Dating the Dark Age, part II

The conventional absolute chronology of Early Iron Age Greece

In this second article in a series on the chronology of Early Iron Age Greece, Matthew looks at the different ways in which archaeologists and historians ascribe absolute or calendar dates to the relative chronology discussed in Part I.

Matthew Lloyd

In part one of this series, I looked at the fairly secure relative chronology of postpalatial Mycenaean and Early Iron Age Greece (ca. 1200-700 BCE), built on a typology of pottery decoration. The quantity of known, stratified material from this period means that there is unlikely to be a radical revision of the relative chronology as we now know it.

The absolute chronology of Early Iron Age Greece, however, is another story. Among other factors, the security of the relative pottery chronology for much of the Aegean – and particularly the parts people like to talk about, such as Athens and Lefkandi on Euboea – means that enthusiasm for absolute dating methods has been limited among archaeologists working on the period.

Nevertheless, three kinds of absolute dating have helped provide a framework into which the pottery chronology fits fairly comfortably. First there are historical chronologies, a broad category encompassing the historical traditions of the Greeks themselves and synchronisation with the historical records of contemporary Egypt and Southwest Asia (i.e. the Near East in Eurocentric terms).

The second and third are scientific dating methods: radiocarbon dating, used in a handful of cases before the excavation of Sindos in central Macedonia that will be the subject of part 3 of this series; and dendrochronology, or tree-ring dating, which relies on the annual growth rings of trees to show the date of a context. Both scientific methods will be explained in more detail below; however, as with historical chronologies, they are often from regions outside, but connected to, Greece.

With these kinds of historical synchronisms, we are typically dealing with what archaeologists call the terminus post quem, point-after-which or upper boundary, and the terminus ante quem, the point-before-which or lower boundary. This means that a certain feature of an archaeological context indicates that the context dates after a certain point – say, the date a coin was minted – or before a certain point – e.g. the destruction of a settlement. Typically, a context will only indicate one of these points.

Fixed points in time

The ancient Greeks recorded the passage of years in several different ways. In Athens, the year was named for the highest-ranking of the nine archons, the eponymous archon, and this convention was employed elsewhere for Athenian history. The Parian Chronicle, a stele from the third century BCE that gives the dates of many events historical and mythical, refers to the expulsion of the Peisistratid tyrants of Athens in the archonship of Harpaktides.

Similar historical records include lists such as those of the kings of Sparta, the victors of the Olympic Games, and the holders of religious offices, particularly the list of priestesses of the Argive Heraion by Hellanicus of Mytilene. The historicity of these lists is often dubious as the versions we know were compiled several centuries after the fact (Hall 2014, pp. 29-32), but that is not really the problem with them for our purposes. The bigger problem here is that they do not really tell us anything that we can cross-reference with the archaeological evidence of the Early Iron Age.

Left: fragment of the Parian Chronicle discovered in the seventeenth century CE, covering the period from 1582/81 to 355/54 BCE. In the Ashmolean Museum, Oxford. Right: fragment of the Parian Chronicle discovered in the nineteenth century CE, covering the period 336/35 to 299/98 BCE. The Archaeological Museum of Paros, Paroikia. Photos: Matthew Lloyd and Zde (via Wikimedia).

The Parian Chronicle records events as far back as what we would consider the sixteenth century BCE, beginning with Kekrops becoming king of Athens (1582/81 BCE) and including other mythical events such as Hades’ abduction of Kore/Persephone (1398/97 BCE) and Theseus’ rule of Athens (from 1259/58 BCE). As such, it’s a useful guide for how the ancient Greeks thought of their own past, but not so useful for our understanding of these earlier periods.

Oliver Dickinson points out that the sites for which we have king lists, such as Sparta, are those where the Early Iron Age archaeology is limited; on the other hand, the best preserved evidence for something resembling “kingship” comes from Lefkandi, for which we have no known historical record – we do not even know the site’s ancient name – and certainly no king list (Dickinson 2020, p. 41). Therefore, if you’re feeling generous, you can argue that a certain archaeological context corresponds to the reign of a Spartan king or Athenian archon, but the absolute date of the context must be derived independently of the historical chronography.

