1912 immigrant study, from Lesson 2, was measuring exactly this phenomenon.

Key idea: Genetic adaptation, developmental adjustment, acclimatization, and cultural adaptation operate on different timescales and different mechanisms, and a population difference in a trait is not evidence of a genetic difference until the other three are excluded.

Common misconceptions

• Dark-skinned people have more melanocytes. Melanocyte number is similar; populations differ in melanosome size, number, packaging, and persistence.
• Light skin evolved once and marks a related group of populations. European and East Asian depigmentation involved largely different genes, so it is convergent rather than shared.
• Africans are uniformly dark-skinned. Pigmentation variation within Africa exceeds that between other continents, and San populations are on average considerably lighter than East African populations.
• All high-altitude peoples adapt the same way. Tibetans do not raise hemoglobin, Andeans do, and Ethiopians resemble neither.
• Adaptation means improvement. The Arctic CPT1A variant helps on a high-fat diet and raises infant hypoglycemia risk; selection optimizes reproduction in a past environment, not welfare.
• Human evolution stopped with agriculture. Lactase persistence, altitude alleles, and arsenic tolerance are all products of the last ten thousand years.

Recap

• Skin reflectance tracks ultraviolet exposure closely, balancing folate protection against vitamin D synthesis, with women slightly lighter in nearly every population.
• Depigmentation evolved separately in Europe and East Asia, and most European pigmentation alleles rose to high frequency only in the last eight thousand years.
• Traits vary clinally and their gradients do not coincide, which is the failure of concordance.
• Tibetan, Andean, and Ethiopian highlanders show three distinct genetic and physiological solutions to hypoxia.
• At least four independent mutations produce lactase persistence in dairying populations, and the advantage may have been episodic survival during famine and illness.
• Genetic adaptation, developmental adjustment, acclimatization, and cultural adaptation must be distinguished before any population difference is called genetic.

Sources

1. Jablonski, N. G., & Chaplin, G. (2000). The evolution of human skin coloration. Journal of Human Evolution, 39(1), 57-106. doi.org
2. Crawford, N. G., et al. (2017). Loci associated with skin pigmentation identified in African populations. Science, 358(6365), eaan8433. doi.org
3. Tishkoff, S. A., et al. (2007). Convergent adaptation of human lactase persistence in Africa and Europe. Nature Genetics, 39, 31-40. doi.org
4. Fumagalli, M., et al. (2015). Greenlandic Inuit show genetic signatures of diet and climate adaptation. Science, 349(6254), 1343-1347. doi.org
5. National Human Genome Research Institute. (n.d.). Genomic variation fact sheets. National Institutes of Health. genome.gov

The big picture

In 1972 Richard Lewontin took data on 17 polymorphic genetic loci, mostly blood groups and serum proteins, from populations around the world, sorted those populations into the seven races then conventional in the literature, and asked a simple statistical question: where does human genetic variation actually sit? His answer was 85.4 percent within local populations, 8.3 percent among populations within a race, and 6.3 percent among races.

Fifty years of better data have moved those numbers only slightly and always in the same direction. Noah Rosenberg's team, typing 377 microsatellite markers in 1,056 individuals from 52 populations in 2002, found 93 to 95 percent of variation within populations. Take two people at random from the same village and then two people at random from opposite sides of the planet, and the second pair is only slightly more genetically different than the first.

This lesson works out what that fact does and does not establish, states the strongest objection to it accurately, applies the criteria zoologists use for subspecies in other mammals, and then asks what race actually is, given that it is plainly not that.

Why human genetic diversity is so low

The apportionment result is not a statistical accident. It follows from three things you have already met. Homo sapiens is a young species, with a long-term effective population size estimated around ten thousand, which is small. The expansion out of Africa proceeded by serial founder effects that stripped diversity rather than creating it. And gene flow has never stopped: human populations have exchanged mates at every scale continuously, which is exactly the force that prevents differentiation.

