Isotopes and impact: a cautionary tale.

Pollard, A.M.

Providers and consumers of science

There can be no doubt that isotopic studies have made a huge contribution to archaeology in recent years, so much so that isotope archaeology is now seen as an essential subdiscipline of archaeology in much the same way as isotope geochemistry is a key subdiscipline of geochemistry. Ignoring for current purposes the contribution made by the measurement of a particular radioactive isotope of carbon (14C) since 1950, we can date the beginnings of isotope archaeology to the mid 1960s with the first measurements of lead isotopes in archaeological metals and slags by Brill and Wampler (1965, 1967). This was followed by carbon stable isotopes in human bone collagen in the late 1970s, building on previous work measuring 813C in archaeological bone for radiocarbon determinations (Vogel & Van der Merwe 1977; Van der Merwe & Vogel 1978). Other isotopes followed rapidly, such as nitrogen, oxygen, sulphur and hydrogen for archaeological, palaeoecological or palaeoclimatological purposes and, more recently, the heavier radiogenic isotopes of strontium and neodymium for determining the provenance of organic and inorganic materials (Pollard & Heron 2008).

There is, therefore, nearly 50 years of accumulated experience available on how to use isotopes in archaeology but also, unfortunately, some evidence of how not to use them. The use of lead isotopes in metals as a provenance tool went through a controversial phase in the 1980s and 1990s (Pollard 2009), resulting in nearly a generation during which lead isotopes were almost totally neglected in archaeology: fortunately, this is now changing. There is a danger that the use of the latest fashionable isotope system (strontium in dental enamel) is heading in the same direction, and this paper is written in the hope of preventing this very promising technique suffering the same fate. These dangers are, if anything, more acute now than they were in the 1980s because of the need to demonstrate impact, the measurable effect of a piece of research outside the discipline. Because of the intense public interest in archaeology, it is inevitable that one manifestation of impact is the reporting of research in the media, with the associated hardening of the story whereby the nuances of interpretation are lost in the need to create good stories. The intention of this essay is not to criticise directly any particular piece of work, which would run the danger of taking the debate into the personal arena. It is primarily intended to launch a dialogue between laboratory and field archaeologists about the use of strontium isotopes in archaeology, with the aim of alerting the consumers of such work to what should realistically be expected, and also to remind the producers of the data about the complexity of archaeological interpretation. By suggesting some simple protocols, the aim is to help make sure that such a promising technique does not get lost in hyperbole and subsequent anti-climax.

Having drawn the parallel between isotope archaeology and isotope geochemistry, it is important to appreciate that the use of isotopes in archaeology is different from that in geochemistry. The difference is us: people. People do strange things, but of course it is the study of what people do that makes archaeology so intellectually interesting and at the same time scientifically challenging. This means that the controls on the isotopes we measure in archaeological material are not simply those of biogeochemistry, but are influenced to some degree by human behaviour. As a minimum, this requires that isotope archaeology needs what might be termed an extra interpretative step to account for the contributions of human agency. In fact, because it is in reality much more complicated than that, it might be argued that it demands a completely different approach. The human factors have to be built in to the research design from the beginning.

Despite recent claims by the manufacturers that the new generation of portable analytical instruments amounts to the invention of the all-purpose tricorder of Star Trek fame, real analytical instruments do not directly provide answers in any branch of science. They produce data, which have to be interpreted before an answer--more strictly, an inference--can be reached. This is why scientific research is an iterative process. We make a set of observations, create a model and produce data which either supports that model, or invalidates it. In either case, we may subsequently produce a new model as a result of new observations, or by redefining the original question more carefully, which causes us to re-interpret the data, or collect some better data, in order to test this new model. This is true in geochemistry, evolutionary biology or classical mechanics. It is perfectly normal and is nothing to be ashamed of. It does not, as is sometimes claimed in the popular media and particularly in highly polarised contexts such as climate change, completely discredit the process of science as in 'how can we believe scientists when they keep changing their minds?'

