Featured post

Gender in the Merovingian World

Wednesday, 5 January 2011

The (Ab)Use of DNA in the Study of Early Medieval Cemeteries

[Much attention has recently been given, especially in popular news sources, to the use of DNA evidence in elucidating the 'Barbarian Migrations', whether in looking at the DNA of modern populations or in the use of ancient DNA from archaeological samples.  It is argued here that such approaches are potentially far more dangerous than helpful.  They risk returning us to ideas of ethnicity as a simple, inherited genetic entity, something akin to those on which the 19th-century idea of the Nation State was based.  The unpleasant uses to which such misleading ideas can be put do not require much imagination to envisage.  Ethnicity is multi-layered, flexible and about ideas.  The genetic approach ignores such things or even individuals' knowledge of their geographical origins.  A worked example or thought experiment argues that even with - from a migrationist perspective - a perfect data-set the explanatory possibilities of DNA are severely circumscribed.  Other questions need to be asked which this potentially valuable source of evidence is capable of contributing meaningfully to answering.]

When I was giving the paper posted below (as ‘Ethnicity and Early Medieval Cemeteries’: http://600transformer.blogspot.com/2010/12/ethnicity-and-early-medieval-cemeteries.html) in Vitoria, one of the questions I was asked concerned the use of DNA as a means of establishing the ethnicity of the deceased and its role in the community. The way the question was phrased was more interesting than usual, in that the argument was that if population elements from outside the area were identified in the skeletal remains, then that would be a means of explaining tension, stress and archaeologically-visible responses to them. As I said, this is more interesting than the usual approach, but I still had to respond fairly negatively, stressing that – when applied to the question of ethnic identity – the potential to distort and mislead hugely outweighed the possibilities offered to advance our knowledge and understanding.

As a convenient (if lazy) means of starting my discussion, here is the relevant section on the topic from my Barbarian Invasions and the Roman West (pp.451-2):

‘Most famous, however, has been the intrusion of the study of DNA into the investigation of the migrations, especially in England. It has, for example, been suggested that DNA can ‘prove’ that there was mass migration and dramatic population change in lowland Britain in the fifth and sixth centuries. The potential of DNA to show significant, precisely-datable, movement within the human population of Europe since the Neolithic can, however, be seriously questioned. Migration is a constant of European history; not something that occurred in specific, discrete episodes. More to the point, the samples used in these (and other physical anthropological analyses) have generally been selected in accordance with a particular view of history (specifically that of migration from northern Germany to Britain). Any similarities that emerge might thus be deemed simply to reaffirm a preconception. No controls are sought from such ‘unlikely’ areas as Italy, France or Spain, let alone Africa. That samples from the much more mixed modern populations of the cities of England and northern Germany show greater variation than those from the highlands of North Wales or Norway might also be unsurprising. Catherine Hills has said that, just as ‘historians hoped archaeologists would answer their questions, now archaeologists look to genetics … [as a] solution to all problems.’ In truth, the geneticists do not often seem to be answering archaeologists’ questions at all, but those of nineteenth- and early twentieth-century historians.

‘Most importantly, however, none of these analyses, even if possibly yielding information on geographical origins, tells us anything about what people thought they were, and ethnicity is a matter of belief. People who crossed the North Sea to Roman Britain in the third century and served in the Roman army adopted Roman culture and ethnicity. Any of their fifth-century relatives who did the same proclaimed their non-Roman identity, and may even have made different choices of which ethnicities based around their place of origin (Frisian, Angle, Saxon, Jute) they did stress. Yet their physical remains, DNA, stable isotopes in their tooth enamel, and so on, would all presumably be extremely similar. The fortuitous discovery of Stilicho’s body, conveniently wrapped and labelled, in a Ravennate grave, would tell us nothing of his own complex identity; what he thought he was, or what other people considered him to be – let alone the way in which his non-Roman antecedents were principally brought out once he had fallen from favour with the western court.

