More about radiocarbon dating
By David Thulman
The great promise of radiocarbon (C14) dating is that it provides a method for dating and sequencing specific prehistoric events. On the Aucilla projects, C14 dating is used as a method for weeding out unpromising sties that are either too old or too young. In addition, it provides a method for determining the relative chronologies at the various sites and their relation to sites elsewhere in the world. Are we working on some of the earliest human cultural remains in North America? C14 dating can help provide the answer.
Given the pervasive reliance on C14 dating in archeology, it is necessary to understand the limitations of C14 dating and how the results can be skewed and misused. A C14 date is not really a “date” at all; it is an estimation of the number of years it would take the radioactive carbon in a dead organism to decay to leave the amount of radioactive carbon actually found when a sample of that organism is analyzed.
For example, at the Aucilla River sites, we typically take samples of buried tree branches for C14 testing. (Wood is a very reliable material for C14 testing.) The ratio of C14 and C12 in that branch is then compared with the ratio of C14 and C12 in a modern standard, and an estimate is made of the C14 remaining in the branch. If half the amount of C14 in the modern standard is left in the branch, the branch should be about 5,730 years old. I say “about” since the decay of C14 is random and the estimation of the amount of C14 is based in part on statistics. That is why C14 dates are always reported with a “±” margin of error. Typically, the margin of error reported is for one standard deviation from the norm. Therefore, a C14 date of 10,000 ± 200 BP on our branch sample means there is a 68% probability (a 2 in 3 chance) the branch died sometime between 9,800 and 10,200 years ago. A common practice is to report a C14 date as the single middle date (in our example, 10,000 years). This is misleading since there is actually an equal chance the true date of the branch will fall anywhere within the 400 year margin of error.
Carbon 14 dating is based upon a number of important assumptions, but only one will be discussed here. In order to compare C14 dates meaningfully, we must assume that all organisms contained the same amount of C14 when they died. Otherwise, organisms with less C14 will appear older because there will be less C14 than expected when the sample is tested. Unfortunately, that assumption is faulty.
As Mary Hudson explained in her Aucilla River Times article two years ago, C14 is created by cosmic radiation in the upper atmosphere. That radiation fluctuates year to year and therefore so does the creation of C14 . That means if our branch grew at a time when relatively lower levels of C14 were in the atmosphere, it would have less C14 when it died and would show an older apparent age than it should. Conversely, if it grew at a time of abundant C14 it would appear younger than it should. This differential C14 concentration may give our branch a younger C14 age than another branch that died hundreds of years after our branch, making comparison of the two samples misleading.
The only way to resolve this uncertainty is to calibrate the C14 dates with calendar dates. This calibration has been done by compiling a dendrochronological (tree-ring) record and painstakingly figuring the C14 age of these tree rings. This tree-ring record now extends back about 11,500 years, and by comparing the calendar age of the tree rings with their radiocarbon age, calibration curves have been created to produce a calendar date for a corresponding C14 date. The differential production of C14 produces “wiggles” in the calibration curves, and these wiggles can result in a single radiocarbon age corresponding to more than one calendar age.
There are presently a few computer programs available over the Internet that automatically calibrate C14 dates. The latest version of OxCal v.3 is fairly simple to use and produces information like the graph in Figure 1. The program can be downloaded from http://units.ox.ac.uk/departments/rlaha/oscal_h.html. By inserting a C14 date of 10,000 ± 100 for our branch, OxCal produced a range of calibrated dates with different confidences. Just like C14 dates, calibrated dates are given in a range. Note that as OxCal translates the lab date to a calendar date it also switches from BP (before present) to BC (before Christ). From Figure 1, we can be fairly confident that the date of our branch is between 9,700 cal BC – 9,280 cal BC, and very confident it falls between 10,200 cal BC – 9,200 cal BC. (“cal BC” means calibrated years before Christ as opposed to “BC” which means radiocarbon years before Christ.)
Carbon 14 dating has revolutionized archaeology by providing a method for dating events and allowing the comparison of events where previously their relative ages could only be indirectly inferred. However, it should be used with caution. Hopefully, even with its limitations, it will help us better understand the relation of our sites to the broader context of Paleoindian archaeology.
Editor's Note: David Thulman is an environmental attorney with the Florida Department of Environmental Protection. This article follows Mary Hudson's excellent description of the basis of radiocarbon dating in the April 1997 edition of the Aucilla River Times.