Science is essentially a Bayesian process which tends towards a more realistic view of the world. Perhaps, however, because of the supposed cultural divide identified in C.P. Snow's famous but subsequently disputed essay on the two cultures (Snow 1959), there is sometimes a lack of communication in archaeology between laboratory scientists on one hand and, on the other, essentially humanistic scholars or professional field archaeologists who have occasionally been persuaded to believe that science provides definitive answers. This, unfortunately, is a widely held misconception, partly but unwisely promoted by scientists themselves as a response to a political framework which demands clear-cut answers--now required in order to produce impact--in return for funding. It is all too clear that the current political debate on climate change is desperately struggling to reconcile the honest answers produced by science, which are often expressed probabilistically, with the enormous costs of doing something; or of doing nothing.

Where lead led

Possibly, it is entirely reasonable for archaeologists to expect, indeed to demand, that if money is spent on scientific analysis then it should produce something definitive.

But how realistic is this? Before considering the most recent case of the use of strontium isotopes, it is instructive to consider what happened with the application of lead isotopes in archaeology. There is nothing wrong, and never was, with the lead isotope technique. The debate was never anything to do with the precision of the measurements or the importance of the data; it was entirely about how those data were interpreted to provide an answer with some archaeological relevance. There were obvious limitations in sampling, both of archaeological material and of geological comparators, but this is inherent in any form of archaeology and dealing with imperfect data is part of the job.

The problem was essentially related to the nature of the questions being asked, as well as a consequence of the expense and technical difficulty associated with making such measurements (Pollard 2009).

The problem with the questions being asked is that they were, although very well-intentioned, ultimately somewhat naive. For the most part the methodology was very simple. By comparing lead isotope signatures in bronze objects to isotope signatures in geological copper ores, we can provenance a metal object to a particular ore source and this will help us to understand an aspect of human behaviour in the past such as trade and exchange. The limitations of such a simple geological determinism are clear to anybody who studies the social history of technology. To pick the obvious issues, what about the mixing of ores, the recycling of metal or the selection of particular materials for particular roles? Even if, as may well be possible, some of these old issues can now be resolved, we are still left with a fundamental question: how useful is it to know that a particular metal came from a particular mine? One can certainly conceive of some circumstances where it might be very useful, such as providing direct evidence for contact between two localities as originally envisaged or suggesting that metals from a particular mine were imbued with particular ritualistic as well as technological or symbolic properties, but it is equally possible that the ancient metalsmith neither knew nor cared where the metal came from. We might counter-argue this by pointing out that facts are facts and that, even if the smith did not care, it is valuable for us to know or simply demonstrate that the smith neither knew nor cared. Whether the information gained is worth the time, money and samples consumed depends on our ability to contextualise this information and, ultimately, on the explicit aims of the research. The question has to be asked: is there a better way to interpret the data?

Archaeology is essentially predicated on the observation of change and difference, either through a stratigraphic sequence (with time) or from site to site (with space). This might be a change in patterns of pottery form or design, a change in costume styles as evidenced by decorative items such as brooches, differences in the spatial patterning and structure of houses or changes in burial practices. When several such changes coincide, then it is reasonable to assume that something significant has happened in the history of that particular locality. The exact relationship of these changes to a change in culture is highly problematic, but making that link is essentially what archaeology is about. So, if the lead isotopes in copper alloy objects change from Layer II to Layer III at a particular site, then that is another significant change to be added to the list. It may indeed be--as the 1980s model would suppose--that this reflected a change in the source of metal supply, but it could equally be a change in the pattern of recycling and deposition, some other cultural phenomenon such as a change in the relationship between form, function and metal source or, perhaps most likely, a combination of several factors. And this is the key: a change in lead isotope values can only be considered in the context of all the other material evidence.

