Towards an Absolute Chronology for the Aegean Iron Age

[Mark]

Grist to Patrick's mill.

http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0083117

[Patrick Waterson]

Grist to Patrick's mill.

Yes.

Exercises like this one, using a concatenation of ad hoc readings and stifling the outcomes with 'Bayesian analysis' alas have all the usefulness of playing 'pin the tail on the donkey' and then insisting the tail is in the right place and the donkey should be moved.

When radiocarbon dating was first introduced, the procedure was to take a single item, run it through the tests, and come up with a date for it.  Many of the dates did not coincide with conventional chronology: for example, a reed mat from Tutankhamun's tomb tested out at 9th-8th century BC, fitting Velikovsky's chronology in which Tutankhamun would have died c.823 BC but not the conventional one which would have him obit around 1323 BC.  Other dates did not seem to fit any chronology at all.

There are in fact two problems with radiocarbon dating and conventional chronology: the first is that conventional chronology is unfortunately incorrect; the second is that carbon-14 concentration, which radiocarbon dating measures and assumes to have been created/deposited at a constant rate, appears to vary with solar activity.  This would appear to be why attempting to reconcile radiocarbon dating with dendrochronology, which on the face of it should provide a clear base line for calibration, seems to create as many problems as it solves.

On top of this are added by way of further distortion attempts at chronology analysis by taking numerous objects from (supposedly) the same period, or a variety of periods, carbon-dating them and then suppressing the 'anomalous' readings by either statistical manipulation or claiming that they must have been 'contaminated' despite other samples subjected to identical precautions being deemed not to have been contaminated.  It is this inconsistency of and partiality in approach that convinces me these exercises are of very questionable value.

What is needed is a rational investigation of radiocarbon dating itself - including testing samples from the same source that have been deliberately contaminated to a known degree together with some for which all precautions have been taken and not writing off the results as an anomaly.

With the above lecture over, what can we actually tell about the dating of the "important transition between the SM and PG periods"?  (In case anyone wonders, 'SM' = Sub-Mycenaean and 'PG' = Proto-Geometric).  I shall attempt to answer this question in a forthcoming Slingshot article, but in brief Dionysius of Halicarnassus, Thucydides and Herodotus all supply important clues, and even Euripides contributes to the picture.  Very revealing is the presence of a chronological anomaly that was already present among Greek classical historians and which we have inherited.

[Justin Swanton]

Looking forward to the article!

[Jim Webster]

Yes, be interesting to see it ;-)

Jim

[Patrick Waterson]

It started as a post on the dating of the Trojan War and it just growed ... 

[Dave Beatty]

A very good discussion of all the problems associated with the various methods used for archaeological dating in the American southwest can be found at The Learning Center of the American Southwest, http://www.southwestlearning.org/topics/methods

To quote this site, "Chronometric dating techniques provide a range of dates that are relative, not absolute.  All chronometric techniques present statistically measurable uncertainty about the dates determined by the techniques.  This uncertainty, or error, is presented as either one-sigma (67% confidence that the date range within one standard deviation is correct) or two-sigma (95% confidence that the date range within two standard deviations is correct).  For example, if a chronometric technique returns the date of 600 B.P., then you have a 67% chance that the true date falls between 520 B.P. and 680 B.P., and a 95% chance that the true date falls between 440 B.P. and 760 B.P.  The two-sigma range is more likely to be correct, but provides a much broader date range than the one-sigma error term."

I have found that both radiocarbon dating and dendrochronology have significant problems, the former highlighted in other discussions above, and the latter due to the fact that many wooden articles that provide useful tree ring samples have been recycled (sometimes repeatedly) from older structures.  Thus, we have a good idea when the tree was cut, but not when the structure it was found in was actually built.

Some better dating techniques that are perhaps of more utility in the US than elsewhere are archaeomagnetic dating, obsidian hydration, thermoluminescence and (my favorite) packrat midden analysis.

[Jim Webster]

As someone who has spent his lifetime 'recycling' timber into other buildings I know exactly what you mean, good wood is too good to waste :-)

Jim

[Erpingham]

One of the worst things you can do with radiocarbon or other dating techniques is use them in isolation.  Not only do you have the statistical deviation to worry about but what you are dealing with could be reused from an earlier period, contaminated by later material in situ or contaminated in lifting, processing and storing.  Ideally, you want multiple dates from contemporary contexts.  You can identify the outliers which appear suspect (I know Patrick doesn't like this but it seems the only logical approach) and then, from your multiple samples, you can get a more statistically precise estimate of date.  One of the problems across the Eastern Med and Near East has been a dearth of scientific dating and what has been done seems to me mainly isolated samples.

We should also be aware of the value of sequential dating when pegged to scientific dating (otherwise it tends to free float).  In artefact-rich environments, some fairly good date ranges can be got from pottery sequences.  As Patrick will no doubt point out, however, these are often calibrated by "known" dates, rather than scientifically, so if there is a chance of systematic error.  However, because the sequences are fairly well understood, a more scientific matrix approach across sites with scientific dating should allow revision and wholesale redating.

