The theoretical tensions between archaeological and (large-scale) aDNA accounts

There are a number of tensions between archaeological and aDNA accounts. Perhaps the most significant critique that is emerging centres upon the relationship between aDNA analyses and culture-historical approaches (e.g. Carlin Reference Carlin2018; Furholt 2018; Reference Furholt2019; Hakenbeck Reference Hakenbeck2019; Heyd Reference Heyd2017; Vander Linden Reference Vander Linden2016). This operates at several levels: some aDNA analyses focus specifically upon addressing similar research questions and themes to those that have been the traditional target of culture historians. Given the wealth of new data generated by aDNA research it is perhaps surprising that the narratives that are emerging from these studies are so familiar. Take the following quote:

This could come as easily from Gordon Childe as from its actual origins in Brace et al. (Reference Brace, Diekmann, Booth, Faltyskova, Rohland, Mallick, Ferry, Michel, Oppenheimer, Broomandkhoshbacht, Stewardson, Walsh, Kayser, Schulting, Craig, Sheridan, Pearson, Stringer, Reich, Thomas and Barnes2018, 6).Footnote ² Just as Childe saw change in prehistory as a product of the movement of peoples bringing with them their material culture and practices, so Brace et al. (2018; Reference Brace, Diekmann, Booth, Faltyskova, Rohland, Mallick, Ferry, Michel, Oppenheimer, Broomandkhoshbacht, Stewardson, Walsh, Kayser, Schulting, Craig, Sheridan, Pearson, Stringer, Reich, Thomas and Barnes2019) see the emergence of the Neolithic in Britain as the product of the movement of groups of genetic signatures. Whilst in Childe’s argument there was no necessary connection between the culture of a particular individual and their biological identity, though this was often implicit, Brace et al. (Reference Brace, Diekmann, Booth, Faltyskova, Rohland, Mallick, Ferry, Michel, Oppenheimer, Broomandkhoshbacht, Stewardson, Walsh, Kayser, Schulting, Craig, Sheridan, Pearson, Stringer, Reich, Thomas and Barnes2019) make this link explicit. For those archaeologists who embrace more complex versions of identity (and indeed change), these arguments can seem reductive. After all, anthropologists have long demonstrated the issues with these concepts (e.g. Barth Reference Barth1969), and we have had fifty years of archaeological critique of culture-historical models of the type that Childe put forward. Indeed, since the 1980s archaeology has embraced a multifaceted understanding of identity (Diaz-Andreu and Lucy Reference Diaz-Andreu, Lucy, Diaz-Andreu, Lucy, Babic and Edwards2005; Fowler Reference Fowler2004; Jones Reference Jones1997; Voss Reference Voss, Casella and Fowler2004). Whilst both Brace et al. (Reference Brace, Diekmann, Booth, Faltyskova, Rohland, Mallick, Ferry, Michel, Oppenheimer, Broomandkhoshbacht, Stewardson, Walsh, Kayser, Schulting, Craig, Sheridan, Pearson, Stringer, Reich, Thomas and Barnes2019) and Olalde et al. (Reference Olalde, Brace, Allentoft, Armit, Kristiansen, Booth, Rohland, Mallick, Szécsényi-Nagy, Mittnik, Altena, Lipson, Lazaridis, Harper, Patterson, Broomandkhoshbacht, Diekmann, Faltyskova, Fernandes, Ferry, Harney, de Knijff, Michel, Oppenheimer, Stewardson, Barclay, Alt, Liesau, Ríos, Blasco, Miguel, García, Fernández, Bánffy, Bernabò-Brea, Billoin, Bonsall, Bonsall, Allen, Büster, Carver, Navarro, Craig, Cook, Cunliffe, Denaire, Dinwiddy, Dodwell, Ernée, Evans, Kuchařík, Farré, Fowler, Gazenbeek, Pena, Haber-Uriarte, Haduch, Hey, Jowett, Knowles, Massy, Pfrengle, Lefranc, Lemercier, Lefebvre, Martínez, Olmo, Ramírez, Maurandi, Majó, McKinley, McSweeney, Mende, Modi, Kulcsár, Kiss, Czene, Patay, Endrődi, Köhler, Hajdu, Szeniczey, Dani, Bernert, Hoole, Cheronet, Keating, Velemínský, Dobeš, Candilio, Brown, Fernández, Herrero-Corral, Tusa, Carnieri, Lentini, Valenti, Zanini, Waddington, Delibes, Guerra-Doce, Neil, Brittain, Luke, Mortimer, Desideri, Besse, Brücken, Furmanek, Hałuszko, Mackiewicz, Rapiński, Leach, Soriano, Lillios, Cardoso, Pearson, Włodarczak, Price, Prieto, Rey, Risch, Rojo Guerra, Schmitt, Serralongue, Silva, Smrčka, Vergnaud, Zilhão, Caramelli, Higham, Thomas, Kennett, Fokkens, Heyd, Sheridan, Sjögren, Stockhammer, Krause, Pinhasi, Haak, Barnes, Lalueza-Fox and Reich2018) set out more complex scenarios in continental Europe, the models that they use to interpret the data in Britain clearly evoke culture-historical parallels. Indeed, these presuppositions more widely underlie aDNA research, as Martin Furholt (Reference Furholt2018) has pointed out (see also Thomas Reference Thomas2006, 52).