‘We know that people moved in the fifth and sixth centuries. The discovery of people of northern German origin in lowland Britain ought not to be a revelation to any sane scholar of this period. And, again, the absence of physical change does not imply a lack of population movement. What differentiated migration in the post-imperial world from that at other periods of history were the complex shifts in political and social identity of which it became one component, and which are studied in the next chapter. This cannot (and indeed can never) be examined through bones or teeth. The current vogue for forcing modern archaeological science to yield answers to old-fashioned and crudely-formulated historical questions runs the risk of returning us to a primordialist view of ethnicity and identity, such as was popular 150 years ago. These scientific techniques may yet yield interesting details about the personal histories of individuals. However, as far as the question of movement from barbaricum into the Roman Empire is concerned, without considerable refinement and greater sophistication, the study of the physical anthropology of the fifth- and sixth-century dead has no capacity to tell us anything that we did not already know, and every potential to set back the understanding of this period by a century or more.’

This – overall – would still represent my position. Here (with apologies for a degree of repetition) are some passages from a forthcoming (I hope) book I am writing at the moment, about post-imperial Britain (Jon Jarrett should note another mention of the King’s German Legion – information, by the way, which I owe to Alex Woolf…):

‘The employment of the DNA of modern populations to study the Anglo-Saxon migration is a deeply problematic and indeed – I would argue – dangerous line of argument. The problems lie in the methods used, not so much in the scientific analysis of the DNA itself but in the movement from such analysis to interpretative conclusions about the early Middle Ages, where the approach becomes decidedly unscientific. European DNA has been inextricably mixed since prehistoric times and the distribution of particular DNA traits reveal very broad areas with these similarities, so that the self-same maps have been argued to reveal the spread of Indo-European language and the movement of the Anglo-Saxons some millennia later. This in turn reveals a further problem: the date at which these patterns became similar cannot convincingly be dated. If, for example, a DNA pattern very similar to that modern inhabitants of Saxony was reported for the inhabitants of a southern English town, need that result from a fifth- or sixth-century migration, or from sexual encounters with a unit of the King’s German Legion (largely from Hanover) stationed there during the Napoleonic Wars? Or from sexual encounters between troops from that town and the women of northern Germany after 1945? A DNA similarity will not tell you the direction of population movement. A Saxon soldier could return home from the Roman army with his British wife and have a family there; Saxon raiders could take British slaves and sire children on them. And so on. We have seen that movement around the North Sea was an important feature of late Roman history.

‘Moving beyond the methodology used, yet further problems arise concerning the analyses’ assumptions. One is that migration is something that happened in discrete periods, so that, for example, the fifth and sixth centuries are often known as the period of the migrations (an appellation long criticised). Thus, in this view, demonstrable population-mixing can be dated to such specific blocks of time. Yet migration is a constant of human existence. We’ve already seen that people were moving from barbaricum east of the Rhine into the Roman Empire for centuries before 400. People moved, within the Empire, on a large scale too. And of course people have continued to move and to marry the inhabitants of other areas ever since. The similarities between the DNA of England and Germany might result from migration from Germany in the late antique period but, as intimated, it might stem from such movement at many other times and, indeed, from movement in the opposite direction. Once again, movement from the Empire to barbaricum, amply demonstrated in the archaeological record, is excluded from reasoning because it does not fit a model derived from problematic written sources. Thus this use (or misuse) of DNA is driven by a particular, and a particularly crude, reading of history and its results chosen to fit this story rather than to examine it.