The comparison with our relatives makes the scale vivid. Nucleotide diversity in humans is roughly 0.1 percent. In chimpanzees and gorillas it is higher, and it is distributed differently: a single chimpanzee community in West Africa can carry more genetic variation than exists in all humans living outside Africa. We are, genetically, an unusually uniform large mammal.

Key idea: About 85 to 95 percent of human genetic variation lies within local populations, a consequence of our young species age, small ancestral population size, serial founder effects, and uninterrupted gene flow.

The strongest objection, stated properly

In 2003 the statistician A. W. F. Edwards published a paper titled Human genetic diversity: Lewontin's fallacy, and it deserves to be presented at full strength rather than waved away, because Edwards was right about the mathematics.

His point was that Lewontin analyzed loci one at a time. Any single locus is a poor classifier. But alleles at different loci are correlated with one another as a consequence of shared population history, and a classifier that uses many loci jointly exploits those correlations. With enough markers, individuals can be assigned to their continental population of origin with very high accuracy. Rosenberg's 2002 analysis demonstrated this concretely: running a clustering algorithm on the microsatellite data and asking for five clusters produced groups corresponding roughly to sub-Saharan Africa, Eurasia west of the Himalayas, East Asia, Oceania, and the Americas. So the low between-group variance and the high classification accuracy are both true.

What follows from that? Less than it appears, for four reasons.

• Assignability is not taxonomy. A classifier that works tells you structure exists. It does not tell you that the categories it recovers are natural biological units. You can classify people accurately by dialect, by surname, or by the ZIP code they grew up in, and none of those are biological taxa either.
• The clusters depend on how you sample and what you ask for. Clustering algorithms of the STRUCTURE family require the analyst to specify the number of clusters. Ask for two and you get two; ask for six and you get six, and the sixth splits out the Kalash of Pakistan, which no folk taxonomy of race has ever recognized. More importantly, Stephane Serre and Svante Paabo showed in 2004 that when individuals are sampled evenly along geographic gradients rather than from widely separated continental populations, the clusters dissolve into clines. Later work confirmed that the apparent discreteness largely reflects sampling gaps: oceans, deserts, and mountain ranges where nobody was sampled because nobody lives there.
• The clusters do not match folk racial categories. The deepest genetic divergences among living humans are inside Africa, between Khoe-San populations and everyone else, a separation on the order of 250,000 to 350,000 years. A folk category like Black therefore groups together the most genetically divergent people on Earth, while the category white splits a shallow, continuous gradient that runs without a break through the Middle East into South Asia. If you built categories that actually tracked the genetic structure, they would not resemble the racial categories any society uses.
• There are essentially no diagnostic markers. Almost all alleles are shared across all populations, differing only in frequency. There is no allele that all members of one folk racial category carry and no member of another does. Group-specific variants exist but are typically rare within their own group, so they classify nobody.

Key idea: Edwards was correct that multi-locus classification works, and that establishes that human population structure is real; it does not establish that the recovered clusters are biological races, because the clusters depend on sampling and cluster number, do not match folk categories, and rest on no diagnostic markers.

Applying the standard zoologists actually use

Biologists who name subspecies in other mammals use quantitative conventions. One is the seventy-five percent rule: at least three quarters of individuals in one subspecies must be distinguishable from essentially all individuals of the other. Another is a threshold on the fixation index, FST, which measures the proportion of genetic variance attributable to population subdivision. Recognized mammalian subspecies typically show FST values above 0.25 to 0.30. Chimpanzee subspecies, which are genuinely recognized, sit around 0.30.

Human continental groups come in around 0.05 to 0.15, depending on marker type. Alan Templeton applied the full set of criteria formally, in 1998 and again in 2013, and reached two conclusions. First, human populations fall far short of every quantitative threshold used for subspecies in other mammals. Second, the pattern of human variation is isolation by distance combined with long-range gene flow, which produces gradients, rather than the tree of separated lineages that a subspecies structure requires. That second point matters more than the first: it is not that humans just miss the cutoff, it is that the shape of the variation is the wrong shape.