There are three central assumptions that underlie both culture-historical research and the work of some aDNA specialists.Footnote ³ First, it is assumed that identity is fixed at birth and consistent through the lifespan; if you are born a member of the Beaker people (whether that is culturally or genetically defined) you remain a member of that group beyond your death (cf. Furholt Reference Furholt2019). Second, these individual identities are strictly exclusionary categories; one cannot be both a Mesolithic hunter–gatherer and a Neolithic farmer (there is no intersectionality for culture historians or many aDNA specialists). Third, that change has two main sources – either diffusion or migration. Change comes from the outside and effectively involves replacement of one group of people (or way of life) with another that is technologically superior. For example, papers in the literature widely accept the replacement of one population in Europe by another at the start of the Neolithic, with various levels of admixture with the former group along the way (e.g. Brace et al. Reference Brace, Diekmann, Booth, Faltyskova, Rohland, Mallick, Ferry, Michel, Oppenheimer, Broomandkhoshbacht, Stewardson, Walsh, Kayser, Schulting, Craig, Sheridan, Pearson, Stringer, Reich, Thomas and Barnes2018; Lipson et al. Reference Lipson, Szécsényi-Nagy, Mallick, Pósa, Stégmár, Keerl, Rohland, Stewardson, Ferry, Michel, Oppenheimer, Broomandkhoshbacht, Harney, Nordenfelt, Llamas, Gusztáv Mende, Köhler, Oross, Bondár, Marton, Osztás, Jakucs, Paluch, Horváth, Csengeri, Koós, Sebők, Anders, Raczky, Regenye, Barna, Fábián, Serlegi, Toldi, Gyöngyvér Nagy, Dani, Molnár, Pálfi, Márk, Melegh, Bánfai, Domboróczki, Fernández-Eraso, Antonio Mujika-Alustiza, Alonso Fernández, Jiménez Echevarría, Bollongino, Orschiedt, Schierhold, Meller, Cooper, Burger, Bánffy, Alt, Lalueza-Fox, Haak and Reich2017). Swiftly, however, this description of genetic change becomes seen as an explanation for what took place. For example, Lipson et al. (Reference Lipson, Szécsényi-Nagy, Mallick, Pósa, Stégmár, Keerl, Rohland, Stewardson, Ferry, Michel, Oppenheimer, Broomandkhoshbacht, Harney, Nordenfelt, Llamas, Gusztáv Mende, Köhler, Oross, Bondár, Marton, Osztás, Jakucs, Paluch, Horváth, Csengeri, Koós, Sebők, Anders, Raczky, Regenye, Barna, Fábián, Serlegi, Toldi, Gyöngyvér Nagy, Dani, Molnár, Pálfi, Márk, Melegh, Bánfai, Domboróczki, Fernández-Eraso, Antonio Mujika-Alustiza, Alonso Fernández, Jiménez Echevarría, Bollongino, Orschiedt, Schierhold, Meller, Cooper, Burger, Bánffy, Alt, Lalueza-Fox, Haak and Reich2017, 368, our emphasis) declare that ‘ancient DNA analysis has validated major migrations from populations related to Neolithic Anatolians as driving the introduction of farming in Europe’. Historically it cannot be the case that the migrations drove the spread of farming, because migration is in itself a historical event in need of explanation. Thus Lipson et al.’s (Reference Lipson, Szécsényi-Nagy, Mallick, Pósa, Stégmár, Keerl, Rohland, Stewardson, Ferry, Michel, Oppenheimer, Broomandkhoshbacht, Harney, Nordenfelt, Llamas, Gusztáv Mende, Köhler, Oross, Bondár, Marton, Osztás, Jakucs, Paluch, Horváth, Csengeri, Koós, Sebők, Anders, Raczky, Regenye, Barna, Fábián, Serlegi, Toldi, Gyöngyvér Nagy, Dani, Molnár, Pálfi, Márk, Melegh, Bánfai, Domboróczki, Fernández-Eraso, Antonio Mujika-Alustiza, Alonso Fernández, Jiménez Echevarría, Bollongino, Orschiedt, Schierhold, Meller, Cooper, Burger, Bánffy, Alt, Lalueza-Fox, Haak and Reich2017) paper, which seeks explicitly in its title and methodology to explore the complexity of genetic histories, reveals how quickly linear, singular models of change become accepted.

