OxCal calibration within 1650–1950: how strong are our inferences?

[Alex Valenzuela]

Hi all,

I was reading the 2020 review by Puckett, Orton & Munshi-South (BioEssays, doi: 10.1002/bies.201900160), where they note that radiocarbon dating becomes less useful for the period between the 18th and early 20th centuries due to the shape of the calibration curve. They suggest that for things like Rattus norvegicus introductions into Europe, it’s hard to get precise dates via ¹⁴C because of this.

I hadn’t come across this limitation before, but it caught my attention because I work with some post-medieval and modern sites, and I’m unsure whether dating those contexts would be worthwhile. In fact, I already have a few radiocarbon results from that time range, and now I’m starting to question how meaningful they really are.

For example, one of my samples gives a calibrated OxCal result for 230 ± 22 BP, with the following 95.4% probability intervals:
• 52.1%: 1639–1680 calAD
• 2.9%: 1740–1753 calAD
• 39.7%: 1762–1800 calAD
• 0.7%: 1940–1945 calAD

This seems to suggest that I can say with reasonable confidence (>94%) that the event predates 1800. So now I’m a bit confused: does the limitation discussed in that paper apply more to distinguishing between dates within the plateau (e.g., 1760 vs. 1790), rather than making broader inferences like “this is probably pre-1800”?

Would love to hear how others approach this. Is it fair to state a terminus ante quem (e.g., “before 1800”) based on this kind of skewed calibration distribution?

Thanks!
Alex Valenzuela

--

--------------------------------------------------------------

Dr. Alejandro Valenzuela Oliver

Mediterranean Institute for Advanced Studies (IMEDEA)
Ecology and evolution (EEG)
C/ Miquel Marquès, 21
07190 Esporles
Illes Balears, SPAIN.
https://imedea.uib-csic.es/en/the-institute/staff/?staff_id=778

https://www.researchgate.net/profile/Alejandro_Valenzuela3

orcid.org/0000-0001-6120-6246

--------------------------------------------------------------

-----------------------------------------------

[Erik Marsh]

Hi Alex – 

I think this is mostly a complaint based on looking a single calibration – but there are plenty of other similar plateaus in the calibration curves.

You're right, at a narrow scale of decades, these plateaus don't allow us to be very precise. At the scale of centuries, it's not really a problem.

More importantly, and I think this forum will agree 😉, is a Bayesian model that adds in other priors and multiple dates. Modeling dates with priors, as a group, is the recipe for overcoming calibration plateaus (Manning et al. 2020; Meadows et al. 2020; Rose et al. 2022). For historic contexts in South America, we have this same problem – European artifacts allow us to introduce priors that address this. Similar to what you suggest, we use a pre-1532 prior, in a Bayesian model, and this gives you a new probability range that takes this into account. Using the Order or Difference queries, OxCal can give you more precise probabilities that an event was pre-1800, for example.

Hope this helps, Erik

[Alex Valenzuela]

Hi Erik,

Thanks so much for the quick reply and for the suggestions. What you mention is really interesting. I’ll definitely take a look at the works you referenced and explore those approaches further.

Much appreciated!

Best regards,
Alex 

Missatge de Erik Marsh <erik....@gmail.com> del dia dt., 27 de maig 2025 a les 19:21:

--
You received this message because you are subscribed to the Google Groups "OxCal" group.
To unsubscribe from this group and stop receiving emails from it, send an email to oxcal+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/oxcal/fefb8166-5d9f-4de5-8fe5-ec37a237bb9fn%40googlegroups.com.