Nothing about this argument is special pleading for our species. It is the same standard applied to wolves, to chimpanzees, and to leopards, run on human data, with the result that no human subspecies are recognized.

So what is race?

Race is a social classification with a documented history, and the clearest evidence for that is how the categories move. In the United States, Irish, Italian, and Jewish immigrants were treated as racially distinct from and inferior to Anglo-Americans in the nineteenth and early twentieth centuries and are now unproblematically white. The one-drop rule made a person with one African great-grandparent Black in Virginia while the same person would be classified differently in Brazil, where classification depends on appearance, and differently again in South Africa under apartheid law.

Two Supreme Court cases, decided three months apart, make the point sharper than any argument could. In Takao Ozawa v. United States, decided in November 1922, the Court denied naturalization to a Japanese applicant on the ground that white meant Caucasian, and Japanese people were not Caucasian. In United States v. Bhagat Singh Thind, decided in February 1923, an Indian applicant argued that he was in fact classified as Caucasian by the anthropology of the day. The Court agreed that he was, and denied him anyway, ruling that white must be understood in the sense of the common person rather than of the scientists. The category was whatever it needed to be.

Key idea: Racial categories have shifted repeatedly by law and custom while the underlying biology did not, which is the signature of a social classification rather than a natural kind.

Social does not mean biologically inert

Here is where careless versions of this argument go wrong. Saying race is not a biological taxonomy is not saying race has no biological consequences. Racism is an environmental exposure, and exposures have physiology.

• Weathering. Arline Geronimus's work documents that health measures in Black Americans deteriorate earlier across the life course in a pattern consistent with cumulative stress exposure. The Black-white gap in preterm birth in the United States persists after adjusting for income and education, and it is not explained by African ancestry: African-born immigrant women have birth outcomes closer to white American women than to Black American women, and their daughters' outcomes converge toward the Black American pattern.
• Instruments calibrated on the wrong people. A 2020 study in the New England Journal of Medicine found that pulse oximeters overestimate blood oxygen saturation roughly three times more often in Black patients than in white patients, so hypoxemia gets missed. This is a device calibration failure against skin pigmentation, and during the COVID-19 pandemic it affected who received treatment.
• Race correction in clinical algorithms. Kidney function estimates carried a race coefficient that raised the estimated filtration rate for Black patients, delaying referral for specialist care and transplant listing; a joint task force recommended removing it in 2021. Spirometry has carried similar corrections. A 2020 review catalogued a dozen such algorithms in routine use.
• Race as a bad proxy for ancestry. Sickle-cell trait tracks ancestry from malarial regions, which includes Greek, Turkish, Arabian, and Indian populations as well as West African ones. Screening by racial category both misses cases and reinforces the error. What is clinically useful is specific ancestry, family history, and actual exposure, none of which race reports reliably.

The claims about intelligence, stated and answered

Claims that racial groups differ innately in cognitive ability recur, and they fail on specific grounds rather than on distaste.

The central error is statistical. Heritability measures the proportion of variance within a population attributable to genetic variance in that population's environment. It carries no information about the causes of differences between populations. Lewontin's illustration remains the clearest: take one bag of genetically variable seed, plant half in rich soil and half in depleted soil. Within each pot, height differences are highly heritable. The difference between the pots is entirely environmental. High within-group heritability and a fully environmental between-group difference are perfectly compatible.

Empirically, population average test scores rose by roughly three points per decade across the twentieth century in many countries, the Flynn effect, a change far too fast for allele frequencies and larger than the gaps under discussion. Gaps between groups in the United States narrowed substantially over the same period. And polygenic scores, the tool that would in principle test genetic hypotheses, lose most of their predictive accuracy when applied to ancestries different from the sample they were derived in, because linkage patterns and environments differ, which makes cross-population comparisons of such scores uninterpretable. The American Society of Human Genetics and the AABA have both issued statements to this effect.