Similarly, Allentoft et al. (Reference Allentoft, Sikora, Sjögren, Rasmussen, Rasmussen, Stenderup, Damgaard, Schroeder, Ahlström, Vinner, Malaspinas, Margaryan, Higham, Chivall, Lynnerup, Harvig, Baron, Casa, Dąbrowski, Duffy, Ebel, Epimakhov, Frei, Furmanek, Gralak, Gromov, Gronkiewicz, Grupe, Hajdu, Jarysz, Khartanovich, Khokhlov, Kiss, Kolář, Kriiska, Lasak, Longhi, McGlynn, Merkevicius, Merkyte, Metspalu, Mkrtchyan, Moiseyev, Paja, Pálfi, Pokutta, Pospieszny, Price, Saag, Sablin, Shishlina, Smrčka, Soenov, Szeverényi, Tóth, Trifanova, Varul, Vicze, Yepiskoposyan, Zhitenev, Orlando, Sicheritz-Pontén, Brunak, Nielsen, Kristiansen and Willerslev2015) reveal at the close of their paper that aDNA research supports a ‘correspondence between cultural changes, migrations, and linguistic patterns’ in the Bronze Age. Although they caution that this cannot always be assumed to be the case, the fact that their paper begins with the declaration that ‘by 3000BC, the Neolithic farming cultures in temperate Eastern Europe appear to be largely replaced by the Early Bronze Age Yamnaya culture’ (Allentoft et al. Reference Allentoft, Sikora, Sjögren, Rasmussen, Rasmussen, Stenderup, Damgaard, Schroeder, Ahlström, Vinner, Malaspinas, Margaryan, Higham, Chivall, Lynnerup, Harvig, Baron, Casa, Dąbrowski, Duffy, Ebel, Epimakhov, Frei, Furmanek, Gralak, Gromov, Gronkiewicz, Grupe, Hajdu, Jarysz, Khartanovich, Khokhlov, Kiss, Kolář, Kriiska, Lasak, Longhi, McGlynn, Merkevicius, Merkyte, Metspalu, Mkrtchyan, Moiseyev, Paja, Pálfi, Pokutta, Pospieszny, Price, Saag, Sablin, Shishlina, Smrčka, Soenov, Szeverényi, Tóth, Trifanova, Varul, Vicze, Yepiskoposyan, Zhitenev, Orlando, Sicheritz-Pontén, Brunak, Nielsen, Kristiansen and Willerslev2015, 167) might lead the cynical reader to conclude that the research outcomes were predetermined by the sets of ideas imposed on the data from the outset.

