sciencebriefs
13:00in productionCh. 1 · A clock made of protein differences/ 13:00 · ceiling 15 min
Evolution · Genetics

Molecular clock

A 1967 protein comparison suggested humans and chimpanzees split far more recently than fossils then implied, and the molecular clock method behind that claim has since been refined to account for the fact that its own ticking rate is not actually constant.

Emile Zuckerkandl and Linus Pauling noticed in 1962 that differences between the same protein in different species tend to grow roughly in step with the time since those species diverged, an idea formalised as the molecular clock. Vincent Sarich and Allan Wilson applied the method to primates in 1967, using blood protein comparisons to put the human-chimpanzee split at roughly four to six million years ago, far more recent than the ten to thirty million years then favoured by many palaeontologists, a dispute later settled largely in the molecular clock's favour by new fossil finds. The method requires calibration against independent fossil evidence, since molecular data alone carries no absolute date, and its founding assumption of a roughly constant mutation rate turns out to be only approximately true, varying by generation time, population size and other factors between lineages. Modern versions use statistical models that allow the rate to vary rather than assuming it is fixed.

Chapters & takeaways6
  1. 0:08
    A clock made of protein differences

    Zuckerkandl and Pauling noticed in 1962 that protein differences between species tend to accumulate roughly in step with time since divergence.

  2. 2:10
    A younger human-chimpanzee split

    Sarich and Wilson's 1967 protein comparison put the human-chimpanzee divergence at four to six million years, far more recent than the fossil estimate of the time.

  3. 4:20
    Why the estimate needed fossils anyway

    Molecular data alone carries no absolute date, so any molecular clock estimate must be calibrated against independent fossil evidence.

  4. 6:30
    The clock does not tick evenly

    Mutation rates vary between lineages depending on generation time, population size and other factors, undermining the assumption of a strictly constant rate.

  5. 8:40
    Where the method breaks down

    Over very long timescales repeated changes at the same site blur the signal, and over very short timescales unfixed variation inflates apparent rates.

  6. 10:50
    Living with an uneven clock

    Modern relaxed-clock methods use statistical models that allow mutation rates to vary across a family tree rather than assuming a single fixed rate.

Worth your time?

Yes. Study the whole thing.

4/ 5
What works
  • shows a clear historical case where molecular evidence challenged and then reshaped fossil-based consensus
  • is specific about which assumption failed and how researchers compensated for it
  • explains calibration honestly as a dependency on fossils, not something molecular data provides alone
What does not
  • cannot supply an absolute date without fossil calibration, however precise the protein or DNA comparison is
  • does not fully resolve disagreement over rates in poorly fossil-documented lineages
Study it if
  • readers who want to know how scientists date evolutionary splits without a time machine
  • anyone curious how a controversial molecular estimate was later vindicated by new fossils
  • people interested in how a useful method survives having its central assumption turn out to be wrong
Skip it if
  • readers wanting a single divergence date that all methods agree on precisely
The written brief3 min read

A clock made of protein differences

The claim is that the accumulated differences between the same gene or protein in two related species carry information about how long ago those species shared a common ancestor, roughly the way rings in a tree trunk record its age. Emile Zuckerkandl and Linus Pauling proposed this idea in 1962 after observing that differences in haemoglobin between lineages seemed to grow in a roughly straight line with time as separately estimated from fossils. If mutations accumulate at something close to a steady rate, then counting the differences between two living species should let researchers estimate when their lineages split, even in the absence of any fossil evidence bridging that specific gap.

A younger human-chimpanzee split

The most consequential early application of this idea came from Vincent Sarich and Allan Wilson at the University of California, Berkeley, who in 1967 compared blood serum albumin across primates and concluded that humans and chimpanzees diverged from a common ancestor around four to six million years ago. This was a startling claim at the time, because many palaeontologists working from the hominid fossil record then favoured a much older split, somewhere between ten and thirty million years ago. The dispute was substantial enough to be genuinely contentious within the field, pitting a new molecular method against established fossil interpretation, and it took further discoveries, including reassessment of certain ape fossils, to move the consensus closer to the molecular estimate.

Why the estimate needed fossils anyway

What the molecular clock cannot do on its own is supply an actual date. A count of protein or DNA differences only produces a number of changes, and translating that into years requires calibrating the rate of change against something with an independently known age, typically a fossil whose position in the family tree and geological age are both reasonably well established. This dependency means molecular clock dates are never purely molecular; they are statistical estimates built on an assumed or fitted mutation rate anchored to fossil evidence, and the accuracy of any given date is only as good as the calibration points used to set that rate.

The clock does not tick evenly

The clock’s founding assumption, that mutations accumulate at roughly the same rate across different lineages, does not hold as cleanly as Zuckerkandl and Pauling’s original observation suggested. Biologist Francisco Ayala identified several factors that can shift a lineage’s effective rate, including differences in generation time, population size, and the intensity of natural selection acting on a given gene. In practice this shows up as measurably different rates between groups, for instance certain seabirds accumulating change more slowly than other birds, and turtles evolving markedly more slowly than small mammals, differences plausibly linked to how quickly each group reproduces.

Where the method breaks down

The method also runs into trouble at the extremes of the timescale it is applied to. Over very long spans, the same site in a gene can be hit by multiple mutations one after another, a phenomenon called saturation, which makes the observed number of differences grow more slowly than the true number of changes and weakens the clock’s reliability. Over very short spans, by contrast, genetic variation that has not yet become fixed across a whole population can inflate the apparent mutation rate, causing molecular clock estimates at that end of the scale to overstate how long ago a divergence occurred, a distortion researchers have had to learn to correct for separately.

Living with an uneven clock

This is worth understanding because it is a genuinely instructive case of a method whose central assumption turned out to be wrong in detail while the underlying idea remained useful. Rather than abandoning the molecular clock once rate variation became clear, researchers developed relaxed-clock statistical models, using Bayesian methods, that explicitly allow mutation rates to differ across branches of a family tree instead of assuming one fixed rate throughout. For readers interested in how science adapts a flawed but productive tool rather than discarding it, and in how a controversial molecular claim about human origins was eventually supported by independent fossil evidence, this is a compact and rewarding case study.

Same field · Evolution4 of 65
Up next in Science

Nash equilibrium

· 13:00

A 28-page dissertation defined stability in a game as nobody being able to do better by switching strategy alone, a simple condition that says nothing about whether the outcome is actually good for anyone.

13:00