The forensic case, which is genuinely awkward

Forensic anthropologists have long estimated ancestry from crania, and software such as FORDISC classifies an unknown skull against reference samples at rates better than chance. This is used to generate investigative leads. It is also the strongest-looking practical argument that racial categories track biology, so it deserves a direct answer.

The method works because population structure is real and because reference samples are drawn from socially defined groups whose members share ancestry, geography, and often developmental environment. What it does not do is validate race as a natural kind. A discriminant function assigns an unknown to the nearest available reference group whether or not that group is the right one; accuracy collapses for individuals whose population is not in the database; and admixed individuals, who are a large and growing share of many populations, are systematically misassigned. Since roughly 2020 the profession has argued openly about this, with some practitioners calling for the abandonment of ancestry estimation entirely and others arguing for reframing it as population affinity with explicit error rates. That argument is live, and it is being conducted by people who do the casework.

Common misconceptions

• Saying race is a social construct means human populations do not differ biologically. They do differ, clinally and by trait; what fails is the claim that those differences sort into discrete taxa.
• Lewontin's result was refuted by Edwards. Edwards showed multi-locus classification works; he did not show the recovered clusters are biological races, and both results are accepted.
• Genetic clustering studies discover natural human groups. The number of clusters is specified by the analyst, and even sampling along gradients replaces clusters with clines.
• Race is biologically useful in medicine. It is a poor proxy for ancestry and exposure, and several race-corrected clinical algorithms have been withdrawn for causing harm.
• High heritability within groups implies genetic causes of between-group differences. This is the seed-and-soil error, and it is the foundational mistake in the intelligence literature.

Recap

• Lewontin found 85 percent of variation within populations in 1972, and genome-scale work has raised that figure rather than lowering it.
• Edwards correctly showed that many loci jointly permit accurate classification, which establishes population structure, not racial taxa.
• Human FST among continental groups is roughly 0.05 to 0.15, far below the 0.25 to 0.30 typical of recognized mammalian subspecies, and the variation is clinal rather than tree-like.
• Racial categories have been redefined repeatedly by courts and censuses, most starkly in Ozawa and Thind three months apart.
• Racism has measurable biological effects through weathering, miscalibrated instruments, and race-corrected clinical algorithms.
• Within-group heritability says nothing about between-group differences, and polygenic scores do not transfer across ancestries.

Sources

1. Rosenberg, N. A., et al. (2002). Genetic structure of human populations. Science, 298(5602), 2381-2385. doi.org
2. Templeton, A. R. (2013). Biological races in humans. Studies in History and Philosophy of Biological and Biomedical Sciences, 44(3), 262-271. doi.org
3. Vyas, D. A., Eisenstein, L. G., & Jones, D. S. (2020). Hidden in plain sight: Reconsidering the use of race correction in clinical algorithms. New England Journal of Medicine, 383, 874-882. doi.org
4. American Association of Biological Anthropologists. (2019). AABA statement on race and racism. bioanth.org
5. Britannica. (2024). Race, human. britannica.com

The big picture

In 1984, a year after the end of Argentina's military dictatorship, a commission investigating the disappearance of thousands of people asked the American Association for the Advancement of Science for technical help. They sent Clyde Snow, an Oklahoma forensic anthropologist. Snow arrived to find graves in Buenos Aires cemeteries being opened by backhoe, bones broken and mixed, evidence destroyed in the act of recovering it. He recruited a group of archaeology and anthropology students, none of whom had done this before, and taught them to excavate a grave the way you would excavate a site.

In 1985, at the trial of the junta commanders, Snow gave evidence about a young woman named Liliana Pereyra, whose skeleton had been recovered from an unmarked grave in Mar del Plata. The bone told the court that she had been killed by a shotgun blast fired at close range from behind, and that features of her pelvis were consistent with a recent pregnancy in a woman recorded as pregnant when she was taken. The students Snow trained became the Argentine Forensic Anthropology Team, which has since worked in more than fifty countries.