Because genetic identity is elided with cultural identity here the notion that we can trace group movement through genetic sequencing is not in itself ever tested. As with the Allentoft et al. (Reference Allentoft, Sikora, Sjögren, Rasmussen, Rasmussen, Stenderup, Damgaard, Schroeder, Ahlström, Vinner, Malaspinas, Margaryan, Higham, Chivall, Lynnerup, Harvig, Baron, Casa, Dąbrowski, Duffy, Ebel, Epimakhov, Frei, Furmanek, Gralak, Gromov, Gronkiewicz, Grupe, Hajdu, Jarysz, Khartanovich, Khokhlov, Kiss, Kolář, Kriiska, Lasak, Longhi, McGlynn, Merkevicius, Merkyte, Metspalu, Mkrtchyan, Moiseyev, Paja, Pálfi, Pokutta, Pospieszny, Price, Saag, Sablin, Shishlina, Smrčka, Soenov, Szeverényi, Tóth, Trifanova, Varul, Vicze, Yepiskoposyan, Zhitenev, Orlando, Sicheritz-Pontén, Brunak, Nielsen, Kristiansen and Willerslev2015) paper above, many aDNA papers do not look for genetic evidence that cultural groups existed in that past but rather work from that premise at the outset. As Martin Furholt (2018, 168; Reference Furholt2019) has demonstrated, there are a number of possible explanations for the genetic changes we see in Europe in the third millennium B.C. (see also Vander Linden Reference Vander Linden2016). Rather than embracing and exploring the multiple ways in which the Neolithic emerged even across as small an area as Britain, Brace et al. (Reference Brace, Diekmann, Booth, Faltyskova, Rohland, Mallick, Ferry, Michel, Oppenheimer, Broomandkhoshbacht, Stewardson, Walsh, Kayser, Schulting, Craig, Sheridan, Pearson, Stringer, Reich, Thomas and Barnes2018) prefer to support a singular hypothesis (cf. Cummings and Morris Reference Cummings and Morris2018). As Horsburgh (Reference Horsburgh2018, 656) highlights, we are at risk of sliding towards hyper-reductionist thinking. There is complexity here as two worlds meet; for archaeologists, whose roots more firmly lie in the humanities, bringing out nuance, multiplicity and complexity are often raisons d’être, whereas for some geneticists simplicity can be its own virtue (Booth Reference Booth2019; Vander Linden Reference Vander Linden2016, 722). Recent papers, including those by Villanea and Schraiber (Reference Villanea and Schraiber2019) on interbreeding between Neanderthals and modern humans, and Schuenemann et al. (Reference Schuenemann, Avanzi, Krause-Kyora, Seitz, Herbig, Inskip, Bonazzi, Reiter, Urban, Dangvard Pedersen, Taylor, Singh, Stewart, Velemínský, Likovsky, Marcsik, Molnár, Pálfi, Mariotti, Riga, Belcastro, Boldsen, Nebel, Mays, Donoghue, Zakrzewski, Benjak, Nieselt, Cole and Krause2018) and Bos et al. (Reference Bos, Herbig, Sahl, Waglechner, Fourment, Forrest, Klunk, Schuenemann, Poinar, Kuch, Golding, Dutour, Keim, Wagner, Holmes, Krause and Poinar2016) on the spread of historical diseases, show this need not be the case.

Some authors have suggested that the argument that a selection of aDNA articles imposes an equation between people and culture is misleading (Eisenmann et al. Reference Eisenmann, Bánffy, van Dommelen, Hofmann, Maran, Lazaridis, Mittnik, McCormick, Krause, Reich and Stockhammer2018, 2). However, given the way this connection repeats itself across numerous papers, in addition to the manner in which the use of the data uncritically echoes culture history, we do not accept that this is the case. Furthermore, as Eisenmann et al. (Reference Eisenmann, Bánffy, van Dommelen, Hofmann, Maran, Lazaridis, Mittnik, McCormick, Krause, Reich and Stockhammer2018, 6) emphasize, even where the connection between archaeological culture and genetic signatures may be debated, the existence of archaeological cultures, as unambiguous historical entities, is not in doubt. Interestingly, even authors who have explicitly argued that aDNA research will help launch a new scientific revolution (e.g. Kristiansen Reference Kristiansen2014) also present interpretations that are fundamentally culture-historical in nature (e.g. Kristiansen et al. Reference Kristiansen, Allentoft, Frei, Iverson, Johannsen, Knussen, Pospieszny, Prices, Rasmussen, Sjören, Sikora and Willerslev2017). aDNA papers in archaeology are not at the forefront of a return to processual archaeology, with models of population dynamics and critical hypothesis testing, as one might expect. Instead they presage a return to older forms of thinking.Footnote ⁴

