GRRRRRR! Why do I have to keep reading about how spearchucky modern humans went around killing Neandertals?

It's all over the science news this week -- Shanidar 3, a 50,000-year-old Neandertal from Iraq, has a partially-healed deep cut to one of its ribs. The kind of cut that gets there when you're stabbed with a knife or spear, or shot with an arrow.

That's pretty interesting in itself, although not news -- the wound has been known and listed as an example of ancient interpersonal violence since the 1960's. It may be the oldest clear example of a wound to a living person from a stone implement, because all earlier cutmarked human remains (there are many) may be post-mortem. Still, not news.

The "news" part is that Churchill and colleagues (2009) have shown that the wound is most consistent with a small projectile, as opposed to a large thrusting spear. In their discussion, they suggest that modern humans may be the culprits:

The nature of the lesion to the left 9th rib of the Shanidar 3 Neandertal ismost consistentwith injury from a low kinetic energy, low momentum weapon. While this does not rule out accidental injury or attack by a conspecific wielding a hand-held weapon, the nature of the traumatic damage, combined with the wound track suggested by the placement and orientation of the rib lesion, is consistent with injury by a long-range projectile weapon traveling along a ballistic trajectory. Given the possible sympatry of this Neandertal with early modern humans, and given possible assymetries in weapon technology between the two species, the case of Shanidar 3 is a good candidate for an instance of Neandertal-modern human interspecific violence (14, emphasis in original).

So what do I think? I guess anything is possible. But any good murder case hinges on motive and opportunity. Churchill and colleagues don't really give us a way to place modern humans at the scene at the crime.

The authors point to the only radiocarbon dates available for Shanidar 3, which at 46,000 and 50,000 radiocarbon years are in a range that we probably should not trust for old, non-AMS dates (as they point out). There's no indication of a non-Mousterian industry in the immediate region before 35,000 radiocarbon years, although an earlier presence of non-Neandertals is possible.

As it stands, what are the odds that the Neandertal Shanidar 3 individual (setting aside the question of whether it or other West Asian specimens really are the same as European Neandertals) would ever have encountered a "modern" human (setting aside the question of whether the two populations were distinct)?

Well, we don't really know. Probably pretty low. Maybe impossible. But that probability is very important to this "forensic" question.

Do we know that Shanidar 3 wasn't hit by a Neandertal weapon? The study shows very well that seven blows from a Mousterian point-hafted spear are not enough to guarantee a wound like the one on the ribs of Shanidar 3. That wound has a clear cut to one rib but little involvement of the adjacent rib. The seven experimental blows from a "high-energy" weapon resulted in five wounds with significant slices in two adjacent ribs, and two that affected no ribs at all. In contrast, the "low-energy" trials generated a high fraction of wounds that involved slices to only one rib. Meanwhile, two of those "low-energy" trials were from a Neandertal-associated tool, stabs with a Levallois point.

But what does this tell us, really? If we just treat the probability of a Shanidar 3-type injury as a binomial with 7 trials in this study, we can't reject the hypothesis that it occurs at a true likelihood of 35 percent. I think the authors have done a cool study, and did a great job evaluating the wounds. The problem is that 7 trials just isn't enough to yield a high confidence. We don't really know that a weapon known to be associated with Neandertals couldn't have made the wound.

So it comes down to opportunity. Can we place modern humans at the scene of the crime?

We may not be able to estimate this probability, but we can use Bayes' theorem to evaluate what the likelihood would have to be to satisfy our jury. Let's say our jury is stacked with Neandertals, and is willing to convict a modern human if there is just a five percent chance he committed the crime.

We don't know the probability that Shanidar 3 encountered modern humans. We'll assume that Shanidar 3 would have been attacked with equal probability by whomever he encountered. Hence, the encounter rate with modern humans, the chief unknown, directly determines the chance of being wounded by a modern human.

We do know something about the probability that a random injury from a high-energy Neandertal-made weapon would have the characteristics of the Shanidar 3 wound -- it's something less than 35 percent.

We'll assume the probability of Levallois point stabs is zero, thereby stacking the deck in favor of a modern human killer.

We know that a random wound from a modern human projectile weapon would generate a wound having those characteristics with something like a 50 percent probability.

Plugging into Bayes' theorem, that means that an encounter rate of 1 modern human for every 28 Neandertals would yield a five percent chance that a modern human was the culprit. Did Shanidar 3 encounter one modern human for every 28 of his own? If not, we have to conclude its very unlikely -- less than five percent -- that his wounds were caused by a modern human.

At the moment, the best evidence from the site suggests that Shanidar 3 lived fifteen thousand years before any archaeological transition to the "low-energy" impactors. Which tilts the scales in favor of the alternative -- a slightly lower probability impact from a much higher probability weapon.

#### References:

Churchill SE, Franciscus RG, McKean-Peraza HA, Daniel JA, Warren BR. 2009. Shanidar 3 Neandertal rib puncture wound and paleolithic weaponry. J Hum Evol (early online) doi:10.1016/j.jhevol.2009.05.010