This last lesson is about biological anthropology applied: to identifying the dead, to medicine, and to understanding how human bodies grow. It ends with where the field is heading.

Forensic anthropology: the biological profile

A forensic anthropologist is asked, in effect, four questions about a set of skeletal remains: are they human, who might this be, what happened to them, and how long have they been here. The answer to the second question is a biological profile, which narrows a missing persons list rather than naming anyone.

Element	Best indicators	Reliability
Age, subadult	Dental development and eruption; long bone lengths; epiphyseal fusion sequence	High; dental development is the single most reliable subadult indicator
Age, adult	Pubic symphysis, auricular surface of the ilium, sternal rib ends	Moderate, and it degrades with age; intervals for older adults are very wide
Sex	Pelvis first: subpubic angle, greater sciatic notch, ventral arc. Skull second: mastoid, nuchal crest, supraorbital margin	Around 95 percent with a complete pelvis, lower from the skull; not estimable before puberty
Stature	Regression equations on long bone lengths	Moderate, with confidence intervals often spanning several centimeters
Individuating features	Healed fractures, surgical hardware, dental work, frontal sinus outline	High; frontal sinus patterns are effectively unique to an individual

Two honest notes belong here. First, skeletal sex estimation reports the pelvis, not a person's gender identity, and the profession is actively working out how to say that in a report without either erasing people or overclaiming. Second, some traditional indicators have not held up: pitting and scarring on the pubic bone were long read as evidence of childbirth, and controlled studies have shown that the association is far weaker than practitioners believed. A field that reports its own failed indicators is doing this correctly.

The most distinctive expertise is reading trauma timing. Antemortem injury shows healing: remodeled bone, rounded margins, callus. Perimortem damage, occurring around the time of death while bone is still fresh and full of collagen, produces a green-bone response: beveled entry margins, plastic deformation, hinged fractures, and fracture surfaces the same color as the surrounding bone. Postmortem damage to dry bone is sharper, more irregular, and lighter in color at the break. Distinguishing a perimortem gunshot from a backhoe strike is exactly what Snow was doing in Buenos Aires.

The discipline's research base includes outdoor decomposition facilities, the first established at the University of Tennessee in 1981 by William Bass. Bass founded it after being asked to date a body found in a disturbed grave, estimating it had been dead a few months, and learning that the man was a Civil War colonel buried in 1864, embalmed and sealed in a cast-iron coffin. He had been wrong by more than a century, and he built a research program rather than an excuse.

Key idea: A biological profile estimates age, sex, stature, and individuating features to narrow a list of candidates, while trauma timing distinguishes injury before, around, and after death from the mechanical response of fresh versus dry bone.

Evolutionary medicine, and how to use it without fooling yourself

Randolph Nesse and George Williams framed the useful question in 1991: not what causes a disease, but why natural selection left the body vulnerable to it. Their answers form a checklist worth carrying.

• Mismatch. Bodies adapted to past environments now live in different ones.
• Pathogen coevolution. Bacteria and viruses evolve on timescales of days; we do not.
• Constraints. Selection modifies what exists. The human pharynx puts the food and air passages across each other, and people choke.
• Trade-offs. An aggressive immune system fights infection and produces autoimmunity.
• Reproduction, not health. Selection favors traits that raise reproductive success even at a cost to longevity or comfort.
• Defenses mistaken for problems. Fever, cough, vomiting, and pain are protective responses, calibrated like a smoke detector to tolerate false alarms because missing a real fire is worse.

Now the criticism, which the field applies to itself. The mismatch hypothesis is the most attractive and the most abused of these. James Neel proposed the thrifty genotype in 1962: alleles favoring efficient fat storage under feast-and-famine conditions now produce type 2 diabetes under constant abundance. It is an elegant story, and genome-wide scans have not found the expected signatures of strong recent positive selection on the major diabetes risk variants. Neel himself walked it back. The related thrifty phenotype proposal from David Barker, in which conditions in utero program later metabolism, has fared better; the Dutch Hunger Winter cohort shows elevated metabolic disease in people whose mothers were starved during early pregnancy, though disentangling that from later circumstances is hard.