[Patrick Waterson]

We should also be aware of the value of sequential dating when pegged to scientific dating (otherwise it tends to free float).  In artefact-rich environments, some fairly good date ranges can be got from pottery sequences.  As Patrick will no doubt point out, however, these are often calibrated by "known" dates, rather than scientifically, so if there is a chance of systematic error.  However, because the sequences are fairly well understood, a more scientific matrix approach across sites with scientific dating should allow revision and wholesale redating.

Very true, and this is the approach I would recommend.  Unfortunately where the stratigraphic sequencing conflicts with the conventionally understood chronology the reaction is not to revise the chronology but to find some bizarre rationalisation for it.

A case in point is the excavation of the Phrygian capital at Gordium, undertaken by Rodney Young in 1950-73.  He dug down to the original Phrygian layer, finding on top of it a 'Hittite' layer and then a Persian period layer.  Logic indicates that the Phrygian layer would date from the 9th century founding to perhaps the 7th century BC, when Phrygian power took serious knocks from the Assyrians, Cimmerians and Chaldeans in quick succession, then we would have a Chaldean layer from the 7th century BC to the mid-6th, then a Persian layer.  A sensibe sequential stratigraphic analysis would have confirmed that we need to redate the 'Hittite' empire to the 7th-6th centuries BC.

Not so for Young.  The obviously 'Hittite' nature of the Chaldean period layer had him conjecturing that the Persians had razed and removed the 'true' Chaldean layer and then overspread the site with unearthed debris imported from a 'Hittite' site elsewhere before settling in themselves. 

Another stratigraphic chronologically-influenced anomaly: Kanesh (Kultepe) in Asia Minor revealed a stratigraphic layer with plenty of Assyrian material from the mid-8th century BC dating mainly from the reign of Sargon - the writing was Assyrian mid-8th century cuneiform and featured Mannu-ki-Ashur, the commander of Sargon's bodyguard.  However it was found under a 'Hittite' layer, and so redated to c.2,000 BC and the time of Sargon of Akkad - and it was assumed that the style of writing was 'readopted' in the mid-8th century BC.  The 'Hittite' layer is of course Chaldean, c.700-550 BC, and the Assyrian colony dates from the mid-8th century BC and the reign of Sargon of Assyria, as the epigraphic evidence suggests (besides which, prior to the establishment of Trebizond/Trapesus in the mid-8th century BC it is hard to see any terminus for the trade route).

However the potential remains to use stratigraphy as the basis of chronology rather than vice-versa.

[Duncan Head]

I'm going to regret getting involved in this, but:

A case in point is the excavation of the Phrygian capital at Gordium, undertaken by Rodney Young in 1950-73.  He dug down to the original Phrygian layer, finding on top of it a 'Hittite' layer and then a Persian period layer.  Logic indicates that the Phrygian layer would date from the 9th century founding to perhaps the 7th century BC, when Phrygian power took serious knocks from the Assyrians, Cimmerians and Chaldeans in quick succession, then we would have a Chaldean layer from the 7th century BC to the mid-6th, then a Persian layer.

Not sharing the doctrine that the Hittites were really Chaldaeans, I wouldn't expect a Chaldaean/Neo-Babylonian layer at Gordion at all. There is evidence for Nabopolassar campaigning in Urartu and Neriglissar in Cilicia, but nothing that I know of suggesting a presence as far west as Phrygia. The recent excavators suggest that the Phrygian period was followed by Lydian rule - "At the time of its destruction, the fortress occupants were using primarily Lydian instead of Phrygian pottery, suggesting that the site was indeed a Lydian garrison".

A sensibe sequential stratigraphic analysis would have confirmed that we need to redate the 'Hittite' empire to the 7th-6th centuries BC.
Not so for Young.  The obviously 'Hittite' nature of the Chaldean period layer had him conjecturing that the Persians had razed and removed the 'true' Chaldean layer and then overspread the site with unearthed debris imported from a 'Hittite' site elsewhere before settling in themselves.

This appears to be based on remarks in Young's 1953 preliminary report. I'm not sure that this was anything more than a preliminary view: in the 1981 "Three Great Early Tumuli" (Final report III) he speaks merely of layers in which Hittite and Phrygian pottery overlap. The recent, and my impression is more thorough, excavators don't seem to acknowledge anything unusual about the stratigraphy (as far as I can tell from http://sites.museum.upenn.edu/gordion/ - I can't claim to have read every word).

So I think a sensible conclusion may just be that Young slightly cocked up a confused stratigraphy. In how wide an area did he claim to find Hittite sherds above Phrygian?

[Patrick Waterson]

So I think a sensible conclusion may just be that Young slightly cocked up a confused stratigraphy. In how wide an area did he claim to find Hittite sherds above Phrygian?

Quite an extensive one, judging by the following:

"The new city built over the clay layer dates from the second half of the sixth century. There is thus a lacuna of about a century and a half in the stratification and history of the site; the clay layer was not accumulated but dumped, apparently all at one time; the clay, brought from elsewhere, contains almost exclusively pottery of the Hittite period." - Young, The Campaign of 1955 at Gordion: Preliminary Report,  American Journal of Archaeology, Vol. 60 (1956), p.264