There are two significant exceptions from within the Greek tradition. The first is the Trojan War, considered by the Greeks – but not the present author – to have been an historical event (Thucydides 1.10.1-5; Herodotos 2.145.4). Herodotos suggests a date for this War eight-hundred years before his own time, around 1250 BCE; the Parian Chronicle gives us 1218/17-1207/08 BCE; the third-century chronographer Eratosthenes ca. 1187 BCE.

Enthusiastic archaeologists have connected this date to the destruction of the settlement at Hissarlik in northwest Turkey in the layer Troy VIIa, dated ca. 1200 BCE (Cline 2013, pp. 71-102). The argument here is largely circular – the identification of Troy VIIa as the “Trojan War” destruction is because its independently derived date correlates with the approximations in ancient sources. It could, however, be argued that this correlation offers something of a confirmation of the chronology.

Regardless of how much credence we give the story of the Trojan War, it doesn’t help us establish an internal chronology for the period between its occurrence and the point at which the epics about it were supposedly written. Herodotos dates Homer no more than four-hundred years before his time (2.53.2; ca. 850 BCE), i.e. some four-hundred years after his date for the war, suggesting that there was a period of about four centuries between the war and the Iliad; this date is more than a century higher than modern Homeric scholars date the epics, no earlier than 700 BCE.

The date of Homer, however, is not the second fixed point from the Greek tradition. Rather, the dates for the end of the Early Iron Age chronology are derived from Thucydides’ account of the foundation of Greek settlements in Sicily and southern Italy (6.1-5). These dates are considered particularly useful for archaeologists, as many of these settlements have been excavated and their earliest levels contain Corinthian pottery, supposedly providing a connection between the relative chronology of pottery and the absolute historical chronology.

Thucydides gives the foundation dates of these settlements in relation to one another, beginning with Naxos, the earliest, and continuing down to Acragas 154 years later. The most significant is the well-excavated site of Megara Hyblaea. Thucydides tells us that it was founded 245 years before its destruction by Gelon, tyrant of Syracuse, which we know to have happened in 483 BCE, thus giving us a date from which to calculate the foundation of the other settlements: 728 BCE.

Megara Hyblaea. Photograph by Davide Mauro; Creative Commons Attribution-Share Alike 4.0 International.

The excavations at Megara Hyblaea revealed, in their earliest layers, pottery of Corinthian Late Geometric and Early Proto-Corinthian. Thus, the foundation date of that settlement, 728 BCE, gives us a terminus ante quem for the beginning of Late Geometric pottery at Corinth, and thus an absolute date from which to hang the relative pottery chronology.

That being said, the course of true history never did run smooth. Historians question how accurately the memory of events supposed to have taken place in the eighth century would survive three-hundred years later (Hall 2014, pp. 108-10). The Greek penchant for keeping lists of rulers, victors, and priestesses seems not to date before the later fifth century, when the lists of archons, Heraian priestesses, and Olympic victors were first compiled. It seems that Greek historians may have been piecing together fragments of tradition and educated guesses, just as we are.

So it goes for the internal historical chronology of Greece. However, the Mediterranean world was an interconnected place and other regions have more secure chronologies than Greece. At Pithekoussai in the bay of Naples, one of the first settlements with an apparent Greek presence in Italy, the burial of an infant contained an Egyptian scarab with a cartouche of the Pharaoh Bokkhoris. On the assumption that this scarab was deposited within or not long after Bokkhoris’ reign, ca. 718-712 BCE, we have a terminus post quem for the Early Protocorinthian aryballoi found alongside it in the grave.

