Tuesday, August 05, 2008

Dating the Predynastic for beginners - Part 2

As previously, all constructive criticism gratefully received! Many thanks for the comments in response to Part 1. Once I have all the comments I'll update both posts and put them somewhere on the web as a single document.

The topic of relative dating was dealt with in Part 1, a previous post. A relative sequence was available for the period between the Badarian and the period when unification was thought to have taken place. However, although this sequence has formed the framework for all other sequences, no calander dates were available. The only date which was relevant was the proposed date of 3000 BC for the unification of Egypt. Apart from the fact that the Predynastic predated it, we would not be any the wiser about exact ages of these periods if it was not for radiocarbon dating.

I said that the major sources of confusion for that system were that

  • Two sets of terminology are still in use for Upper Egypt (Amratian = Naqada I and Gerzean = Naqada II)
  • Petrie’s original sequence has been refined several times and a number of different schemes exist. Although Kaiser’s 1956 and 1990 versions are the most commonly used, it is rare that an author will say which scheme is in use
  • Before radiocarbon dating it was impossible to know exactly how long these periods lasted, and how old they extended back to
  • A different set of terms is used to describe the technology and tradition of Lower Egypt during the Upper Egyptian periods of the Badarian, Naqada I and Naqada II., with which they were contemporary
  • Contemporary prehistoric cultures outside the Nile Valley in Egypt are usually treated completely separately, although there is often chronological overlap with the Badarian.
  • Dynasty 00 and Dynasty 0 are terms which cause a lot of confusion

The matter was particularly highlighted on the course by discussion about the date given for a newly discovered site in the Faiyum Depression, currently under excavation by UCLA. The site is near the Roman town of Karanis and apparently bears a very strong resemblance to the Faiyum A or Faiyum Neolithic sites which were found on the north short of Lake Qarun by Gertrude Caton Thompson and Elinor Gardiner in the 1920s. The Faiyum Neolithic is dated to 5450-4400BC. The new date which has appeared in a number of news articles for the new site is 7000BCE, which puts it more or less on a par with the previous occupation which are associated with an entirely Epipalaeolithic (hunting, gathering and fishing) industry known as the Qarunian which is dated to c.7050BC. This date clearly challenges quite a few assumptions if it is to be accepted. The UCLA team are currently preparing a report for the SCA and cannot release any data without the SCA‘s sanction, as per the SCA’s regulations. However, we will hopefully hear more details in the future about whether the date is a one-off or whether it is the average of a series of dates, and where the sample from which it was taken came from. But the discussion about that one date opened up the whole chronological can of worms.

I am going to focus on radiocarbon dating because it is the one used most frequently, and the one which seems to generate the most misunderstandings, but I will mention other relevant techniques at the end.

So what actually is radiocarbon dating? Radiocarbon dating (also written as Carbon 14, 14C or C14) was invented by Willard Libby in 1949. Carbon 14 is an unstable isotope which exists in the atmosphere with other isotopes and is absorbed by all living things (plant and animal). Because it is unstable it begins to decay at the time of the death of the organism that absorbed it. The rate of decay is regular and can Libby was able to measure this rate using a Geiger counter. Libby calculated how long it took for half of any carbon 14 in a sample to decay (it’s so-called “half life”). He also estimated that atmospheric levels of carbon 14 would remain the same throughout time. He realized that if this was the case the rate of decay of carbon 14 could be used on samples from ancient contexts to date those contexts. The organic materials on which it can be used include wood, charcoal, bone (but not cremated bone), plant remains, skin, etc. However, a sample must be at least 10g for wood and charcoal to be dated and at least 100g for bone. Of course one of the problems experienced by archaeologists is that it is not always possible to obtain a sample of this size (a point made by Dirk Huyge in his rock art lecture). Radiocarbon dates have a practical use in terms of reliability for samples which are between 200 and 10,000 years old. Reliability decreases after 10,000 years ago and becomes unusable for practical purposes after 40,000 years ago.

When Libby first released the details of his technique he caused a sensation, because for the first time archaeologists were able to obtain real calendar dates for artefacts and sites, rather than trying to fit them into often clumsy relative frameworks. Many of these frameworks were found to be too old or too young, and the face of archaeology was changed forever.

