Thank you for the work you have done. I have found it very helpful in becoming familiar with how Oxcal works. I would like to ask you a question or two about the use of Oxcal and which model would be most appropriate for the following situation. I have seven radiocarbon dates from an historic Alaska Native archaeological site. Four of them are from two house pits that have been excavated. The house pits are from the same component and the radiocarbon dates all overlap nicely. Consequently, the site has been interpreted as having a single component. However, the site has been used subsequent to the abandonment of these same houses as evidenced by a cash pit that had been dug into the roof of one of the collapsed structures. The remaining three dates are from the spoil pile of a vandalized midden at the site. The midden has not been excavated. The three midden dates cluster together but are later than the oldest three house pit dates, with the 95.4 confidence interval the the midden dates only with that of the youngest house date. I am uncertain if the midden is contemporary with the houses or not. Now the questions. Which model would be the best for presenting a suitably conservative estimate on the age of the midden? Should I include the house pit dates in the analysis of the midden dates or should I treat the midden as independent?
It's a bit difficult to answer this without knowing all the facts but, if you have strong cause for believing the midden is later in date (besides the simple fact that your dates are later), I would model it accordingly. You could try both abutting and "gap" models and see how much difference that makes. Keep in mind that you shouldn't just go with the one that you like best. Instead, show the results for both models (assuming that they give you different results). With a "gap" model, you can also estimate the time difference between the use of the house pits and the use of the midden to see if it's likely to be large or small. If it seems pretty small, that would make an abutting model more reasonable than it would otherwise be.
Thank you for this interesting tutorial on Chronometric dating and OxCal use. I am currently trying to use the software to create an accurate chronostratigraphic model for a series of OSL ages. Since I have some slight age reversals, I would like to use Bayesian analysis to solve these issues. I would be grateful if you could direct me on how to use the software to create a model for my OSL data.
As long as you have a stratigraphic sequence for your dated specimens, it really doesn't matter if they're radiocarbon or OSL. You should be able to put them into a model. Of course, only the radiocarbon dates would need calibration. You'd enter the OSL dates without calibrations. You might want to consult a Google group that deals with this topic: groups.google.com/g/oxcal/c/1wbghTC3TFk?pli=1
Hey Question, what is the difference between modelled and unmodelled? Also, when I use Marine20 as my curve, and have a Delta_R, Why does the Delta_R get it's own row in the table? isn't it just a part of the calibration calculation?
As for the delta-R, it's important for people to know which value was used for the calibration. As for the modelled dates, those are the ones that depend on the parameters of your model (the second set of "dates" in the OxCal output), and, if the model makes much difference, they'll always have narrower credible intervals than the unmodelled ones. The latter only take the calibration curve into consideration, and ignore the stratigraphic order and any other prior information. That's why people need to know if your result is modelled or not.
I wouldn't want to characterize it exactly that way, as an uncalibrated "date" is not really a date at all: it's just a statement about how much carbon-14 is left in the material. More colloquially, archaeologists talk about "radiocarbon years" but the main point is that they're not actually years. Over most of the last 10,000 years, though, it is true that the uncalibrated "age" is a smaller number than the actual age in calendar years, but there are exceptions, as in the post-bomb era.
Thanks for this guide, its been very helpfull! i keep wondering tho, why does OxCal not show the use of sigma1 or sigma2 on these calibrations, is there something im missing here? thanks again for your work
Also, if its not much trouble, could you indicate me the way to properly introduce a Delta R for the Marine 20 Curve, cause i have not been able to do so
You might want to query the Oxford group about this. They always default to 95.4% credible intervals and only have check boxes for that and, I believe, 68.2%. BCal, by contrast, allows you to pick any credible interval you want.
Just go to the OxCal main web page (c14.arch.ox.ac.uk/oxcal.html) and click on the "Download OxCal" icon in the bar at right. However, unless you're using a pretty fast computer, or are only running pretty simple models, you might want to do your analysis online instead (click on "OxCal Online").