‘Another, perhaps even more serious problem concerns the movement from DNA to conclusions about ethnic or political identity. As we have seen, ethnic identity is multi-layered. It is deployed (or not) in particular situations as the occasion demands, and it can be changed. A person’s DNA will not give you a sense of all of the layers of that person’s ethnicity, or of which s/he thought the most important or even if s/he generally used a completely different one, or of when and where such identities were stressed or concealed. Let me illustrate this. A male Saxon immigrant into the Empire in, say, the fourth century, would – one assumes – have DNA revealing the area where he grew up, but he would probably increasingly see himself, and act, as a Roman. His Saxon origins would have no part in his social, cultural or political life, and even less for his children, if he stayed in the Empire. If he returned home with all the cachet of his imperial service, even then it might have been his Roman identity that gave him local status. However, if a distant male relative of his moved into Britain a hundred and fifty years later, his DNA might be very similar but, in complete distinction, this man might make a very big deal of his Saxon origins for they would, or could, propel him to the upper echelons of society. DNA tells us nothing about any of this. What is pernicious about this use of genetic data is that it is essentialist. That is to say that it views a person’s identity as one-dimensional and unchanging, and it sees that dimension as entirely derived from that person’s biological and geographical origins. It is, in short, a view that reduces identity to something very similar to nineteenth-century nationalist ideas of race. Everyone sane knows that people moved from northern Germany to Britain in the fifth and sixth centuries. In that sense, these expensive analyses tell us nothing that we did not know already. In their implicit reduction of identity to a form of race, and by masking all the other contingent and interesting aspects of cultural interaction and identity-change, they not only risk setting back the understanding of this period by more than a century but provide pseudo-historical and pseudo-scientific ammunition for present-day nationalists, xenophobes and racists. Before leaving the topic of recent DNA analyses, the last thing to flag up is that the historical sources cited to show racial segregation are used very uncritically and rarely say what they are claimed to. ’

These extracts are, however, as stated, largely concerned with the attempts to draw conclusions about the extent of migration from the DNA of modern populations. When that section of Barbarian Migrations was written (around summer 2005), my understanding of the scientific situation at that time was that there was profound scepticism about the possibility of extracting usable DNA from ancient skeletal samples because of the dangers of contamination. As I understand things now, developments in science have reduced those dangers and there is apparently now the possibility even of using old skeletal data from excavations conducted years ago, to extract usable samples.

If so, these developments, to be sure, present us with much greater possibilities. But we must be sure that we are asking the right questions. The problem with the use of modern genetic data (quite apart from the fact that it seems to me to be principally aimed at getting the researcher in question into the headlines) is that it is (as I said in Barbarian Migrations, as quoted above) framed around questions aimed at a view of post-imperial history from over a century ago. The genetic evidence (assuming that the claims newly made for it are sound) must be used on its own terms, and historians of the period should be involved in framing sophisticated questions. Here is a potential area where I think that DNA analyses could be hugely valuable in the study of furnished inhumation (grave-goods) cemeteries:

Identifying family groups: Results could be compared with the detailed rites used in the burials of the deceased to see whether assumptions about preferred familial burial practices are well-founded, whether assumptions (made by me, for one) about similar rites possibly indicating kinship are soundly based. Where are the burials of genetically related burials found within a cemetery? I have suggested, for example, that the sixth-century phases of Frankish ‘row-grave’ cemeteries were laid out according to community rules, rather than organised by family. That said, I still originally interpreted clusters of similar graves (within the rows) as representing families. Subsequently I mooted the idea that clusters of similar burials were to be explained more plausibly according to their closeness in time rather than the nearness of kinship between the deceased. DNA would be the only way of deciding between these options – as also perhaps in investigating things like the double burial, ‘graves’ 6 and 8, at Ennery (Moselle); were they brothers? Are the seventh-century clusters of burials indeed, as I and others have always assumed, family plots. Etc.

So, what of ethnicity? What if, as my questioner asked, one could show that some people within a cemetery had significantly different DNA profiles from those of the rest of the population? Would that not help us to look at migration and its effects?

In the first instance, of course it would. Yet, there would be some fairly stringent methodological requirements, and some pretty strict limits to what it did tell us, and more importantly some very gave dangers. Let’s start with the methodological requirements.

One would have to know where these people came from, and that would have to mean the accumulation of a huge body of samples from all over Europe, North Africa and the Near East. The problem to date is that the samples have been chosen according to pre-conceived ideas and (given the huge degree of mixing in European genetics) often one geographical area of origin has been chosen from several equally plausible alternatives according to the (usually pre-determined) story that the researcher wants to tell. All the possibilities or variables (however implausible, however much they contradict what we think we know) have to be entered into the equation.