The modern sociology of advanced analytical science in archaeology is also important. In the lead isotope case, there were effectively only three labs in the world involved in making a significant number of measurements on archaeological materials. Each had a different interpretative philosophy, and some were less good than others at publishing their raw data. In certain cases, helpful and constructive critiques of the interpretations from outside the fraternity were not always welcomed. This is not, and was not, healthy and does not make for good science. As a basic minimum, all the raw data must be published (not just the interpretations), other labs should be positively encouraged to get involved in the measurements and comments should be welcomed (or at least tolerated) from outside the immediate research area. Science at its best is collaborative rather than competitive despite the counter-pressures from limited funding. Healthy science requires debate created by a critical mass of active and informed participants, not a clique of adepts who retain sole rights to interpretation.

The strontium saga

Archaeological interest in strontium was first aroused by the observation that strontium is discriminated against as it passes up the food chain. Thus, the ratio of strontium to calcium in bone decreases from herbivore to carnivore (as reviewed by Sillen & Kavanagh 1982). It was quickly realised, however, that in an archaeological context the strontium content of bone is strongly affected by diagenetic processes in the burial environment. Ericson (1985) was the first to suggest that the isotopic ratio of strontium (specifically [sup.87]Sr/[sup.86]Sr) is a powerful indicator of geological source when measured in human dental enamel which is preferred to bone because of the higher density and therefore greater resistance to postmortem contamination. Since then, strontium isotope studies have been widely applied in archaeology, and have been reviewed by, for example, Bentley (2006).

Most commonly, strontium isotopes in dental enamel are measured in conjunction with oxygen isotopes ([sup.18]O/[sup.16]O) in the carbonate or phosphate of the enamel. This signal is believed to be largely controlled by the isotopic ratio of oxygen in drinking water and is, therefore, a powerful geographical indicator (Huertas et al. 1995). Similar comments to those made here about strontium isotopes could and should be made about oxygen, but because the complicating factors are somewhat different, that is being considered elsewhere (Pellegrini pers. comm.).

So what lessons does the sorry history of lead isotopes have for the current application of strontium isotopes to archaeology? There is a danger that the use of strontium isotopes in dental enamel as a tool to determine the provenance of humans could go down the same rocky road. There is a popular view, promulgated strongly by the media, that the measurement of strontium isotopes (usually combined with oxygen isotopes) can provide a simple archaeological service: send in a tooth and get an unequivocal postcode for the domicile of that individual as a child. This simplified perception is almost certainly not being actively promoted by anybody from within the isotope community although it is possible that occasionally it is not being discouraged as strongly as is necessary. It is a perception which could be fostered by a partial and uncritical reading of the literature, but also, and most worryingly, by the tendency of popular television shows in particular to harden up scientific evidence. All academics are now under pressure from their institutions and funding bodies to demonstrate impact, and how better than to be interviewed on prime time television and radio? The price, however, is that the 'could have come from' statement becomes 'did come from' and uncertainties, ambiguities and contradictory evidence are forgotten. As a consequence, however, increasing pressure is being put on funding bodies and the managers of post-excavation research in both commercial and academic contexts to get the strontium isotopes done.

Is this approach a bad thing? In principle it isn't, and often represents the best approach to a serious and well-formulated archaeological question. There are many case studies in which the strontium isotope evidence has been critical in revealing new and hitherto unobtainable insights about human mobility in the past. Inevitably, however, in other cases it appears to be simply following a fashion, to see if anything interesting emerges and without a pre-formulated research strategy. This, although not itself entirely wasteful, is perhaps a poor use of expensive resources. It is almost as if the attraction of the apparently instant answers provided by science is taking the place of real archaeological analysis and interpretation. More critically, however, there appears to be emerging a simple model which sees determining origin as a valid end in itself and it is this aspect which strikes a parallel with the limited objectives used in the early lead isotope studies.

In both cases we can ask the following question: how valuable is it to know where something/someone comes from? In the case of humans, it is perhaps trivial to point out that identity is a social construct. Certainly birthplace may be significant, but it isn't the only consideration. So how useful is it to know where somebody lived as a child? We might feel that it is, but did it matter to the individual or the society in which she or he lived? As with lead isotopes, we might feel that facts are facts and it is important for us to know, but it again depends very much on the aims of the research. Much more significantly, however, the tendency towards simple geological determinism in the identification of origin should be resisted.

