Last year, Davis et al. (2019) made a claim that the Cooper’s Ferry location in Idaho has archaeological remains older than 16 000 years. This would make them by far the oldest securely dated remains in the Americas. Manning (2020) just published a critique of their claim, accompanied by a response from Davis et al. (2020).

There are two basic disagreements between Manning and Davis et al. They disagree on how to use the Bayesian age estimation functions of the OxCal software package, and they disagree on which remains at Cooper’s Ferry are actually archaeological and therefore indicative of human activity.

Unfortunately, Davis et al. don’t engage the disagreement about which remains are archaeological, beyond simply reiterating that the bone and charcoal at the bottom of their LU3 layer are indeed associated with human activity. This, in my view, is the more important critique of the two and I hope they address it more fully in their future work.

The most obviously archaeological remains in LU3 are above 411.6 meters above sea level. Based on the published account, at least, there is still work to do to show that there are clearly archaeological remains at the bottom of LU3.

As for the disagreement on the finer points of age estimation in OxCal, Davis et al.’s response to Manning’s critique amounts to a fairly withering RTFM. Basically, they disagree on whether the boundary function or the first query is most appropriate for estimating the oldest age of the archaeological remains in layer LU3.

Davis et al. are interested in knowing how old Cooper’s Ferry could be, and so they use boundary. Manning is interested in the age of the oldest securely dated archaeological material, and so uses first. Not surprisingly, first gives a more conservative age estimate, to which Davis et al. vociferously object.

Focusing on the most obviously archaeological material and using the first query, Manning estimates the age at (calibrated Before Present) 15,935 ± 75 to 15,130 ± 20 using one subset of dates, and 15,735 ± 20 to 14,740 ± 90  with a slight different subset. Davis et al, using the boundary function and all the dates in LU3, have it at 16,560 to 15,280.

Both Manning and Davis et al. acknowledge that all these estimates, whether they result from boundary or from first, are very early and very interesting, but Manning’s estimate, which creeps into the 14k range, is a lot less shocking that Davis et al’s, which goes into the mid 16k range.

Instead of arguing which function is better, I think it might be more productive to explore this parameter space a bit further. This is what modeling is good at.

The boundary function assumes a uniform distribution of events over time. OxCal also has tau_boundary and zero_boundary functions which assume an increasing density of events over time. In the case of Cooper’s Ferry, or for any very early activity, this might be quite appropriate. The tau_boundary function assumes exponential growth in activity over time, and zero_boundary assumes a more linear growth.

Method (OxCal files attached at the bottom of this page)

I compared the age estimates given by boundary, tau_boundary, zero_boundary, and first , using two sets of dates. I used the full set of dates claimed as archaeologically relevant by Davis et al., except for the OSL dates, whose error range covers pretty much the entire period of interest. The OSL dates are nice to have, but not terribly useful for the specific purpose of estimating the oldest age of LU3.

I didn’t run the OxCal general outlier model used by Davis et al. to adjust the contribution of each date to the age estimate. I simply excluded the three LU3 dates that are clear outliers. The other dates have excellent agreement, and the outlier model makes little difference for them.

I also built a more restricted set of dates that focuses on what Manning and I seem to agree are the more archaeologically relevant remains. For that set, I excluded the LU3 dates that are from below 411.6 meters above sea level, since I don’t believe they are securely associated with human activity at the moment. I could also have assigned low prior probability to the dates from below 411.6 meters, but it would have been so low that it would not have been different from excluding them in practice.


Calibrated BP Full set Strict set
Boundary 16 608 – 15 312 16 380 – 14 569
Tau boundary 17 092 – 13 582 17 643 – 13 073
Zero boundary 19 227 – 15 570 18 862 – 14 910
First 15 848 – 15 295 15 165 – 14 628


As expected, running boundary on the full set of dates gives a result very similar to what is reported by Davis et al., even though the OSL dates and the outliers are ignored. While running boundary on the stricter set of dates doesn’t much affect the earliest boundary, it does bring the youngest age into the 14k range, which is an important detail to note.


Figure 1: boundary on restricted set

Running tau boundary on both sets allows a much greater possible age range. Assuming exponential growth of activity at the site allows estimates as old as the mid 17k range, but also as young as the mid to late 13k, which would be quite in line with previously known sites of the same kind.


Figure 2: tau boundary on restricted set

Playing around with the exponent of the tau function, obviously, would allow different age range estimates, but I just used the default value in OxCal. The take-home message is that assuming exponential growth allows this group of dates to give estimates that reach both farther into the past and come closer to the present.

Running the first query on the strict set brings us into the mid-14ks, which would be only a mildly shocking estimate, but still a very interesting one.

In the end, I have to agree with Manning that “more and better evidence is required to be confident of human presence at Cooper’s Ferry before ~15,000 years ago.” I would even say that I am not convinced there is archaeological material older than about 14.5k right now.

What is needed is much more convincing and more obviously archaeological material in the older parts of LU3. Perhaps that material is there already, but based on the published account, I don’t see it.

OxCal Files: CoopersFerryBoundary


Davis LG, DB Madsen, L Becerra-Valdivia, T Higham, DA Sisson, SM Skinner, D Stueber, AJ Nyers, A Keen-Zebert, C Neudorf, M Cheyney, M Izuho 2019 Late Upper Paleolithic occupation at Cooper’s Ferry, Idaho, USA, ~16,000 years ago, Science 365, 891–897.

Davis LG, L Becerra-Valdivia, DB Madsen, T Higham 2020. Response to Comment on “Late Upper Paleolithic occupation at Cooper’s Ferry, Idaho, USA, ~16,000 years ago”

Manning SW 2020. Comment on “Late Upper Paleolithic occupation at Cooper’s Ferry, Idaho, USA, ~16,000 years ago”, Science 10.1126/science.aaz4695

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s