Second, we need to know just how different these genetic samples are and thus, related to that, how recently the ‘newcomers’ came into the community. Obviously, a cemetery founded in, say c.350 AD, with new rites adopted c.450 AD, cannot be explained according to the existence of a genetic ‘out-group’ that arrived in the region say 6 generations before the appearance of the new rite (i.e. in c.300). How fine can the chronologies given by DNA be (I ask because I don’t know, not as a rhetorical question)?

The strict limits on the conclusions would be that they were purely descriptive. Some people moved from A to B at some period between X and Y. As I have been at pains to argue, migration is a constant of human history, and we know perfectly well that significant numbers of people moved around in the fourth, fifth and sixth centuries, though whether more or fewer than in, say, the third or seventh is sadly a question rarely asked. We know that some people descended from people who lived around the Danube washed up in Spain in the late fifth and early sixth century. We know perfectly well that people moved from northern Germany to lowland Britain during the fifth and sixth centuries. DNA analyses that show that – after satisfying all the methodological requirements given above – are not telling as anything (at all) that we didn’t already know. Let’s (considering the cost of these analyses – is it $10,000 per sample?) be quite clear about that.

So the possibilities of DNA to tell us much about migration – at least until such time as we have so many (hundreds of thousands of) samples from across Europe and the Mediterranean basin for us to be able to make statistically significant statements about the scale of migration and the extent of intermarriage (this would be very important, but let’s remember the cost of the exercise and the likelihood that it might only confirm what sophisticated analyses of written and archaeological data already suggest…) – are meagre.

Migration happened. It is always going on. The real issue is why, during the fifth and sixth centuries, it became such an important feature of social change, and here DNA and other scientific techniques can tell us nothing. At all.

Let me illustrate with a hypothetical example. Let us suppose that we have a rural cemetery with a significant number (say 600: 200 per century of the site’s use, 50 per generation) of intact burials, founded in the Meseta (Spain) around 350 AD and lasting until 650. A new rite of burial with grave-goods, appears on this site c.450 AD and, of the people of that generation a clear genetic, DNA difference is found between the burials with grave-goods and those without, the latter showing the same profile as the burials in the four generations of the pre-grave-good period. Rigorous comparisons show that it is likely that the newcomers came from the Danube area about 3-5 generations earlier. The newcomers are thus buried in a different fashion from the ‘locals’. Over time, the DNA differences between the furnished (with grave-goods) and the unfurnished (without goods) graves decreases and, while the number of grave-goods burials increases, the range and number of objects decreases. In the last generation or so of the site’s use, the population is more or less genetically homogenous and grave-goods have been more or less abandoned.

This would be a case study that would be pretty much ideal (superficially at least) for the ‘migrationist’ perspective: we’d have a clear example of an out-group moving in from outside and their place of origin would seem to tie in unproblematically with their identification as Goths. They mark themselves out from the locals by the use of a new burial rite but over time they intermarry and eventually create a new community. This would tally with what we see as the historical narrative of the Goths in Spain. Hurrah! Case closed. Or is it? First of all, let’s be quite clear that not only would this be the ideal case for the migrationist view but – and this is especially important to stress for those not au fait with late antique and early medieval cemetery archaeology – the chances of finding such an ‘ideal type’ are slim indeed, for a whole range of reasons concerning each of the variables (the chances of such clear and neat distinctions, the length of use and the dates of foundation and abandonment of the cemetery; the chances of so many intact burials being found; the even spread of burials across the period of use; the ability to date burials with, let alone without, goods to within a generation, etc., etc.). Nevertheless, let us assume that this ‘dream data-set’ has indeed been located and scientifically excavated (and the DNA analysed, at the cost of $6,000,000 or so). What’s wrong with the migrationist interpretation?

