Help with Programming R_Combine and Combine Commands

1,092 views
Skip to first unread message

Brian D.

unread,
Mar 11, 2018, 3:23:19 AM3/11/18
to OxCal

Hi All:

I could use some help with programming using the R_Combine and Combine commands.  My understanding is that the former is for samples from the same 14C source (i.e., a split sample), and the latter is for two different samples from the same context.  I have the following scenario.

Three supine skeletons were excavated from the same context, and they had interlocking arms indicating that they died at essentially the same time.  Skel1 and Skel2 had single radiocarbon dates, while Skel3 had two dates from two separate samples (bones) from the skeleton.  The results were as follows: Skel1 - 2625 +/- 20, Skel2 - 2655 +/- 20, and Skel3 - 2700 +/- 20 and 2725 +/- 20.

How do you nestle the Combine command with the R_Command for this case?

Any help is appreciated.

Brian D.



MILLARD, ANDREW R.

unread,
Mar 11, 2018, 3:10:12 PM3/11/18
to ox...@googlegroups.com
Dear Brian,
Dear Brian,

You would nest the different combine commands like this:

Combine("Skeletons")
{
R_Date("Skel1 ", 2625, 20);
R_Date("Skel2", 2655, 20);
R_Combine("Skel3")
{
R_Date("Skel3a", 2700, 20);
R_Date("Skel3b", 2725, 20);
};
};

The result has poor agreement so you may need to consider other factors like bone turnover slowing with age or a reservoir offset which varies between individuals.

Best wishes

Andrew
--
Dr. Andrew Millard
Associate Professor of Archaeology,
Durham University, UK
e: A.R.M...@durham.ac.uk | t: +44 191 334 1147
Personal page: https://www.dur.ac.uk/archaeology/staff/?id=160
Scottish Soldiers Project: https://www.dur.ac.uk/scottishsoldiers
Facial reconstruction: https://www.youtube.com/watch?v=JXOL_kw9sB8





________________________________________
From: 'Brian D.' via OxCal <ox...@googlegroups.com>
Sent: 11 March 2018 07:23
To: OxCal
Subject: Help with Programming R_Combine and Combine Commands
--
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<mailto:oxcal+un...@googlegroups.com>.
For more options, visit https://groups.google.com/d/optout.

Christopher Ramsey

unread,
Mar 12, 2018, 5:12:12 AM3/12/18
to ox...@googlegroups.com
Yes - that will work. However, both R_Combine and Combine should really be used where the event is the same rather than just the same context (which could imply some duration). For a context that might span some time you should really use a Phase. What classifies as the 'same event' is somewhat qualitative: in practice it means within the resolution of the dating technique - so in practice here the resolution of the calibration curve at <5 years.

It would be useful to use R_Combine nested within Combine in the way Andrew suggests if you had a different reservoir effect or a different calibration curve for two coeval events.

Mathematically the main difference between R_Combine and Combine is that R_Combine does the combination first and then does the calibration, whereas Combine will do the calibration first and then the combination. The Combine function assumes that the errors from the different elements are completely independent - which is not the case if they are radiocarbon dates calibrated against the same curve. Combine will underestimate the error when applied to radiocarbon dates for this reason.

More generally Combine can be used with other types of date or where the date comes from a mixture of radiocarbon and other techniques.

Best wishes

Christopher
> To unsubscribe from this group and stop receiving emails from it, send an email to oxcal+un...@googlegroups.com.

Brian D.

unread,
Mar 24, 2018, 5:16:33 PM3/24/18
to OxCal

Andrew/Christopher:

Thank you for your responses.

I actually made an error in the initial description.

Skel1, Skel2 and Skel3 initially had the following radiocarbon ages:  2625 +/- 20 (humerus), 2700 +/- 20 (tibia) and 2725 +/- 20 (humerus), respectively.  Subsequently, I reanalyzed Skel1, which yielded 2655 +/- 20 (tibia).  Nonetheless, the modeling results still show a poor agreement.  I think the same-time-of-death scenario should be regarded as an 'event' rather than 'context'.  I am in the process of AMS dating other samples from the three individuals.

