3.1 — The Evidence

A whale's flipper contains a humerus, a radius, an ulna, and five sets of finger bones. It uses none of them as fingers. A bat's wing contains the same bones, enormously elongated to stretch a membrane. A horse's leg contains them too, with most of the fingers reduced to splinters and the animal standing on the tip of one enlarged middle digit.

If each of these animals had been designed for its job, none of them would be built this way. A flipper does not need finger bones. A wing does not need a wrist. The only reason to build a paddle out of a hand is that you started with a hand.

That is the shape of the argument this page makes: not one decisive proof, but many independent lines of evidence that agree with each other for no reason except that they are all recording the same history.

Line 1: the same bones, rearranged

The same forelimb, four times. Corresponding bones are shown in the same colour: one upper arm bone, two forearm bones, a cluster of wrist bones, then digits. The proportions differ wildly; the plan never does. Image: Wikimedia Commons.

Structures that share an underlying plan because they were inherited from a common ancestor are called homologous. The forelimb is the standard example, and the pattern — one bone, two bones, many small bones, digits — holds across every land vertebrate and their aquatic descendants, from a frog to a bird to you.

Contrast this with analogy. A bird's wing and an insect's wing both fly, and they have nothing structurally in common: one is a modified forelimb with bones inside, the other is a membrane growing from the body wall with no bones at all. Two solutions to the same problem, arrived at separately. That is convergent evolution, and telling homology from analogy is the basic skill of comparative anatomy.

The strongest version of this evidence is not the useful structures but the useless ones. A whale has a pelvis — a pair of small bones floating in the body wall, attached to no leg, doing nothing for locomotion. Some whales retain vestigial femurs. Flightless birds have wing bones. Cave fish that have lived in darkness for millions of years still develop eyes in the embryo and then degenerate them.

Your own body carries several. The coccyx is three to five fused vertebrae — the remnant of a tail, and human embryos have a genuine tail with 10 to 12 vertebrae at around week five, which is then removed by the programmed cell death of Chapter 1.8. The muscles that would move a human ear are present and non-functional in most people. The appendix is much reduced from the large fermentation chamber it is in herbivores, though it is not entirely useless — it appears to serve as a reservoir of gut bacteria. Goosebumps are the contraction of muscles that would raise fur you no longer have.

A vestigial structure is not evidence of poor design so much as evidence of history. There is no reason to build a whale with a pelvis unless whales descend from animals that walked.

Line 2: the fossils, and the predictions they confirmed

Archaeopteryx, found in 1861, two years after Darwin published. Feathers and a wishbone, like a bird. Teeth in the jaw, clawed fingers on the wing, and a long bony tail, like a small dinosaur. It is not an ancestor of modern birds, but it is exactly the kind of mixture the theory required to exist. Image: Wikimedia Commons.

Fossils are rare, because fossilisation requires rapid burial in the right sediment and most organisms simply decay. So the fossil record is a heavily incomplete sample, and any argument built on gaps in it is an argument about sampling rather than about biology.

What matters is that where the record is good, it shows the expected sequence, and it never shows the unexpected one. The order is consistent everywhere on Earth: single-celled organisms in the oldest rocks, then simple multicellular life, then fish, then amphibians, then reptiles, then mammals and birds. No rabbit has ever been found in Precambrian rock, and J. B. S. Haldane's remark that such a find would destroy the theory is the point — the theory is falsifiable and has not been falsified.

The transitional sequences are now numerous. Fish to tetrapod, through Tiktaalik and its relatives. Reptile to mammal, through a long series in which the jaw joint bones migrate into the middle ear — a sequence so well documented that intermediate forms exist with two functioning jaw joints side by side. Land mammal to whale, through Pakicetus and Ambulocetus.

Tiktaalik. Scales, gills and fins like a fish; a flat head, a mobile neck and fins containing shoulder, elbow and wrist joints capable of propping the animal up, like a tetrapod. It was found where and when it was predicted to be. Image: Wikimedia Commons.

The Tiktaalik case deserves emphasis because it was a prediction, not a discovery after the fact. Neil Shubin's team reasoned that the fish-to-tetrapod transition happened around 375 million years ago, so they needed rocks of that age, laid down in shallow freshwater, and exposed at the surface. Those criteria pointed to Ellesmere Island in the Canadian Arctic. They searched for four summers and found Tiktaalik in 2004, in exactly the predicted age and environment.

A theory that can tell you where to dig is doing real work. This is the same kind of test as the prediction of a planet's position from Newton's laws (Volume IV, Chapter 11).

Line 3: the molecular evidence, which is now the strongest

Darwin had no access to this, and it is where the case has become overwhelming, because it provides independent confirmation of relationships worked out from anatomy a century earlier.

The genetic code is shared by everything (Chapter 1.9). So is ATP, so is the ribosome, so is the direction of DNA replication.

Sequence similarity tracks the anatomical tree exactly. Human and chimpanzee protein-coding DNA is about 98.8 percent identical; human and mouse around 85 percent in the coding regions; human and fruit fly far less. Nobody arranged this. The molecular tree and the anatomical tree were built from completely different data, decades apart, and they agree.

The really compelling evidence is the shared mistakes. Two students with the same right answers may both be competent. Two students with the same wrong answers copied from each other.