Greek pottery is also found overseas, where it can be tied into the historical dating of the Southwest Asian empires. A destruction horizon at the neo-Hittite city of Hama in North Syria provides a terminus ante quem for several Greek imports found at the site. The destruction has been attributed to the sack by neo-Assyrian king Sargon II at the beginning of his reign between 722-720 BCE; after this point, the city was abandoned for some time. In the area around the temple, five sherds of an Attic Middle Geometric II krater (wine-mixing vessel) were discovered; in the cemetery and several unstratified burn deposits, Sub-Protogeometric pendant concentric semi-circle skyphoi (two-handled cup or bowl) were discovered.

Nicholas Coldstream argued that these vessels indicated that Sub-Protogeometric and Middle Geometric must extend late into the eighth century (Coldstream 2008, pp. 310-313). As the vessels were not found on the settlement floor, which would suggest that they were in use during the destruction, he did not push the date too late in the century, but nevertheless he ended up with a Middle Geometric period in Attica that was nearly a century in length (ca. 850-760 BCE).

Connections between the Aegean and Egypt also give us dates for the Late Bronze Age pottery chronology. The presence of vast quantities of Late Helladic IIIA2 pottery in the short-lived capital of the Pharaoh Akhenaten (ca. 1353-1337 BCE) at El Amarna, occupied for a maximum of two decades, provides a strong terminus ante quem for the development of that style (Dickinson 2020, pp. 47-48).

Earlier still, Middle and Late Minoan pottery has been found in the Egyptian Late Middle Kingdom sites of Lisht, Kahun, and Harageh in Egypt; dates for Late Minoan IIIA and IIIB suggest that one style is developed into the other between 1375 and 1350 BCE (Kemp and Merrillees 1980).

A few less-substantial links connect Aegean pottery to Egypt in the following centuries. Late Helladic IIIC Early has been synchronized with the reign of Ramesses III (ca. 1184-1153 BCE). This claim, however, is dependent on the association of the Philistines in the southern Levant with one of the groups of “Sea People” Ramesses claimed to have repelled from Egypt, whose early pottery seems influenced by the first phase of Late Helladic IIIC. After this, the possible links between Greek archaeology and Southwest Asian history almost entirely disappear until the seventh century BCE.

Wary readers of Ancient World Magazine may feel a tickling in their brain that there’s something off about the association of archaeological destruction levels with historical events. Josho and I have both referenced what Anthony Snodgrass calls “the positivist fallacy” – the association of what is archaeologically visible with what is historically significant – in particular in the relationship of destruction levels to sacks discussed in historical sources.

I do not believe that this article is the place to go into a deep discussion about the reliability of these absolute dates; only to note that concerns have been raised. I will discuss the destruction event at Gordion in Phrygia below to offer some insight into how these dates can be challenged. I will turn now to a dating method that gives arguably more precise dates, but which is not without its own problems.

Radiocarbon before Sindos

There have been some attempts to provide more secure dates for the EIA on the basis of radiocarbon dating. Radiocarbon dating measures the decay of carbon-14 or 14C, an unstable radioactive isotope present in the Earth’s atmosphere which is absorbed by plants as carbon dioxide through photosynthesis. These plants are then eaten by animals, which are in turn eaten by other animals. Through this process, carbon-14 is present in all living things on the planet – plants, animals, fungi, etcetera.

Carbon-14 is accumulated until an animal stops eating or a plant stops photosynthesising – i.e. when it dies. At this point the amount of carbon-14 in the plant or animal is the same as the level of carbon-14 in the atmosphere, but because carbon-14 is an unstable isotope it begins to decay. The decay rate or half-life of carbon-14 is 5,730 years – it takes 5,730 years for half of the atoms to decay. Thus, if the level of atmospheric carbon-14 when the creature or plant died is known, the level of decay can be calculated and from this the date at which it died.

When radiocarbon dates were first calculated in the 1940s CE, it was assumed that the level of atmospheric carbon-14 was stable. This proved not to be the case, with radiocarbon samples of tree rings older than 1000 BCE indicating that the level of carbon-14 in the atmosphere was much greater. Fortunately, tree-rings provide an independent dating method, discussed below, which has allowed the creation of a calibration curve to provide more precision to radiocarbon dates.