Andrew, as you noted as a possibility, I think this may be a case of bone turnover. The stable isotopes do not indicate any appreciable differences in diet.  Skel2 is the oldest of the three (estimated age 35-45), whereas the others are younger (I'm awaiting confirmation but believe Skel1 was the youngest at 20-30 years).  I know there has been attempts to model bone turnover for modern (bomb) time frame (e.g., Hedges et al., 2007) based on 1 year resolution for the bomb curve, but I am not aware of other attempts using the the radiocarbon curve which has a 5 year resolution. As an initial attempt, I was thinking of approximating (assuming), for example, bone turnover of 10 or 20 % for the initial four 5-year increments (i.e., say 0-5 years of age: 10%, 5-10 years: 10%, 10-15 years: 10%, and 15-20 years: 10%), and then 2 or 3 % thereafter (i.e., 20-25 years: 2%, etc.).  Does this sound reasonable?

Thanks again for your input.

Brian 
> To unsubscribe from this group and stop receiving emails from it, send an email to oxcal+un...@googlegroups.com<mailto:oxcal+unsub...@googlegroups.com>.

Christopher Ramsey

unread,
Mar 26, 2018, 11:27:02 AM3/26/18
to ox...@googlegroups.com
Brian

This does not really answer your question but within OxCal you can offset dates before combination. So if for example you think that a date is likely to be about 25 years older than context and with an uncertainty of about half the mean you could use N(25,12.5) or better (because it does not go negative) lnN(ln(25),ln(1.5)). With several of these you can then combine them as in:

Combine()
{
R_Date(3000,30)+lnN(ln(20),ln(1.5));
R_Date(2990,30)+lnN(ln(25),ln(1.5));
R_Date(3050,30)+lnN(ln(30),ln(1.5));
};

You do however have to have some information on bone turnover to justify this - that is something which others might be better advising on.

Best wishes

Christopher
> > To unsubscribe from this group and stop receiving emails from it, send an email to oxcal+un...@googlegroups.com<mailto:oxcal+un...@googlegroups.com>.

Bayliss, Alex

unread,
Mar 26, 2018, 12:18:17 PM3/26/18
to ox...@googlegroups.com
I had a go at working out the average offset of individuals of different ages and sex (from the Hedges et al. 2007 paper) in the Bayliss et al. 2013 (fig 2.23). But it is complicated because how the calendar year turnover changes to an offset of radiocarbon in the bone obviously depends on the concentration of radiocarbon in (contemporary) food sources. The subjects for the Hedges et al. 2007 were modern Australians, and we have no guarantee that they were eating no fish. Similarly, when I do FRUITS models to assess diet from stable isotopes, there is never none. There is almost always 2.0+/-2.0% marine or 2.0+/-2.0% freshwater (maybe it would be better if you had 4 or 5 isotopes, or if your isotopic baseline is more differentiated than that for European prehistory). When you add this additional uncertainty to your models, they will probably combine satisfactorily. Whether this is realistic is another matter.

Good luck!

Alex

Bayliss, A, Hines, J, Høilund Nielsen, K, McCormac, F G, and Scull, C, 2013 Anglo-Saxon Graves and Grave Goods of the Sixth and Seventh Centuries AD: A Chronological Framework, Society of Medieval Archaeology Monograph, London (Maney)




We help people understand, enjoy and value the historic environment, and protect it for the future. Historic England is a public body, and we champion everyone’s heritage, across England.

We have moved! Our new London office is at 4th Floor, Cannon Bridge House, 25 Dowgate Hill, London, EC4R 2YA.

This e-mail (and any attachments) is confidential and may contain personal views which are not the views of Historic England unless specifically stated. If you have received it in error, please delete it from your system and notify the sender immediately. Do not use, copy or disclose the information in any way nor act in reliance on it. Any information sent to Historic England may become publicly available.