As a first guess of what the strontium isotopic signal should be in the teeth of individuals growing up in a particular area, it is entirely reasonable to look at the isotopic values in the rocks beneath the landscape. The underlying solid rock geology is obviously a prime determinant of the strontium isotope ratio in the local biosphere, but it is only one of the factors which might contribute to the signal of humans living in a particular place. Other factors might include drift geology, water supply sources, farming and culinary practices, trade in foodstuffs and dietary taboos (Montgomery 2010). In some areas, for example, the strontium isotope ratios of the underlying geology can vary significantly over short distances, and thus the detail of food procurement strategies might affect the human values quite markedly. We are neglecting for the purposes of this discussion the obvious and potentially very serious problems which might be caused by the recrystallisation of dental enamel, or postmortem chemical alteration, which have to be considered in any study of biological tissues in archaeology.

As well as considering geological uncertainty, we need to understand more about the biology of strontium uptake in dental enamel. Until weaning, it seems likely that most of the strontium will enter the child through the mother's or wet-nurse's milk and therefore the signal is related indirectly to the local biosphere. It seems possible that a child might acquire a non-local strontium signal through being nursed by an immigrant mother, even if that child never moved from the locality. Early-formed enamel might therefore differ from later-formed enamel. That of itself would be archaeologically interesting, but might not be correctly interpreted or possibly even interpretable.

If the isotopic values in teeth cannot necessarily be directly related to the values in the rocks beneath our feet, then how else might we approach the important archaeological discrimination between locals and incomers? Some studies have addressed this question by, at least in the first instance, ignoring the solid geology and focusing on the evidence within the isotope data themselves. Is there a discernable dominant value within the teeth of a large number of individuals which might plausibly be seen as local? One can then ask whether this matches other evidence for a local signal, such as that from local faunal data (modern and/or archaeological), and only after that if it matches the local floral and geological signal. In the ideal instance, of course, the geological data will support all the other evidence and a confident answer can be given about who is local and who is not. It seems preferable, however, to approach the problem from the isotopic data to the geology rather than the other way round in order to allow a natural isotopic boundary for the local signal to emerge, which may take into account some of the confounding issues described above, rather than having an external boundary imposed from the solid rock geology alone. This poses a numerical challenge in deciding where such a natural boundary might fall, but where enough data have been accumulated and the geological differences are large then this is often so obvious that it does not require convoluted mathematics. In general, it seems likely that this natural boundary will encompass a larger region on an isotope diagram (i.e. a plot of strontium vs oxygen isotope ratio) than that defined purely by the underlying local geology and rainfall values alone. Multiple factors might be responsible for this broadening, related to the complications of regional geology and human behaviour listed above. For example, it looks as if some of the strontium values in Britain lie on a mixing line between that of the local solid geology and the local rainwater values (i.e. the strontium isotope ratio of modern seawater) but there is no reason to expect this particular relationship to be a general phenomenon (Montgomery 2010). Other mixing lines might be important, such as that between lowland and upland areas if these differ markedly in geology. We might also anticipate differences between social or gender groups: elites having access to food resources from a wider region than other social groups.

There are, therefore, many reasons why the strontium isotope values of a range of individuals who lived and died in the same locality may differ significantly. Only if the local isotope field is robustly and comprehensively defined can foreigners be identified, and speculation begin, on their origins. From a simplistic point of view, one would expect that if the local isotope field is too tightly defined, then an unknown number of locals are being misclassified as foreigners.

Assuming that incomers can be securely identified in the data, the obvious next question is to ask where they come from, but this is full of potential pitfalls, circular arguments, wishful thinking and, in general, too few samples. It is important to distinguish clearly between a hypothesis-driven approach and one which simply questions where all the data fit. The former is a much more rigorous process. If some other archaeological (or literary, linguistic or genetic) evidence suggests an origin in a particular location, then it is perfectly legitimate to test the isotope data against what one would expect from that location whilst bearing in mind all the reasons listed above why this might not have a simple unequivocal answer. The key point, and the lesson to be learned from lead isotopes, is that all evidence must a priori be given equal weight. Scientific evidence, simply because it is quantifiable, does not automatically trump other forms of evidence.