First of all, any blurring at all of any of the clear lines I have proposed for this data-set weaken the strength of the hypothesis, even at its most basic descriptive level. Next, for us to explain the change according to the migration, we would need to be certain of two things: that this was the case generally in Spanish cemeteries with grave-goods, across the different regions of the Iberian Peninsula, and, secondly, that Spanish cemeteries without grave-goods, across the rest of Iberia, didn’t contain dead people with this ‘Gothic’ DNA profile (incrementally increasing the cost of the exercise…). Why do we need to know this?

First, because grave-goods burial is not a traditional Gothic rite. No amount of new, ‘ideal’ archaeology can change that fact. For the 80 or so years that the Goths were in Aquitaine, hardly anyone at all was buried in this fashion, during the 30 or so years of their migration from the Danube to the Garonne pretty much no one was, and before that it could hardly be described as a typical Gothic rite either. Thus we need to be sure that this was a new development associated exclusively with ‘being Gothic’.

But, even if the new rite was clearly Gothic, if there are regions of Spain (or other cemeteries in the same area) where, either, there are people with ‘Gothic DNA’ who didn’t get buried with grave-goods at the same time as our Goths on this ‘ideal site’ were being distinctively buried with goods, or, where people with ‘non-Gothic’ DNA were buried with grave-goods in the same generation as (or earlier than) the newcomers appeared and started interment with grave-goods in our ‘dream cemetery’, either alternative fatally weakens the migrationist hypothesis. It does so for the very simple reason that it means that, although we can see that (in this case) the new rite was associated with the intrusion of a new genetic group, it cannot be argued that the rite was in itself a straightforward sign of Gothic identity, because it can be shown that Goths themselves did not always use it and that non-Goths did use it. Therefore, we would have to conclude that, although in our community the Goths did adopt a rite which marked them out from their indigenous neighbours, the simple facts of their ‘Gothic-ness’ and of their migration into the area did not explain their choice and the change in the material cultural record. Immediately the explanation of the change must become much more subtle and complex. (All the more so, in fact, since in fact the new rite isn’t traditionally Gothic.) The grave-goods symbolise something else, or result from another causal factor, meaning that migration is largely if not entirely (in our specific case study) removed from the explanation.

Making a scientific theory stick does not mean just showing that one possibility fits the data; it also involves showing that all the others don’t fit it as well. Since human behaviour is messy, unpredictable and the rules that govern it are never better than fuzzy, the likelihood of being able to prove that all possibilities except one can (in the current state of knowledge) be ruled out range between slim and non-existent. Therefore explanations have to be more subtle and flexible than migrationist hypotheses invariably are. They must be able to accommodate a number of alternative possibilities.

But let us return to our hypothetical case-study and assume that – amazingly – it could be shown that in the earliest generations of grave-goods burial the rite was always and exclusively associated with ‘Gothic DNA’. Would this at least not prove the migrationist line?

Overall, of course, it would – to some extent. But, as mentioned above, the fact that the rite is, before the Goths’ arrival in Spain, not traditionally Gothic would immediately mean that some contingent variables (explaining the choice of artefacts and their symbolism) had to be incorporated. That would mean that elements to do with the contingencies of local and regional politics and social structure would need to be brought in to explain why the Goths did this. Migration alone would not explain the issue. After all, in other areas where the Goths had settled (such as Aquitaine, where they had lived for 70 years or more) this material cultural response to migration had not ensued.

What, however, if association between grave-goods and ‘genetic Goths’ could not be proven across Iberia (as is much more likely to be the case)? What about our (from a migrationist perspective) ideal, dream data-set? We’ve already had to conclude that ‘although in our community the Goths did adopt a rite which marked them out from their indigenous neighbours, the simple facts of their ‘Gothic-ness’ and of their migration into the area did not explain their choice and the change in the material cultural record.’ This is already a conclusion that would be too subtle for most migrationists to comprehend. The problem with migrationist hypotheses is that they narrow the range of interpretations and prejudge the explanation. If there is an observable material cultural difference between two groups with different genetic make-up, then that genetic difference must explain the material cultural variation. Think about that for a moment and how essentialist it is – as well as how essentially dubious its political implications are.

