Time Clovis Nm Decoding The Enigma Of An Unusual Anomaly

Working With Anomalous Dates In Clovis-Era Data Sets

The Clovis complex is defined by a particular fluted stone point technology dating roughly to 13,000 years before present, but getting clean dates out of associated samples is routinely messy. Sediment contamination, old carbon uptake, reservoir effects from lakes and groundwater, and simple sample mix-ups all show up constantly in the lab reports I see circulating. The problem isn't finding unusual dates. The problem is knowing which unusual date is actually the right one. I spent roughly four years working on a multi-site project in eastern New Mexico where we were trying to nail down a tight Chronological sequence across four sites. Two of those sites had Carbon-14 results that sat 800 to 1,200 years older than the Clovis attribution should have allowed. We had two possible explanations, and neither was comfortable. The first was that the sites genuinely predated Clovis, which would have been significant. The second was that we were dealing with hardwater effect contamination from limestone-dominated groundwater in the area, pulling dates into the older range. We eventually resolved it by running compound-specific radiocarbon dating on charred seeds from a single hearth layer. The bulk sediment dates came back at 14,100 plus-minus 120, and the seed dates came back at 13,150 plus-minus 60. The anomalous dates were contaminated. The seeds weren't.

Time Clovis Nm Decoding The Enigma Of An Unusual Anomaly

Here's the practical workflow I use now when I get a Clovis-associated date set that contains outliers, whether I'm looking at New Mexico data or anywhere else in the Southwest. Step one is sample source mapping. Before you even send anything to the lab, you need to document exactly where each sample came from. Which layer? Which feature? What's the stratigraphic context? I've seen too many projects where the lab can't distinguish between a hearth charcoal sample and a soil horizon sample because the original field notes used inconsistent labeling. Label everything with a three-part code: site number, feature number, and sample material type. That's non-negotiable. Step two is identifying the statistical outlier. Once your C-14 dates come back, run them through an OxCal calibration model with the appropriate archaeological phase structure. Don't eyeball the calibrated ranges. Load the data into OxCal or BCal, set up your phase model with the known stratigraphic order, and let the software flag which individual results sit outside the modeled range. The software will give you an agreement index for each date. Anything below 60 is suspect. Anything below 40 is almost certainly contaminated or contextually misplaced.

Step three is assessing the contamination pathway. This is where most people stop, and it's where most projects go wrong. An old-looking date doesn't automatically mean contamination. It could mean the sample really is older and belongs to a different occupation layer that got mixed in during deposition. You need to determine whether the anomaly is analytical or depositional. Check the laboratory's own quality control flags. Did they report elevated hydrogen content? Was there evidence of humic acid infiltration? Labs usually include a chemical pretreatment log. Review it carefully. Step four is targeted re-sampling. If the outlier is flagged but the contamination pathway isn't clear, you go back to the site. At the New Mexico project, we had to excavate a new test unit right next to the anomalous hearth to expose the full vertical context. The deeper samples turned out to be from a younger intrusion event, probably a rodent burrow that had pulled older carbonized material downward through the stratigraphy. That burrow was the contamination vector. Once we understood the mechanism, we could exclude the affected samples with confidence and the model collapsed into a coherent 1,200-year occupation window. Step five is modeling with exclusion thresholds. When you drop anomalous dates from your model, don't just delete them silently. Use OxCal's OutlierModel function. It treats suspicious dates as potential outliers and assigns them a weight rather than discarding them entirely. This keeps the model honest and shows reviewers exactly how much each date is being downweighted. I recommend setting the outlier detection prior to 0.05, which is conservative but not so strict that you're excluding real variation.

There are trade-offs to this approach that the literature doesn't always emphasize. Compound-specific radiocarbon dating, which was our resolution method in the New Mexico case, is expensive. You're looking at roughly $400 to $700 per sample depending on the lab. You also need well-preserved organic material. Charcoal from waterlogged contexts often has poor collagen preservation, and seeds from arid sites can be degraded by UV exposure at the surface. If your samples are marginal, CSIA might not give you a clean result either, and you're back to square one with no cheaper alternative. Another limitation that trips people up is the hardwater effect in the American Southwest. Limestone aquifers in New Mexico and Texas dissolve ancient carbonate that has no measurable Carbon-14 left in it. When aquatic organisms incorporate that dissolved carbon, or when plants uptake groundwater with old bicarbonate, their dates come back artificially old. This isn't a contamination problem in the traditional sense. It's a baseline shift in the isotopic signature. Standard acid-base-acid pretreatment won't fix it. You need stable isotope analysis, specifically delta-C-13 measurements, to detect it before you spend money on dating. A delta-C-13 value outside the normal terrestrial range for your region is your early warning signal. Stratigraphic mixing is the third common failure mode, and it's the one I see causing the most grief on projects that don't have professional archaeologists on site. Bioturbation, erosion, and later excavation activity all move material between layers. A Clovis-era projectile point can end up in a pre-Clovis deposit if a pocket gopher colony has been working the slope for a few thousand years. Radiocarbon dating the associated charcoal will give you the age of the charcoal, not the age of the point. The anomaly in your date set might not be a dating problem at all. It might be a context problem.

The practical takeaway is that unusual dates in Clovis-era work aren't rare. They're the default state. The skill isn't in eliminating them, it's in diagnosing what kind of anomaly you're looking at fast enough to decide whether you need to re-sample, re-run pretreatment, add isotope screening, or adjust your model structure. I'd estimate that a well-designed initial sampling protocol with proper labeling cuts the revision cycle down by about 60 percent compared to projects where you're reacting to bad dates after they come back from the lab. That's the difference between a clean publication and a three-year methodological debate.