I don't get it, how does it make my C14 dated fossil more accurate that some dendrochronology has been performed somewhere in the same region as my fossil, how will that increase any precicion for the date whatsoever? :/
The reason is that tree-ring sequences have been well developed in some regions, and we can carbon date the tree rings of known date. That allows us to make corrections to radiocarbon dates so that they're a lot more accurate (calibration). Precision is a different matter. That depends on a number of factors, including the way the carbon was analyzed, and where the results end up on the calibration curve (steep or plateau or wiggle regions).
@@thearchaeologistslaborator6591 yeah thanks but I don't get it: if I find an old shoe laying somewhere in the south of Finland, how does it help that I callibrate it using the Intcal20 program, since all that's specified in that program seems to be the northern atmosphere, in contrast to SHcal which is the south? If I type in my carbon age for the shoe in IntCal20, how does the callibration accuracy go up if the program doesn't know which exact forest and at which exact spot the shoe was find? I don't think that information is shared in IntCal at all... So pretty much my question: why does it matter to my shoe radiocarbon dating found in southern Finland if some some guys measured some threes somewhere in the northern hemosphere, of it's not the exact spot where I did find my shoe?
The accuracy does not go up (well, only in the sense that the calibration corrects for all the biases in uncalibrated radiocarbon dates), but the calibration can either increase, or decrease, precision, usually the latter, because of wiggles and plateaus in the calibration curve. As Finland is in the northern hemisphere, the IntCal20 calibration for the northern hemisphere will work well. It doesn't matter which forest or where the shoe was found (other than being in the northern hemisphere), because the amount of radiocarbon in the entire northern hemisphere in any given year is pretty much the same because of mixing by winds, and all the trees, cows, people, grass, etc. absorb that radiocarbon. In fact, the forest is irrelevant. The calibration curves are built up from tree-ring sequences from several parts of the world, including, for the northern hemisphere one, the United States, Ireland, Germany and elsewhere. The reason for the calibration, mostly on the basis of tree rings of known date, is that there are all kinds of biases in the uncalibrated radiocarbon determinations, mostly related to the incorrect assumptions that Libby made, and especially to the assumption that the abundance of radiocarbon in the atmosphere was pretty constant. We now know that radiocarbon abundance fluctuates quite a lot, and also has long-term trends. Fortunately, because we can make radiocarbon determinations on tree rings of known date, we can build these calibrations that solve those problems.
@@tedbanning9090 Thanks a lot for the interesting answer, Ted. But then one thing has to be determined: how can dendrochronology callibration be accurately done? I've heard they do it by extracting the cellulouse from the the cell walls of the wood. But does that mean that they take out the cellulouse of every single tree ring layer of every single individual tree sample, and then measure the rate based on that? I wonder if you who seem knowledgeable could expound a little on this subject, to someone who isn't and haven't found much helpful material online...?
I'm not sure I'm understanding your question fully, but they used to date groups of tree rings of known date (say 5 or 10 consecutive rings), which obviously isn't totally satisfactory because it gives a kind of average over that period. But at one time it was the only way to get enough carbon for a reasonable date. Now, with high-efficiency AMS dating, it's typically possible to date rings individually, and this is done for tree rings from samples taken in different parts of the world where we have good dendrochronological sequences, sometimes going back thousands of years. Obviously this is slow and somewhat expensive, but eventually it builds up a very large sample of results that can be analyzed to produce the calibration curve (they use Bayesian statistical methods to do this). But the key feature is that the tree rings have known dates, because we know these trees produce one new ring every year, and we can just count the rings backwards, and current technology allows us independently to date indigenous carbon in those same rings by radiocarbon dating to find out how much the radiocarbon dates for a given year diverge from the known date.
It's so very good to listen somebody to explain exactly bayesian analysis
Thanks. I think there are some very good ones on Bayesian analysis more generally. I'll try to add some links to them if I haven't already.
Thank you for the work you have done. I have found it very helpful in becoming familiar with how Oxcal works. I would like to ask you a question or two about the use of Oxcal and which model would be most appropriate for the following situation.