In our hypothetical case-study, the genetic difference would not tell us what this variation in the presence of grave-goods meant. Were the incomers staking a claim to power? Or was it, by contrast, a sign of the group-solidarity of a group with less social power (after all, generally speaking, groups with securely established authority do not tend to display that power to their local neighbours in grave-goods burial)? Did it proclaim Gothicness? Or (given the non-traditional nature of the rite) a claim to being, within the circumstances of the fifth century, ‘more Roman than the Romans’? Put another way, does such a display mark a ‘strategy of distinction’ (something which historiography has, in my view, been too preoccupied with) or a ‘strategy of identification’ (that is to say of identifying with the other people, already inhabiting the area)? What do I mean by the latter possibility? We can safely assume that the non-grave-goods-burying folk did not walk around naked, but their unhelpful burial practices remove the possibility of studying their costume. If the grave-goods-burying people interred their dead not in a different costume but in a form of ‘Sunday best’ that was actually normal in the region then their practice of depositing their dead in public ritual wearing costume that would be accepted for a person of that gender, age and status might in fact be a way of saying ‘we are just like you’. In other words, their origin as outsiders is being played down rather than played up. (On supposedly barbarian costume I can do no better than refer you to the works of Philipp von Rummel, notably his book, Habitus Barbarus).

If we could not make a link between the genetic out-group and Gothicness or – still more so – if all we had was genetic difference, we would not be able to say for sure whether the material cultural variation resulted from locals trying to look Gothic, incomers into a Gothic community trying to look Gothic, or locals or incomers trying to look Roman. Scientifically, these would all be possible, and the reduction of the range of alternatives in accordance with pre-conceived notions drawn from one old reading of the documentary sources would be neither scientific nor even historical.

Underlying all these points, furthermore, is the fact mentioned in the quotes from my books given above, that the DNA or other genetic information does not (and cannot) tell us what people thought they (or others) were, or what other people thought they were. People with ‘Gothic DNA’ (were we able to identify such a thing) need not have known they were Gothic. By the time they arrived in Spain, as many as six generations can have passed since the immigrant Goths had lived in Gothia. Can we be sure they still thought of themselves as Gothic? If they did, what did they think that that meant? Did any sense of a genetic origin mean anything, compared with claims to social or political status? Did the people in our cemetery even know that they were of such different biological origins? We can’t know. Answering that they must have done because of the material cultural difference yet again prejudges the issue and takes us close to a circular argument.

Further, again as mentioned above, the migrationist interpretation assumes that people only have one form or level of ethnic identity. There were different Roman and there were different Gothic identities, based on civitas, pagus (perhaps) or province for the former, based on different Gothic group, different village, different unit within the Gothic forces (I assume), possibly even differences the Goths brought with them only from Aquitaine (and thus quite Roman in their structure). Which, if any, level of such identities does the material culture represent? In other circumstances, genetic difference might mark difference between Goths of different origins.

And of course, the migrationist argument shares with other interpretations an assumption that these differences relate to identities in what I have called the ‘ethnic arc’ of the spectrum. They might also refer to socio-political function (soldier families for example – in the fifth century military status was associated with ethnic identity, to be sure, but this is a more complex issue than is often assumed, and in the late Roman world, we ought to remember that military service was likewise hereditary), or simple familial/kin-group differences, or social rank, or any number of other things.

Thus, even where we have what might seem to be the perfect data to fit a migrationist hypothesis, the explanatory value of DNA evidence is severely circumscribed. Every incidence or variable where the data fail to live up to the ideal type described above weakens its value yet further. Even in descriptive terms alone, for the foreseeable future it is highly unlikely that DNA evidence will do more than confirm what we already know. More seriously, though, the latching on to one form of (in its employment) pseudo-scientific evidence as a catch all explanation is dangerous in that in reducing ethnicity to a primordialist, monolithic genetically-based reality it runs the risk of taking us right back to nineteenth-century ideas of the nation-state and thus turning the clock back on our understanding of the early middle ages by a century and a half. The horrific political uses to which this sort of argument can lend itself do not require much imagination to envisage.