Pseudogenes are broken copies of genes, disabled by mutation and no longer read. The vitamin C gene is the standard case. Almost all mammals make their own vitamin C using an enzyme called GULO. Humans cannot — which is why scurvy exists and why sailors needed limes. But we still carry the GULO gene, sitting in our genome, disabled by specific mutations. Chimpanzees, gorillas and orangutans carry the same broken gene with the same disabling mutations at the same positions. Guinea pigs also cannot make vitamin C and also carry a broken GULO — but broken in a different way, by different mutations.

There is no design explanation for a shared identical error. Shared inheritance explains it in one sentence: the gene broke once in a common ancestor of the great apes, and separately in the guinea pig lineage.

Endogenous retroviruses are the same argument, more forcefully. Some viruses insert their DNA into the host genome. If that happens in a germ cell, the insertion is inherited. About 8 percent of your genome is old viral sequence — more DNA than all your protein-coding genes combined. Humans and chimpanzees share hundreds of thousands of these insertions at identical positions. A retrovirus inserts more or less at random among millions of possible sites, so the chance of the same insertion landing at the same base in two lineages independently is negligible. Shared position means shared ancestor who was infected.

Line 4: geography

Islands that were never connected to a continent have no native land mammals other than bats, and no native amphibians. They have birds, insects and plants — organisms that can cross water. This has nothing to do with what an island can support: introduce a rabbit to such an island and it thrives, often catastrophically. The pattern is not about suitability, it is about arrival.

Four of the Galápagos finches as drawn by John Gould. All are close relatives of a single South American species, and the beaks differ according to what each population eats — heavy for crushing seeds, fine for probing. The island is not what varies. The food is. Image: Wikimedia Commons.

Island species most closely resemble the nearest mainland species, not species from similar climates elsewhere. Galápagos animals resemble South American ones; Cape Verde animals, at a similar latitude and climate off Africa, resemble African ones. Darwin noticed this before he had a mechanism, and it was one of the observations that would not leave him alone.

And the continental distributions match plate tectonics. Marsupials are found in Australia and South America, with fossils in Antarctica in between — exactly the pattern expected if those three landmasses were joined and drifted apart, which the geology independently establishes (Volume VIII, Chapter 4).

Line 5: evolution observed directly

This is not something inferred only from the past. It is measured now.

The peppered moth in its pale form, nearly invisible against lichen-covered bark. When industrial soot killed the lichen and blackened the trunks, the dark form — previously rare — became the one predators could not see, and it went from under 2 percent to over 95 percent of the population in industrial England within about fifty years. Clean air legislation reversed it. Image: Wikimedia Commons.

Antibiotic resistance is evolution happening in hospitals, and Chapter 17.12 covers it as a clinical emergency. It is the same process as the moth: a variant that was rare becomes common because the environment changed to favour it.

The Galápagos finches were measured directly by Peter and Rosemary Grant over forty years. After the drought of 1977, small soft seeds ran out and only large hard ones remained. Birds with deeper beaks survived at higher rates, and the average beak depth of the population measurably increased in a single generation. When wet years returned and small seeds became abundant, it went back down. Selection was observed, quantified, and seen to reverse.

Richard Lenski's long-term experiment has grown E. coli continuously since 1988, now past 75,000 generations, freezing samples along the way so any point in the past can be revived and re-run. Around generation 31,500 one population evolved the ability to use citrate as a food source under oxygen — something E. coli famously cannot do. Reviving frozen ancestors showed the innovation required an earlier, apparently unrelated mutation that had occurred thousands of generations before and made the later change possible. A genuinely new capability was observed appearing, and its history could be replayed.

What the theory does and does not claim

Three corrections, because these misunderstandings do real damage to people's ability to reason about medicine.

Evolution does not mean progress toward humans. There is no ladder. Bacteria are not primitive attempts at being something else; they are extremely successful organisms that have been optimising for far longer than we have. The tree has no top.

Individuals do not evolve. Populations do. Nothing that happens to you changes your descendants' genes, apart from the mutations in your gametes. A giraffe stretching its neck does not lengthen its offspring's neck; giraffes with longer necks left more offspring.

"Only a theory" misuses the word. In ordinary speech a theory is a guess. In science a theory is a well-supported explanatory framework — germ theory, atomic theory, the theory of gravitation. The fact of evolution — that populations change over time and species share ancestry — is as securely established as anything in biology. The theory is the explanation of how, and it is refined continuously, which is what a healthy scientific field does.

Why this matters to your health

This is not a historical aside. Evolutionary reasoning is used in the clinic every week.

Antibiotic prescribing policy is applied evolutionary biology. Chapter 17.12.

Cancer is evolution inside your body. A tumour is a population of cells with heritable variation, competing for resources, under selection. Chemotherapy is a selection pressure, and resistance emerges the same way antibiotic resistance does. Chapter 19.3.

Influenza vaccines are reformulated annually because the virus evolves away from immunity — the process is tracked in real time and next season's strains are predicted from this year's trees. Chapter 17.13.

And drug development depends on shared ancestry. A drug is tested in mice because mouse biochemistry is close enough to ours to be informative, and it is close enough because of common descent.

What the next page fixes

This page showed that it happened. It did not explain how — what the actual mechanism is that turns a fish fin into a hand, or a moth population white then black then white again. Chapter 3.2 takes natural selection apart into its three requirements, shows exactly what it can and cannot do, and answers the objection that always comes up: how something as complicated as an eye could arise by accumulated accident.