What the Fossil Couldn't Carry

Luke Fannin filmed it. He and W. Scott McGraw, both at Ohio State, measured what the footage showed and published it in Proceedings of the National Academy of Sciences: mangabeys averaged 45.7 degrees of peak midfoot flexion during vertical climbs, against 23.3 degrees at the ankle. The foot bends nearly twice as far in the middle as at the joint you would think was doing the work.

The number to hold that against is roughly 45 degrees — the ankle flexion a chimpanzee achieves going up a comparable trunk. The match is close enough to be quotable and it should not be quoted. Two angles at two joints in two animals are not the same quantity; equal degrees do not buy equal mechanics, and a near-identity like that is a coincidence of units doing work the numbers cannot do. What the authors claim is convergence — comparable performance getting up the same tree by different routes. Different joint. Different bones. Same tree.

Fannin calls this functional equivalence, and it would be a footnote in a biomechanics journal except for where it lands. One of the load-bearing arguments in human origins runs straight through feet. The question is whether the last common ancestor of humans and chimpanzees was a vertical climber — whether bipedalism emerged from an animal still spending real time going up trunks, or from one that had already committed to the ground. Feet were the evidence. Chimpanzee feet are mobile and grasping and ape-like. Early hominin feet, where we have them, look stiffer, more arched, more terrestrial. The inference followed cleanly: no ape-like foot, no vertical climbing.

The mangabey does not have an ape-like foot. It goes up trunks anyway, using a joint the argument was not watching.

The objection is worth stating before someone else states it. The mangabey climbs by being flexible in the midfoot, and reduced midfoot mobility is precisely what early hominin feet are said to display. Read uncharitably, the counterexample bends the wrong way: it shows you still need a mobile foot, only somewhere else. So the finding does not say that any foot can get up a trunk. It says something narrower and more corrosive — that the argument was being conducted at the ankle, and the ankle was not where the climbing was happening. Once the joint you choose to measure determines the answer you get, a fossil that preserves some joint surfaces well and others badly stops being a neutral witness.

Notice what the fossil actually did here, because it isn't the villain. It didn't lie. It didn't mislead. It carried, faithfully and for millions of years, precisely what it was capable of carrying: mineralised bone, and the shapes of the surfaces where bones met. Everything else about the animal — how far a joint would really travel with ligament and cartilage in place, which joint it preferred under load, what it did on a wet Tuesday in the rainy season — was never in the rock. It couldn't be. Soft tissue and behaviour don't fossilise. They aren't hidden in the record. They were never entered into it.

So we did the reasonable thing, which is also the thing that got us here. We took the bones we had, found the nearest living animal whose bones looked similar, and borrowed its behaviour. The chimpanzee became the default template for the ancestor not because the evidence pointed there but because the evidence stopped, and something had to occupy the other side of the stop. An analogy was installed in the gap, and then it sat there long enough to stop looking like an analogy.

That isn't a failure of paleoanthropology. That is paleoanthropology working. An inference chain is sound right up until it becomes furniture, and then you can't see it anymore — you just walk around it. What broke this one wasn't a better fossil. It was a camera pointed at an animal that was alive.

The restraint worth keeping: one monkey species, one forest. Nothing here says the last common ancestor climbed with a midfoot break. It says you cannot rule out that it climbed on the grounds that its feet were the wrong shape — which is a smaller claim than the headline and a stranger one than it sounds. The constraint was never in the past. The past has been sitting there fully specified this whole time, entirely indifferent to whether we recover it. The constraint was in the medium — in what a bone is physically able to hold — and then in us, for reading the limits of the medium as the limits of the animal.

And then correcting it the only way available, which is by pointing a camera at a different living animal. That is the same move run again with a wider reference class. The mangabey does not get us out of the trap. It shows the trap was baited with one species when it could have been baited with dozens, and that there is no route to the ancestor except analogy — the error was never using a proxy, it was owning exactly one. Every improvement from here is a better-populated comparison set, not a closer look at the thing itself. The thing itself does not come back.

The mangabeys have been climbing like this the entire time. Nobody had to dig them up. That is not the same as having looked.