Radiocarbon dating gives a length of time between the death of the organism and the present, presented in years Before Present (BP). Conventionally, the radiocarbon “present” is 1950 CE, just after the first publication of radiocarbon dates in 1949 CE. Radiocarbon dates are also presented with an associated measurement error, ± so many years, indicating that the likely date extends so many years before and after the radiocarbon age.

Radiocarbon dates can theoretically be taken from anything that was once alive, although short-use objects usually provide more accurate dates for their contexts. A radiocarbon date from wood used in construction of a building will only give the date of construction – if that, see below – not the span of the building’s use.

Other limitations include the amount of carbon required, but new methods have reduced this from 5g of pure carbon when the process was developed in the 1940s to 5-10mg if the accelerated mass spectrometry method is used. This necessity has created particular problems in regions such as central Greece where soil conditions are not good for the preservation of organic matter, making it difficult to collect samples of the necessary size.

Contamination is also a problem. Radiocarbon dates are only useful if they come from secure deposits that can show that the material tested has not migrated from another stratum. At tell sites, such as Sindos and Lefkandi, the constant rebuilding can secure earlier contexts, but human pit-digging and construction – among other processes – often rework younger material into older contexts, and vice versa.

A calibration problem toward the end of the Greek Early Iron Age is the phenomenon known as the “Hallstatt Plateau”. This term refers to the fact that radiocarbon dates around 2450 BP (ca. 500 BC) always calibrate with a wide margin of error, resulting in a date of 800-400 BCE, the time of the Hallstatt culture in central and western Europe. This results in a lack of specificity in the radiocarbon dates of certain strata compared to the apparently more precise dates of pottery styles. This problem can be worked around in some cases where a more secure date is known through some other means – a radiocarbon date from an adjacent stratum that does not fall on the plateau, or dendrochronological dates – as we shall see in the case of Sindos.

Also significant is a lack of enthusiasm for the method among more traditional archaeologists of early Greece, who do not see a problem with the chronological framework as it stands. As the Hallstatt Plateau covers much of early Greek history, it has discouraged archaeologists of this period from using radiocarbon dates to scientifically validate their assumptions about the chronology.

Therefore, radiocarbon dating was not used to create the chronology of Early Iron Age Greece in the same way as historical records and the relative dating methods discussed in part 1 of this series. Rather, where it has been applied, the questions asked of radiocarbon dating have been to confirm or to challenge the accepted structure.

Some radiocarbon dates indicate that the dates assigned to the relative phases of the pottery chronology are near enough accurate. Dates from Lefkandi, Kalapodi, and Corinth seem to indicate that the transition from Sub-Mycenaean to Early Protogeometric should be dated in the late eleventh century BCE, ca. 1020 BCE (Toffolo et al. 2013).

Four tombs at in the Early Iron Age cemetery of Torone, containing pottery in the Sub-Mycenaean and Protogeometric styles, seem to confirm the standard ranges for these periods, but all have margins of error in the region of 80-90 years (Papadopoulos et al 2011, pp. 198-199).

Other radiocarbon dates suggest radically different dates for Greek chronology. Most notable, prior to the radiocarbon dates from Sindos that will be discussed in the next part of this series, are the dates from Assiros, also in Macedonia. These dates suggest that Attic Protogeometric should begin no later than 1080 BCE, if not several decades earlier, while the beginning of Late Helladic IIIC should date to the mid-thirteenth century BCE.

The dates from Assiros have caused some consternation among Aegean archaeologists. Dickinson comments that the publication of the Assiros dates “offers no suggestion for coping with the knock-on effect of raising the beginning of PG on later phases” (Dickinson 2020, p. 48). That several of these dates are determined from timber samples has also led to the suggestion that “old wood” was used in construction (Papadopoulos et al 2011, p. 195-6), although the more recent work uses more short-lived samples such as grains.

The wood from the trees

I learned as a child that you could tell how old a tree was by counting the number of rings in its trunk. Each year, many species of trees grow a new ring, the thickness of which varies based on fluctuations in the climate and the tree’s age. Depending on the local climate, variations in annual rainfall, sunlight, and temperature will affect the thickness of the tree ring, although they typically getting narrower as the tree ages.

