top of page

How Not To Date A Hominin

Seeing at close hand that science is not always about the search for truth came as a bit of a surprise

Chimp Investor is ranked #3 of 90 by Blog Overview and #10 of 100 by FeedSpot in their relevant categories. A review on WealthTender would be much appreciated.

I got invoved as part of my geology BSc last year in an interesting project relating to hominin evolution in Africa. Hominins are those species, including our own, Homo sapiens, in the branch of the evolutionary tree that split over 7ma from that which includes chimps and bonobos, our closest living relatives. Bipedalism went one way, knuckle walking the other.

I thought Chimp Investor readers may be interested to read more about my project, not least because during it I was accused by authors of one paper that had been published in the presitigious Nature magazine of being "on a witch hunt" and "out to ruin reputations". I was quite surprised by these accusations as all I had done was ask them politely for details of their methodology for dating their rock samples, details that were not in the paper's supplementary material. I required these details in order to replicate their results, the basis of empirical science.

What has hominin evolution got to do with geology? An understanding of evolution - hominin or otherwise - is based on the fossil record, and fossils are found in rocks. Rocks of course are the domain of geology, though there is more to it than that. To be scientifically meaningful, fossils must be dated, and dating of fossils - and rocks generally - is a branch of geology.

The claim of the paper whose authors accused me of behaving badly was that the distribution of ages of 29 rock layers knowns as flowstones (see diagram below) from South African caves between which fossils had been found was such that certain interesting conclusions about hominin evolution could be drawn.

My project supervisors believed that some of the ages used in the paper were wrong and also that spurious statistical techniques had been used to discern a meaningful distribution. They suggested that an investigation of these alleged errors would make for an interesting project.

They were right!

Human evolution is fascinating, scientifically. It is also engages on an emotional level - to know our ancestry is a primal urge. The project involved lots of mathematics - one of my very limited number of abilities. It also involved me being exposed to the ugly side of science - the reality that data gets faked in order to derive an interesting finding - known in the trade as p-hacking - which in turn leads to getting published and receiving all the benefits that come with that.

On a more positive note, my project later formed the basis of a NERC (Natural Environment Research Council) grant application submitted by my supervisors to re-date some of the 29 rocks cited in the offending paper - this ultimately is the better way to expose untruths than via an undergraduate dissertation.

In Africa, hominin fossils have been found in three distinct environments: the East African Rift, carbonate caves in South Africa, and the Chad basin in central Africa (see diagrams below).

These three areas were not necessarily the only places hominins lived, just that they were ones that had good fossil preservation potential.

In South Africa, remains of hominins that fell or were dragged by predators into caves could quickly be covered by material washed in shortly after and thus preserved (see diagram below).

In East Africa, remains that ended up on lake beds could get buried by ash erupted by volcanoes along the rift. The Chad endorheic basin - one that loses water only through evaporation or seepage into the ground - would also have provided good fossil preservation potential.

Moreover, dating techniques have varied across the three areas. The volcanic sediments of East Africa have lent themselves to argon-based radiometric dating. Mammal association - estimating the age of something based on knowledge of the age of a mammalian fossil found with it - and uranium disequilibrium dating, among others, have been used to date remains found in South African caves. The ages of the remains of the two species found in Chad were estimated using mammal association and cosmogenic nuclide dating techniques.

While argon-based dating is both accurate and precise, the methods applied in Chad and South Africa are generally far less so. Mammal association often comes with significant uncertainty. Uranium disequilibrium dating used in South Africa, as you will see, comes with large error for older rocks. As for Chad, one of the two remains was a surface find. This anyway would have introduced considerable uncertainty but, in addition, the remains showed evidence of having been tampered with. The skeleton, it is alleged, had been arranged facing Mecca and its jaw bone contained a false tooth!

Traditional classification schemes for living things are based on what they look like - anatomy, structure etc. They are thus often disputed, since what an organism looks like is something that gets qualitatively not quantitatively measured. New schemes that are based on genetic makeup provide a quantitative and thus reliable indicator of relationships between organisms. Findings of these new schemes suggest that traditional looks-based schemes may contain significant errors.

Within the hominin branch, we are the only extant species so classification gets based only on looks. Two such classification frameworks - branch networks - are as below. There are several differences between them but an example of one is whether the species rudolfensis is a member of the Homo or Kenyanthropus genus. Given that classification by looks is essentially arbitrary, evolutionary trees are an area where lumpers and splitters continue to do battle.