Radiocarbon dates are always allocated a unique laboratory reference which shows an abbreviated form of the laboratory’s name and the number assigned to the sample. Some academic publications are kind enough to add the lab reference which makes it easy to double check them or find out more information about the sample used. Many, however, do not. A typical lab code would be GrN-54432

Radocarbon dates are always expressed with a range of error e.g. 5500+/-100 bc. This is because the process of deriving a date from a sample comes with a probability factor of it being correct or incorrect of 68%. So the date 5500+/-1000 bc (5400 - 5600) has a 68% probability (just over 2 in 3) of being correct but also has a 1 in 3 chance of being wrong. In order to improve the likelihood of it being correct the standard deviation (in my example 100) can be doubled to obtain a 95% probability of the date range being correct. So instead of a range of 5400-5600 bc which has a 68% chance of being correct we have a date range of 5300-5700 bc, which has a 95% chance of being correct. Although this is not always a problem for prehistorians it is not particularly helpful where greater granularity of time is required!

Unfortunately life is never simple, and the radiocarbon dates being produced using Libby’s system on samples of known dates obtained from tree-ring dating (dendrochronology) were found to be too young. The effect was increased over time, so that some prehistoric dates were between 800 and 900 years too recent. The cause of the problem was that Libby had not realized that the concentration of C14 in the atmosphere has not remained constant throughout time - it has fluctuated. The answer to the problem lay in an unlikely place - the Bristlecone Pine. This specie of tree has been used to build up tree ring sequences, which provide accurate dendro-chronological dates back for a 5000 year period. By dating samples with a known date scientists were able to determine when and to what degree the carbon 14 fluctuations occurred and then use that data to work out a correction curve whereby radiocarbon dates could be calibrated by reference to the dendro dates.

These calibration curves look simple when you see them on paper but in practice they are far from straight forward, due to the inbuilt probability distribution of radiocarbon dates mentioned above. To produce a correctly calibrated date range requires the use of a highly specialized and complex statistical technique called Bayesian statistics (named after the man who developed it). The calculations required are so complex that they must be carried out by a computer programme. A number of programmes now exist, with some of them freely available online (e.g. OxCal). The uncalibrated radiocarbon date together with its probability range, are fed into the programme which then makes a calculation to produce a date which is expressed without a deviation. Instead, the calibrated dates are expressed as a possible range between two dates e.g. 4900-3400 BC.

Radiocarbon dates can be expressed BP or BC/BCE. BP is Before the Present, the present having been set for practical purposes at 1950. BC (Before Christ) and BCE (Before Calendar Era (the politically correct version of BC) both refer to the time before the modern calendar begins - the year 0. Hence 100BC is approximately 2108 years ago, whereas 100BP would only be AD 1850. In fact, historians wouldn’t use BP because their dates are much more fine-grained, but it was just easy to use it to illustrate the point!

But of course there is the complication of having uncalibrated dates and calibrated dates. Either can be used in publications, and sometimes both are shown. This has lead to considerable confusion. Fekri Hassan expresses it very well in his book Droughts Food and Culture: “One of the key methodological issues hampering interdisciplinary communication and correlation of environmental and cultural data is the lack of standardization in reporting radiocarbon dating” (2002 p.8).

In my undergraduate days the uncalibrated dates were expressed in lower case as bp and bc, whereas the calibrated dates were expressed BC and BP. This is somewhat error prone, so nowadays a more popular method is to express uncalibrated dates BC or BP and calibrated ones Cal BC or Cal BP. If you’re lucky your author will explain which conventions he or she is using at the beginning of the publication, but most don’t. I think that for those new to prehistory this is the main source of confusion - between bp, bc, BP, BC, Cal BP and Cal BC there is a lot of space for serious confusion - and it creeps into one or two publications in the form of typographical and editing errors. In his above-mentioned book Hassan is good enough to provide a list of standard abbreviations used in the text and a conversion table of radiocarbon age estimates, which are immensely useful. He also indicates which calibration programme that he used, which in this case was Stuiver and Reimer‘s Calib 3.0.3c programme. But Professor Hassan is the exception, not the rule.

Just to add to the fun there is now a new form of radiocarbon dating called Accelerator Mass Spectrometry carbon 14 dating. This is a refined version of radiocarbon dating and offers the same benefits of ordinary radiocarbon dating but has the massive benefit or requiring a much smaller sample.

Finally, radiocarbon dates are only as good as the people handling and interpreting them. There are a number of points in the life of a sample where its value can diminish - for example, in errors made in the laboratory analysis, and contamination from other samples. Background cosmic radiation can also impact the value of a sample. For this reason it is rarely wise to depend upon one radiocarbon date. In order for a site to be dated securely a range of dates should, in an ideal world, be taken. These give additional statistical validity to a date range, and any anomalies will show up quite clearly - the Nabta Playa dates are a good example of this. In the Nabta reports there are so many dates taken at sites in the area that the anomalies tend to show up quite clearly.