I have seven radiocarbon dates from an historic Alaska Native archaeological site. Four of them are from two house pits that have been excavated. The house pits are from the same component and the radiocarbon dates all overlap nicely. Consequently, the site has been interpreted as having a single component. However, the site has been used subsequent to the abandonment of these same houses as evidenced by a cash pit that had been dug into the roof of one of the collapsed structures.
The remaining three dates are from the spoil pile of a vandalized midden at the site. The midden has not been excavated. The three midden dates cluster together but are later than the oldest three house pit dates, with the 95.4 confidence interval the the midden dates only with that of the youngest house date. I am uncertain if the midden is contemporary with the houses or not.
Now the questions. Which model would be the best for presenting a suitably conservative estimate on the age of the midden? Should I include the house pit dates in the analysis of the midden dates or should I treat the midden as independent?
It's a bit difficult to answer this without knowing all the facts but, if you have strong cause for believing the midden is later in date (besides the simple fact that your dates are later), I would model it accordingly. You could try both abutting and "gap" models and see how much difference that makes. Keep in mind that you shouldn't just go with the one that you like best. Instead, show the results for both models (assuming that they give you different results). With a "gap" model, you can also estimate the time difference between the use of the house pits and the use of the midden to see if it's likely to be large or small. If it seems pretty small, that would make an abutting model more reasonable than it would otherwise be.
Thank you for this interesting tutorial on Chronometric dating and OxCal use. I am currently trying to use the software to create an accurate chronostratigraphic model for a series of OSL ages. Since I have some slight age reversals, I would like to use Bayesian analysis to solve these issues. I would be grateful if you could direct me on how to use the software to create a model for my OSL data.
As long as you have a stratigraphic sequence for your dated specimens, it really doesn't matter if they're radiocarbon or OSL. You should be able to put them into a model. Of course, only the radiocarbon dates would need calibration. You'd enter the OSL dates without calibrations. You might want to consult a Google group that deals with this topic: groups.google.com/g/oxcal/c/1wbghTC3TFk?pli=1
@@thearchaeologistslaborator6591 Thank you for the quick reply - I will check this out
Hey Question, what is the difference between modelled and unmodelled? Also, when I use Marine20 as my curve, and have a Delta_R, Why does the Delta_R get it's own row in the table? isn't it just a part of the calibration calculation?
As for the delta-R, it's important for people to know which value was used for the calibration.
As for the modelled dates, those are the ones that depend on the parameters of your model (the second set of "dates" in the OxCal output), and, if the model makes much difference, they'll always have narrower credible intervals than the unmodelled ones. The latter only take the calibration curve into consideration, and ignore the stratigraphic order and any other prior information. That's why people need to know if your result is modelled or not.
Usually calibration increases age?
Am I correct?
I wouldn't want to characterize it exactly that way, as an uncalibrated "date" is not really a date at all: it's just a statement about how much carbon-14 is left in the material. More colloquially, archaeologists talk about "radiocarbon years" but the main point is that they're not actually years. Over most of the last 10,000 years, though, it is true that the uncalibrated "age" is a smaller number than the actual age in calendar years, but there are exceptions, as in the post-bomb era.
Thanks for this guide, its been very helpfull! i keep wondering tho, why does OxCal not show the use of sigma1 or sigma2 on these calibrations, is there something im missing here? thanks again for your work
Also, if its not much trouble, could you indicate me the way to properly introduce a Delta R for the Marine 20 Curve, cause i have not been able to do so
You might want to query the Oxford group about this. They always default to 95.4% credible intervals and only have check boxes for that and, I believe, 68.2%. BCal, by contrast, allows you to pick any credible interval you want.
@@dmcupitty Hi. You'll find the tool for entering your Delta R values under the "Insert" menu in OxCal. It's right after "Date" in the drop-down menu
how to download oxcal software
Just go to the OxCal main web page (c14.arch.ox.ac.uk/oxcal.html) and click on the "Download OxCal" icon in the bar at right. However, unless you're using a pretty fast computer, or are only running pretty simple models, you might want to do your analysis online instead (click on "OxCal Online").