Source: Encyclopedia Britannica (2005)

Most palaeoanthropologists agree that the four key traits that distinguish hominins are: coming down from trees (terrestriality), walking upright (bipedalism), tool use, and acquiring bigger brains (encephalisation). However, there is disagreement as to the order in which these occurred.

The current evidence favours the bipedalism-then-terrestriality-then-tool-use-then-large-brains model. Indeed, the earliest hominin fossils (5-7 Ma) indicate a primitive form of bipedalism but also arboreality, suggesting that walking upright was, at least at first, an adaptation that aided movement around or, at most, between trees. Fossilised footprints found in Laetoli, Tanzania and dated to 3.7 Ma indicate more advanced bipedalism i.e. travel over longer distances, not just from tree to tree. The oldest stone tools, found on the western shore of Lake Turkana in Kenya, have been dated to 3.3 Ma, while cranial evidence indicates that rapidly increasing brain size did not begin until the appearance of the genus Homo around 2.4 Ma.

Current fossil evidence still supports the proposition that the human-chimpanzee/bonobo last common ancestor originated in East Africa. Other geological evidence, specifically the formation of the East African Rift Valley, also supports this theory - jungle that was uplifted at rift margins became less dense forest that would have favoured bipedal creatures.

That said, theories that it originated elsewhere are still valid and the lack of strong evidence supporting them may simply mean that it was destroyed or has yet to be found. The 7-6 Ma remains found in Chad have also served to weaken the argument for lineage origination in East Africa, though controversy continues to surround them.

As for my project, it was essentially about radiometric dating, specifically the method used by the aforementioned paper to date cave rocks known as flowstones. The paper states:

"Here we show that flowstones from eight Cradle caves date to six narrow time intervals between 3.2 and 1.3 million years ago. We use a kernel density estimate to combine 29 U–Pb ages into a single record of flowstone growth intervals. We argue that the entire early Cradle fossil record is restricted to these limited time intervals."

The diagram below shows the age estimates of the flowstones cited in the paper - the dark and light green bars represent uncertainties. As stated in the paper, the flowstones range in age from 3.2 to 1.3Ma. It can also be seen that the age uncertainties are on the whole small - many of them are single digit in % terms. The scale has been set such that visual comparison with the diagram shown further below of age estimates as calculated by my project is easier.

The authors then derive a statistical distribution for ages of all flowstones in the eight caves, based on their 29 ages, as below. This shows the "six narrow time intervals" cited by the paper.

To derive the below type of distribution, a parameter, the so-called bandwidth, must be used that reflects both the number of samples and the uncertainty of the samples. The paper used a bandwidth of 0.03Ma (30,000 years) which essentially increased the resolution of the distribution and allowed the six peaks to be discerned. Why the authors used this bandwidth is unclear because, based on the number of samples and the uncertainties, they should have used a bandwidth of 230,000 years.

Had they used the correct bandwidth, their distribution would have looked like the diagram below. No "six narrow time intervals". No interesting conclusion.

Although the above uses the appropriate bandwidth, it is still based on the paper's ages and their uncertainties. A large part of my project involved re-dating samples using the program IsoplotR that was written by one of my supervisors. The below shows the ages of the samples as calculated by IsoplotR. Quite a difference, obviously.

I also scoured the literature for other cave samples that had not been considered by the paper and came across quite a few. Adding in their ages and the ages of samples cited in the paper as calculated by IsoplotR, and using an appropriate bandwidth, you get the below distribution. It would certainly not have warranted publication in Nature or any other journal for that matter.

You may be wondering why there was such a big difference in ages and uncertainties between the paper and my project. The dating methodology in question is called 234/238 uranium disequilibrium dating and is based on the principle that the drip water from which flowstones and other cave sediments form contain amounts of 234-uranium and 238-uranium that are out of balance i.e., in disequilibrium. Once in a flowstone or stalactite, the 234U and 238U decay undisturbed and after around 2ma the amounts get back into balance i.e., any residual imbalance is unmeasurable. If residual imbalance can be measured, the time of formation of the rock - its age - can be estimated.

Given the residual imbalance, together with the uncertainties of the measurement, there is essentially only one way to derive an age estimate - it involves a precise if somewhat complicated mathematical formula. In other words, it is hard to know why there was such a big difference. As I said, I did ask the authors for details of their calculations but instead of sending them they just threw the aforementioned accusations at me.

For anyone really interested, my dissertation is here.

53 views0 comments

Related Posts

See All


bottom of page