Combine and outlier

[Chris Chinnock (Chris)]

Hi all, 

Im struggling to get my head around the following problem and would appreciate some help. 

I have a few dates from a small group of early medieval furnaces/ore roasting pits. Historically, 3 dates were sent to Beta for analysis but I had no detail on their species. I subsequently had them confirmed as oak and wanted to confirm the dates, anticpating possible old wood effect. I got two new dates on short-lived samples from two of the same contexts (31) and (74). The two dates from (31) are consistent and I have used combine given they are assumed to relate to the same firing event. The two dates from (74) are not consistent, with the confirmed short-lived sample actually being older than the initial date from Beta. Im not sure how best to treat this older date 'FTMC-QG69-34 (74)'. They are from the same context and believed to be from the same or near contemporary firing events but as they aren't consistent, am I right in thinking I cannot/should not model them with 'combine'? Left within the phase but with no further constraints there is a good general agreement but I had thought that this date probably should be marked out using 'outlier'. Im just not fully confident on how best to apply it in this case. 

Code below:

 Plot()
 {
  Sequence()
  {
   Boundary("Start Furnaces");
   Phase("Furnaces")
   {
    Combine("Fill (31) of Pit [32]")
    {
     R_Date("FTMC-QG69-33 (31)",1253,29);
     R_Date("Beta-479936 (31)",1260,30);
    };
    R_Date("Beta-479938 (58)",1420,30);
    R_Date("Beta-479937 (74)",1390,30);
    R_Date("FTMC-QG69-34 (74)",1512,29);
   };
   Boundary("End Furnaces");
  };
 };

Many thanks

Chris

[MILLARD, ANDREW R.]

Hi Chris,

 

Outlier models should represent beliefs before obtaining the dates, so they apply to all your samples or none, as you don’t know which is an outlier before making the measurement. As these are all charcoal samples and there is always a risk of an old wood effect it might be appropriate to apply a charcoal outlier model to all of them, which would allow for them to be older than their context, but not younger. Personally, given the possibility of an old wood effect or multiple occasions of use of the pits, even if not visible in the measurements, I would not combine the samples in (31) but leave them as separate dates within a phase.

 

Best wishes 

Andrew 

-- 

Prof. Andrew Millard 

Department of Archaeology,

Durham University, UK 

Email: A.R.M...@durham.ac.uk  

Personal page: https://www.durham.ac.uk/staff/a-r-millard/ 

Dunbar 1650 MOOC: https://www.futurelearn.com/courses/battle-of-dunbar-1650 

 

 

From: 'Chris Chinnock (Chris)' via OxCal <ox...@googlegroups.com>
Sent: 05 August 2025 15:46
To: OxCal <ox...@googlegroups.com>
Subject: Combine and outlier

 

[EXTERNAL EMAIL]

--
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/7c1cb4e9-5666-4d8f-9e08-10ba1ec41aban%40googlegroups.com.

[Erik Marsh]

Hi Chris, I think Andrew's right about not combining those dates post-hoc. I would model it like this, tagging all dates as Charcoal outliers.

Grouping them as a KDE_Plot works the same as Phase, but also gives you a summary KDE plot.

The boundaries give you a good idea of when the wood was harvested (that was later used in the furnaces).

If you think the oldest date is not very close in time to the use of the furnaces, you'll need to constrain it with other info, perhaps dates from adjacent contexts or historic information. It seems like they could have been used multiple times over the centuries – hard to say with five dates.

Hope this helps,

Erik

Plot()
{
Outlier_Model("Charcoal",Exp(1,-10,0),U(0,3),"t");

Sequence()
{
Boundary("Start Furnaces");

KDE_Plot("Furnaces")
{
R_Date("FTMC-QG69-33 (31)",1253,29) { Outlier(1); };
R_Date("Beta-479936 (31)",1260,30) { Outlier(1); };
R_Date("Beta-479938 (58)",1420,30) { Outlier(1); };
R_Date("Beta-479937 (74)",1390,30) { Outlier(1); };
R_Date("FTMC-QG69-34 (74)",1512,29) { Outlier(1); };
};
Boundary("End Furnaces");
};
};

[Seren Griffiths]

Hi Chris

As well as agreeing with everything that has been said, I think that Erik’s point about what how representative the 14C measurements are of the underlying archaeological activity; given those samples with potential old wood effects, and those short lived samples, and the stratigraphic relationships you describe, the 14C results might suggest a much more multiphase and/or long lived focus of activity…

Can you commission more measurements to explore?

All the best 

S

Prof Seren Griffiths

MA (Oxon.), MSc, PhD, MCIfA, FSA

BBC/AHRC New Generation Thinker 2020

To view this discussion visit https://groups.google.com/d/msgid/oxcal/3c40d5d7-68e9-422c-89c5-4dc72aad42bcn%40googlegroups.com.