I don't get it, how does it make my C14 dated fossil more accurate that some dendrochronology has been performed somewhere in the same region as my fossil, how will that increase any precicion for the date whatsoever? :/
The reason is that tree-ring sequences have been well developed in some regions, and we can carbon date the tree rings of known date. That allows us to make corrections to radiocarbon dates so that they're a lot more accurate (calibration). Precision is a different matter. That depends on a number of factors, including the way the carbon was analyzed, and where the results end up on the calibration curve (steep or plateau or wiggle regions).
@@thearchaeologistslaborator6591 yeah thanks but I don't get it: if I find an old shoe laying somewhere in the south of Finland, how does it help that I callibrate it using the Intcal20 program, since all that's specified in that program seems to be the northern atmosphere, in contrast to SHcal which is the south?
If I type in my carbon age for the shoe in IntCal20, how does the callibration accuracy go up if the program doesn't know which exact forest and at which exact spot the shoe was find? I don't think that information is shared in IntCal at all...
So pretty much my question: why does it matter to my shoe radiocarbon dating found in southern Finland if some some guys measured some threes somewhere in the northern hemosphere, of it's not the exact spot where I did find my shoe?
The accuracy does not go up (well, only in the sense that the calibration corrects for all the biases in uncalibrated radiocarbon dates), but the calibration can either increase, or decrease, precision, usually the latter, because of wiggles and plateaus in the calibration curve. As Finland is in the northern hemisphere, the IntCal20 calibration for the northern hemisphere will work well. It doesn't matter which forest or where the shoe was found (other than being in the northern hemisphere), because the amount of radiocarbon in the entire northern hemisphere in any given year is pretty much the same because of mixing by winds, and all the trees, cows, people, grass, etc. absorb that radiocarbon. In fact, the forest is irrelevant. The calibration curves are built up from tree-ring sequences from several parts of the world, including, for the northern hemisphere one, the United States, Ireland, Germany and elsewhere. The reason for the calibration, mostly on the basis of tree rings of known date, is that there are all kinds of biases in the uncalibrated radiocarbon determinations, mostly related to the incorrect assumptions that Libby made, and especially to the assumption that the abundance of radiocarbon in the atmosphere was pretty constant. We now know that radiocarbon abundance fluctuates quite a lot, and also has long-term trends. Fortunately, because we can make radiocarbon determinations on tree rings of known date, we can build these calibrations that solve those problems.
@@tedbanning9090 Thanks a lot for the interesting answer, Ted.
But then one thing has to be determined: how can dendrochronology callibration be accurately done? I've heard they do it by extracting the cellulouse from the the cell walls of the wood.
But does that mean that they take out the cellulouse of every single tree ring layer of every single individual tree sample, and then measure the rate based on that?
I wonder if you who seem knowledgeable could expound a little on this subject, to someone who isn't and haven't found much helpful material online...?
I'm not sure I'm understanding your question fully, but they used to date groups of tree rings of known date (say 5 or 10 consecutive rings), which obviously isn't totally satisfactory because it gives a kind of average over that period. But at one time it was the only way to get enough carbon for a reasonable date. Now, with high-efficiency AMS dating, it's typically possible to date rings individually, and this is done for tree rings from samples taken in different parts of the world where we have good dendrochronological sequences, sometimes going back thousands of years. Obviously this is slow and somewhat expensive, but eventually it builds up a very large sample of results that can be analyzed to produce the calibration curve (they use Bayesian statistical methods to do this). But the key feature is that the tree rings have known dates, because we know these trees produce one new ring every year, and we can just count the rings backwards, and current technology allows us independently to date indigenous carbon in those same rings by radiocarbon dating to find out how much the radiocarbon dates for a given year diverge from the known date.
What is the best calibration curve for India?
As India is in the northern hemisphere, as long as you're dating terrestrial material, you can use the IntCal20 curve for the northern hemisphere