This is quite a different approach from simply using a geological/rainfall map and saying 'I think they came from here'. We must also remember clearly one of the fundamental constraints on provenance studies of any description. It is only ever possible to disprove a source hypothesis, never to prove one. all that can be said with certainty is that this did not come from here. If a match exists, then it is possible to say that the evidence is consistent with a source at place X, but we can never discount the possibility that somewhere else would look just like it.

Making the links

So what is the overall lesson to be learnt from this brief study of the history of isotopes in archaeology? Strontium isotopes in dental enamel--usually combined with oxygen isotopes, and even sometimes with lead--offer a very powerful tool for looking at the mobility of humans, and other animals. We must therefore press forward enthusiastically with research into strontium and these other isotopes, but we must also use the techniques wisely and cautiously. The first key to good research in archaeological science is the quality of the archaeological question and the consequent structure of the proposed research. It is important to consider first and foremost the isotope data in a wider archaeological context, and be prepared to look for something which has changed or perhaps, in this case, something which is different rather than looking ab initio for a specific postcode of origin. A key test is to ask whether the rest of the archaeological evidence supports or refutes such observations. A more general requirement is to make sure that there is a multiplicity of voices to be heard in terms of posing the questions and interpreting the data. Several labs should be involved in producing such data, using common standards and publishing all their data. This is not to advocate that all measurements should be duplicated in more than one lab, although that might be regarded as good practice for a limited number of cases, simply that several labs should be involved in such research. If these simple lessons from the past are applied, then we might avoid the opprobrium that has fallen on anybody who mentioned lead isotopes in polite circles during the last 20 years. And we might also learn something interesting and new about the history of our species, rather than simply generating ill-supported speculations, which will ultimately bring our science into disrepute.

References

BENTLEY, R.A. 2006. Strontium isotopes flora the earth to the archaeological skeleton: a review. Journal of Archaeological Method and Theory 13: 135-87.

BRILL, R.H. & J.M. WAMPLER. 1965. Isotope studies of ancient lead. American Journal of Archaeology 69: 165-66.

--1967. Isotope studies of ancient lead. American Journal of Archaeology 71: 63-77.

ERICSON, J.E. 1985. Strontium isotope characterization in the study of human prehistoric ecology. Journal of Human Evolution 14: 503-514.

HUERTAS, A.D., P. IACUMIN, B. STENNI, B.S. CHILLON & A. LONGINELLI. 1995. Oxygen-isotope variations of phosphate in mammalian bone and tooth enamel. Geochimica et Cosmochimica Acta 59: 4299-305.

MONTGOMERY, J. 2010. Passports from the past: investigating human dispersals using strontium isotope analysis of tooth enamel. Annals of Human Biology 37: 325-46.

POLLARD, A.M. 2009. What a long strange trip it's been: lead isotopes in archaeology, in A.J. Shortland, I.C. Freestone & T. Rehren (ed). From mine to microscope: advances in the study of ancient technology: 181-9. Oxford: Oxbow.

POLLARD, A.M. & C. HERON. 2008. Archaeological chemistry. Cambridge: Royal Society of Chemistry.

SILLEN, A. & M. KAVANAGH. 1982. Strontium and paleodietary research: a review. Yearbook of Physical Anthropology 25: 67-90.

SNOW, C.P. 1959. The two cultures and the scientific revolution (The Rede Lecture 1959). Cambridge: Cambridge University Press.

VAN DER MERWE, N.J. & J.C. VOGEL. 1978. [sup.13]C content of human collagen as a measure of prehistoric diet in woodland North America. Nature 276:815-16.

VOGEL, J.C. & N.J. VAN DER MERWE. 1977. Isotopic evidence for early maize cultivation in New York State. American Antiquity 42: 238-42.

A.M. Pollard, Research Laboratory for Archaeology &the History of Art, Dyson Perrins Building, South Parks Road, Oxford, OX1 3QY, UK (Email: mark.pollard@rlaha.ox.ac.uk)