Erik

unread,
Jul 7, 2018, 11:11:17 AM7/7/18
to OxCal
Hi Brian, interesting problem.
Christopher and Alex have some great suggestions. Since the dates still do not agree (with a full century difference), you may want to return to your original assumption that they died at the same time (though I agree that interlocking arms seems quite clear at first). At least in the Andes, it is common for people to open graves and add or remove body parts, which can be powerful totems. It is also common (in many cultures) to return to a grave and add family members who died later – and perhaps place them in certain position. This might be another reason for the difference in the dates. Because of this potential uncertainty in the timing of deposition of the bodies, I would suggest modeling these dates as a Phase, as Christopher mentioned.
Erik

Patricio

unread,
Jul 8, 2018, 3:17:45 PM7/8/18
to OxCal
Hi to all, I have a related situation and I need some opinions. From a burial I dated two objets asocciated to a newborn child: a basket and a turban. Because of the caracteristics of the context (no perturbation, primary position of objets around the body), I assume that the desposition of both objets is synchronous. Nevertheless, there is the possibility of some curation of this objets: particulary, the turban, like other textiles in the Andes, may be passed through generations before be desposited in the burial. So, ¿what modeling procedure could bring the more "realistic" range?¿R_Combine or Phase? I´m not able to model an offset, because I have no idea of the lifetime of the turban before deposition. The dates are 2,450 +/- 30 (turban) and 2,410 +/- 30 /Basket). Thanks for your suggestions. 

Patricio.

Erik

unread,
Jul 8, 2018, 3:35:10 PM7/8/18
to OxCal
Patricio –
In general, Phase is the more conservative choice because it assumes less.

You don't want R_Combine because that's for two pieces of a single bone or plant, for example.

Your other option is Combine, but it assumes that the wool or plant fiber to make the basket and hat were harvested at the same time (perhaps a different time from when the finished objects were deposited), probably by people who wanted to make a grave good close to the time of burial.
If the materials were harvested within about five years (as Christopher points out), you could call it a single event. This is not unreasonable and you could make a case for using Combine. But, as you point out, these objects could have been curated and made decades before burial. Since you can't rule this out, I would suggest Phase.

Hope this helps
Erik

Patricio Felipe De Souza

unread,
Jul 10, 2018, 10:51:24 AM7/10/18
to ox...@googlegroups.com
Erick, thanks for your opinion, maybe to use Phase is actually the better option in this case. Nevertheless, if one avoid considerations about curation of the objects, I understand (following Christopher) that R_Combine could be effectivilly used in a case like this, because both objets are terrestrial and pertain to the same reservoir. R_Combine is an option if objects require to be calibrated before combination with a different curve or mixed-curve (for example, if one object is a marine shell or a human bone and the other is a basket). That is correct? Or I'm missing something?

--
You received this message because you are subscribed to a topic in the Google Groups "OxCal" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/oxcal/XIi3Ts15-xk/unsubscribe.
To unsubscribe from this group and all its topics, send an email to oxcal+un...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.


--
Prof. Patricio de Souza Herreros
Departamento de Antropología
Facultad de Ciencias Sociales
Universidad de Chile
Capitán Ignacio Carrera Pinto 1045, Ñuñoa, Santiago, Chile
Teléfono: +56 229772472

Erik

unread,
Jul 11, 2018, 12:12:48 PM7/11/18
to OxCal
As far as I understand it, R_Combine is for two samples from the same reservoir type (terrestrial v. marine) AND that died at the same time (as reservoirs vary over time). This could only very rarely apply to two separate objects. Usually it is used for two bones from a single individual, for example. As the manual puts it:

"If the dates are all from the same sample or object then the radiocarbon dates should be combined before calibration (R_Combine)."

"Radiocarbon dates that all relate to the same event (in other words where all of the samples should should all derive from the same radiocarbon reservoir at the same time) can be combined together before calibration."

This issue has been addressed in a number of forum discussions:

Patricio De Souza

unread,
Jul 12, 2018, 3:17:20 PM7/12/18
to OxCal
Erick, yes, you are right, time of death should be the same to use R_Combine, because reservoirs varies over time. I have it clear now. Thanks for your time responding me. 

Best regards,

Patricio.
Reply all
Reply to author
Forward
0 new messages