Some mismatch claims are well supported. Myopia has risen steeply in populations with intensive schooling, and randomized and cohort studies both indicate that time spent outdoors in bright light during childhood is protective, which is a specific, testable, actionable finding. Others are weaker. The hygiene hypothesis, in its modern old friends form, proposes that reduced exposure to helminths and commensal organisms leaves immune regulation miscalibrated; the epidemiological correlations are real and the clinical trials of deliberate helminth therapy have mostly disappointed.

The failure mode to watch for is the ancestral diet argument, which assumes a single environment our bodies were designed for. There was no such environment. Hadza foragers take a substantial share of calories from honey in season; traditional Inuit diets contained almost no plant food; Tsimane horticulturalists eat a high-carbohydrate diet and have among the lowest rates of coronary artery disease ever measured. Herman Pontzer's team measured total daily energy expenditure in Hadza adults with doubly labeled water in 2012 and found it statistically indistinguishable from that of sedentary Western adults once body size was accounted for. That does not mean exercise is useless; it means the story in which modern obesity is simply a matter of moving less than our ancestors does not survive measurement. The rule to apply is that an evolutionary hypothesis in medicine earns its keep by making a prediction that could fail.

Where the framework pays off is concrete. Antibiotic resistance is evolution observable in a hospital ward, and stewardship programs are applied population genetics. Cancer is clonal evolution under selection, which is the reasoning behind adaptive therapy strategies that dose to contain rather than to eradicate. Pharmacogenetics is direct: CYP2D6 variants alter codeine metabolism dangerously in both directions, and screening for the HLA-B 5701 variant before prescribing abacavir prevents a potentially fatal hypersensitivity reaction.

Key idea: Evolutionary medicine is productive when it generates falsifiable predictions, as with outdoor light and myopia or resistance management, and unreliable when it reasons backward from a single imagined ancestral environment that never existed.

Growing up human

Human life history contains a contradiction that has to be explained. Compared with chimpanzees, we wean our infants early, at roughly two and a half to three years rather than four to five. Yet we mature far later, reproduce for the first time years later, and live much longer. Weaning early should mean growing up fast. We do the opposite.

Barry Bogin's resolution is that humans inserted an extra stage. After infancy and before the juvenile period, humans have a childhood: a stage in which the individual is weaned but cannot feed itself, requires specially prepared food, grows slowly in body while the brain continues to consume an enormous share of energy, and depends on adults other than the mother. No other primate has it. And it only works if someone besides the mother provisions the child, which is Sarah Hrdy's argument for cooperative breeding as a foundation of human evolution: early weaning is possible precisely because fathers, grandmothers, siblings, and others feed the weaned child. Adolescence adds a second oddity, a growth spurt of a magnitude no other primate shows.

Then there is menopause. Human females typically cease reproduction around fifty and, in populations with low mortality, may live decades afterward. Among mammals a substantial post-reproductive lifespan is otherwise essentially restricted to a few toothed whales. Kristen Hawkes and colleagues proposed the grandmother hypothesis in 1998 from Hadza foraging data: older women who stop reproducing can provision daughters and grandchildren, raising their daughters' fertility and their grandchildren's survival enough to outweigh the offspring they forgo. Hillard Kaplan's embodied capital model instead emphasizes the long investment required to produce a competent adult. A third line, developed from Finnish and Canadian church and parish records by Mirkka Lahdenpera and others, models reproductive conflict between generations of women in the same household, showing that overlapping reproduction imposed measurable costs on both. These are not settled, and the grandmother hypothesis in particular is quoted far more confidently in popular writing than in the literature.