The study of tree-rings is known as dendrochronology. Starting from trees in the present and then incorporating older timber, dendrochronologists can cross-check the ring growth of trees of the same species, but different ages, and so start to build a continuous sequence running backward from the present for thousands of years. In some cases, sequences “float” disconnected from the main sequence, but with other dating methods they can be used to give accurate dates for their contexts.

The long master sequences of tree rings are also used to calibrate radiocarbon dates. The long-lived tree species California bristle-cone pine allowed the creation of a sequence nearly 9,000 years back for southwestern North America, while oak sequences in Germany stretch back more than 10,000 years.

The growth rings of an unknown tree species, at Bristol Zoo, Bristol, England. This trunk is three feet (1 metre) across. Photographed by Adrian Pingstone in September 2005 and released to the public domain.

With these sequences in hand, archaeologists fortunate to discover wood from one of these tree species on their sites can usually get an accurate felling date for the timber in question. It is important to note that the date derived from dendrochronology is just that: the date at which the tree was felled. Wood needs to dry out before it can be used in construction, meaning that it was often stored before it was used. It is also possible that old timber was re-used or that younger timber was added to a structure for repairs, again throwing off the dates. As dendrochronological dates, like carbon dates, tend to suggest that a context is older than archaeologists think, this phenomenon is sometimes called the “old wood” problem.

It must also be noted that the felling date can only be obtained from the outermost ring of the tree, beneath the bark. This is the latest growth and thus gives the date the tree was killed. Where the wood has been shaped in such a way that the outermost growth is not visible, dendrochronological dates may be inaccurate. In more recent logging activity, tree trunks are squared off to make them easier to transport, removing the outermost layer.

The impact of dendrochronology on the Early Iron Age Greek chronology is, like the historical chronologies, more significant for its links to surrounding areas than the evidence from what is currently Greece. However, significant dendrochronological dates in conjunction with radiocarbon dates come from the Phrygian site of Gordion in Anatolia.

A destruction level at Gordion, on original excavation, was identified as the result of the Kimmerian invasion at the end of the reign of Midas, ca. 709 BCE (Strabo 1.3.21). Radiocarbon results in the 1960s suggested that the date of the destruction should be higher; dendrochronological evidence seemed to confirm this, pushing the destruction level back a century, into the ninth century. These dates were explained away as re-use of old wood.

Further radiocarbon dates from the destruction level, this time of short-lived seeds, confirmed that the destruction had been dated too early. It is now accepted that the destruction level dates to the last quarter of the ninth century.

Another dendrochronological date from Gordion challenged the interpretation of Tumulus MM as the tomb of Midas. Timbers used in the construction of the burial chamber were dated to 740 BCE, more than thirty years before Midas could possibly have died.

These raised dates seem to confirm certain aspects of the Greek chronology, as the re-built citadel contained Corinthian Late Geometric and Protocorinthian pottery, dated to the mid-to-late eighth century BCE.

End of Part II

The phenomenon of “old wood” is one of several convenient get-out-of-chronological-revision-free cards used for many of the challenges that have been presented to the chronology of Early Iron Age Greece. But there are certainly difficulties with establishing an absolute chronology that should not be overlooked.

One such problem is the question of synchronisation. Radiocarbon dating and dendrochronological dates have both developed in such a way as to reinforce one another, each apparently making the chronology provided by the other more secure. But cross-referencing these dates with those provided by historical sources has proved difficult, and must result in questioning whether it is the archaeological identification of the historical event that is wrong, the measure of dates provided by the ancient sources, or the methodology behind the scientific dating methods.

In the next part of this series, I will focus very specifically on the excavations at Sindos and the recently published radiocarbon dates from that settlement. These dates seem to be much more reliable than those discovered in the past as they are from well-stratified contexts, short-lived seed samples, and were discovered alongside significant evidence of the accuracy of the relative dating system of Early Iron Age Greece.