I mentioned at the beginning of this treatise (I had no idea that it would be so long when I sat down to type it on my lunch break two weeks ago) that there are other types of absolute / scientific dating technique available. These are usually used when radiocarbon dates are impractical - either because there are no organic remains present or because the date range of the site falls outside that for which radiocarbon is useful. I don’t want to completely overload anyone who was mad enough to get this far, but I thought that a couple of examples might be of use/interest.

Thermoluminescence dating (usually just abbreviated to TL) is used for dating fired clays - like pottery. It can be used to date items and baked clay layers from the present until 400,000 years ago but it is relatively unreliable when compared with radiocarbon dating so it is only used in circumstances when radiocarbon dating is unavailable I.e. when there are no organic remains or when the site is over 40,000 years old. Clay contains quartz crystals which experience radioactive decay which builds up to create an electrical charge. This charge builds up at a known rate. When an item made of clay is exposed to heat under lab conditions the light is released and this can be measured to calculate the length of time since it was fired (I.e. the tie at which the light was first released).

Potassium Argon dating (usually abbreviated to K-Ar) is similar in principle to TL. It is only of use with very old samples - 200,000 to several millions of years ago. But its other major limitation is that it requires the right sort of geological formations and can therefore be used only in certain parts of the planet. Rock crystals contain potassium, which produces the gas argon as it decays. Examination and measurement under laboratory conditions can reveal when the crystal was formed. It is only of use where rock lies above and/or below a sample to give an approximate date, a Terminus Ante Quem (a date after which) or a Terminus Post Quem (a date before which).

Uranium Series is potentially useful for obtaining dates from calcium-rich samples like teeth, shells and calcium carbonate deposits (stalagmites or stalactites or stalagmitic floors, generically and colloquially termed “stal”). Again it operates by knowing the decay rate of isotopes - in this case U235 and U238, which are water soluable and produce thorium and protactinium. It needs a high content of uranium in the sample to be of any use.

If anyone desperately wants me to explain how Pharaonic Egyptian chronology has evolved let me know and I’ll do a piece on that too, but not until after Bloomsbury.


Anonymous said...

This is what I knew about 14C. I hope to be right: I studied this argument more than 20 years ago and my mind is older now...

Carbon is the chemical element of life: all molecules useful for life processes contain carbon (DNA, proteins, sugars, fats, etc.).

Carbon has 3 isotopes: 12C (almost 99%), 13C [around 1%; it is useful in analytical organic chemistry: 13C and 1H-NMR (Nuclear Magnetic Resonance) are powerful complementary metods, with MS (Mass Spectometry), to determine molecular structures] and traces of 14C: it has very low percentual concentration, but in absolute value is an enormous number and this great concentrastion is considered to be costant (this is a reasonable approsimation; what else can we do?).

Talking about enormuos numbers of atoms / molecules, think for example that a glass of water contains 100 ml = 100 grams of H2O, that is equal to 33 x 10exp23 molecules (1 x 10exp6 = 1 million; 1 x 10exp9 = 1 billion)!

Even if 14C decays, losing a nuclear electron, we can consider, as said, its concentration costant and the decay is a tipical first order kinetic: it is relatively simple resolve the differential equation descriving this process and it is possible to calculate half life, that means time from death to half concentration of 14C in the sample (bones, papirus, ashes etc.).

I have a nice episode to say: years ago, an university of a city in Toscana (I can’t remember if it was Siena or Pisa) tried to date 4 horses of San Marco basilica in Venice.

The horses are in bronze, but in antiquity they were gilded.

Ancients used the tecnique of “amalgama” [chemists calls amalgama a mixture of one metal, gold (Au) in this case, and quicksilver (Hg)]: covering horses with Au / Hg and heating, Hg evaporates (vapours of this metal are extremly toxic; I can’t think of poor workers lungs!), leting Au on bronze.

For heating, they burned wood in the belly of horses, and on ashes remains scholars used the radiocarbon method.

The result was little strange: around 1500 A.D.! What they dated? Simple, the LAST gildation, not the bronze...

Anonymous said...

I was partially wrong: I've forgotten that "costant" concentration is due to tranformation of nitrogen (14N) in carbon (14C) in higher atmosphere.

But I'm happy: I remember the theory of beta decay after a plenty of years I got my degree in chemistry, and now my job is in industrial chemistry: great difference between theory and practical work.....