Growth research is also the field's most direct public health contribution. Height-for-age is an integrated record of nutrition and infection during development, which is why stunting is a standard population health indicator. The World Health Organization's growth standards were built by following well-nourished children in Brazil, Ghana, India, Norway, Oman, and the United States, and the study found that children from all six grew closely enough alike that a single standard applies. That result, arrived at empirically, is the clinical restatement of everything Module 6 has argued: the differences we see between populations are overwhelmingly about circumstances.

Where the field is going

Four directions are clear. Ancient proteins are pushing molecular evidence past the thermal limits of DNA, with enamel proteomes now reaching close to two million years. Sediment DNA is recovering populations from caves containing no bones. Functional genomics is testing what human-specific variants actually do, with brain organoid work on genes such as SRGAP2C, ARHGAP11B, and NOVA1, though the leap from a dish of cells to a claim about human cognition is long and frequently overstated in press releases. And synchrotron imaging plus routine three-dimensional scanning are making it possible to study fossils without cutting them and to share data so that anyone can measure a specimen.

The ethical direction is just as consequential. Repatriation obligations are expanding, consent frameworks for genomic work with communities are becoming standard, and there is pressure toward genuinely equitable partnerships with institutions in the countries where the fossils and the people actually are. Data-sharing norms are being written to sit alongside open science rather than against it, giving source communities a say in how data about their ancestors are used.

And most of the record has not been found. Vast areas of Africa have never been systematically surveyed, the tropical forest fossil record is nearly empty, and every year produces specimens that do not fit. If you want to work in this field, the United States Bureau of Labor Statistics tracks anthropologists and archeologists as an occupation, most positions are outside universities, and the reliable route in is a strong foundation in biology and statistics, field or laboratory experience, and a willingness to be wrong in public.

Common misconceptions

• Forensic anthropologists identify people from bones. They build a biological profile that narrows a candidate list; identification normally requires DNA, dental records, or unique features.
• Age at death can be estimated precisely in adults. Precision drops sharply after about thirty, and honest reports give wide intervals.
• The thrifty genotype explains type 2 diabetes. Genome scans have not supported it, and the developmental thrifty phenotype account has more evidence behind it.
• Foragers were highly physically active, which is why they were lean. Measured total daily energy expenditure in Hadza adults matches that of sedentary Westerners once body size is controlled.
• Grandmothers evolved menopause, and this is established. The grandmother hypothesis is one of at least three competing models and is not settled.
• Growth differences between countries are mostly genetic. The WHO standards were built precisely because well-nourished children across six countries grow closely alike.

Recap

• The biological profile estimates age, sex, stature, and individuating features; the pelvis gives the best sex estimate and dental development the best subadult age.
• Trauma timing rests on the green-bone response, distinguishing perimortem from antemortem and postmortem damage.
• Nesse and Williams's six categories explain evolutionary vulnerability; mismatch claims must make falsifiable predictions rather than invoke one imagined ancestral environment.
• Antibiotic stewardship, adaptive cancer therapy, and pharmacogenetics are evolutionary reasoning with clinical payoffs.
• Human childhood is an inserted life history stage made possible by cooperative provisioning, and menopause has at least three competing explanations.
• Palaeoproteomics, sediment DNA, functional genomics, non-destructive imaging, repatriation, and equitable partnership define the field's next decade.

Sources

1. Britannica. (2024). Forensic anthropology. britannica.com
2. Pontzer, H., et al. (2012). Hunter-gatherer energetics and human obesity. PLoS ONE, 7(7), e40503. doi.org
3. Hawkes, K., O'Connell, J. F., Blurton Jones, N. G., Alvarez, H., & Charnov, E. L. (1998). Grandmothering, menopause, and the evolution of human life histories. PNAS, 95(3), 1336-1339. doi.org
4. World Health Organization. (n.d.). Child growth standards. who.int
5. U.S. Bureau of Labor Statistics. (n.d.). Anthropologists and archeologists. Occupational Outlook Handbook. bls.gov