Data, regardless of the mechanisms through which they are produced, always remain theory-laden. There is no way out of an approach that hypothesizes the existence of bounded (and opposed) cultural identities in the past once you have started with that presupposition. As Lewis Binford (Reference Binford1968) and David Clark (Reference Clarke1973, 15) demonstrated so clearly, these kinds of approach merely describe the data they encounter rather than seeking to explain the historical and anthropological circumstances through which they arose. What this shows is that generating more aDNA sequences will in itself not be sufficient to create more complex narratives of the past; instead what is required is a different approach to the data themselves. Tensions arise here because of the social and political power that scientific – and especially aDNA – narratives have (Horsburgh Reference Horsburgh2015). These narratives are accepted by the public, by some aDNA specialists and by some archaeologists as basic facts against which other kinds of evidence can be weighed. They are explicitly held up as something archaeology must be tested against (e.g. Brandt et al. Reference Brandt, Szécsényi-Nagy, Roth, Alt and Haak2015, 87) and not the other way around (Sørensen Reference Sørensen2017, 101–102). Interpretations of aDNA present these data as neutral arbiters in a way that they could never be. They are a product of research strategies, sampling selections, statistical modelling, particular forms of presentation, generalization and expansion that reflect the aims and ambitions of the people generating the data (see, for example, Fujimura et al. Reference Fujimura, Bolnick, Rajagopalan, Kaufman, Lewontin, Duster, Ossorio and Marks2014 on the effect of principal-component analysis). To be clear, that is not a criticism of aDNA research as against any other form of investigation; this is true of all research in all disciplines (Latour Reference Latour1987; Shanks and Tilley Reference Shanks and Tilley1987; Sørensen Reference Sørensen2017). The issue is that its contextual nature is not embraced and acknowledged. These tensions thus create debates about how to get genetic and archaeological data to match (e.g. Eisenmann et al. Reference Eisenmann, Bánffy, van Dommelen, Hofmann, Maran, Lazaridis, Mittnik, McCormick, Krause, Reich and Stockhammer2018) or demands for more integrative ways of working (e.g. Johannsen et al. Reference Johannsen, Larson, Meltzer and Vander Linden2017, 1120). As we will see below, this tension reveals a deeper underlying reliance on binaries that requires attention, and a philosophical reconsideration.

The political tensions in aDNA

Before we move on to set out the philosophical critique of aDNA approaches, we need also to address the political and ethical elephants in the corner (see Frieman and Hofmann Reference Frieman and Hofmann2019; Hakenbeck Reference Hakenbeck2019). Culture history is not simply problematic because of its presuppositions about identity and change. The model it rests upon also implicitly embraces a political position that reflects a reified version of 19th- and 20th-century Europe. All theory is of its time. This was a world where people sought to clearly define the boundaries of nation states and the identities of those who lived within them, and it was a world where Europeans saw themselves as technologically and intellectually superior to their counterparts elsewhere. These approaches have been thoroughly critiqued and deconstructed through post-colonial theory (Bhabha Reference Bhabha1994; Said Reference Said1978; Spivak Reference Spivak, Nelson and Grossberg1988; Wolf Reference Wolf1982). Not only do such approaches fail to give adequate credit to the complexity of non-European civilizations, they also deny the always multicultural and multifaceted nature of all cultural groups (European or otherwise). Today, post-colonialism in archaeology brings to the fore the multiple and complex nature of colonial engagement; it emphasizes hybrids and creolization as well as highlighting oppression and resistance rather than discussing simplistic models of invasion and replacement (Battle-Baptiste Reference Battle-Baptiste2011; Gosden Reference Gosden2004; Van Dommelen Reference van Dommelen, Lyons and Papadopoulos2002; Voss 2008; Reference Voss2015). As Furholt (Reference Furholt2018, 170) has argued, as long as our approaches to genetic evidence continue to rest upon these kinds of assumption, we risk lending spurious scientific legitimacy to nationalist politics. The utilization of culture history in the 1930s stands as a stark warning here.

The risk is not just in the political ramifications of the data and narratives we produce but in how we treat those with whom we work. It might be relatively uncontroversial to talk about different groups of Europeans 5,000 years ago, but it is not unproblematic to talk in the same way about the histories of those who have been victim to colonial oppression (cf. MacEachern Reference MacEachern2012). This requires not only, as Mary Prendergast and Elizabeth Sawchuk (Reference Prendergast and Sawchuk2018) have recently highlighted, consultation with local communities and the development of protocols for informed consent and the sampling of human remains for aDNA analysis, but also an understanding that the narratives we generate in the present can have significant political implications for people today. In David Reich’s recent book on aDNA (Reference Reich2018, 163), he states that modern studies of DNA variation in Native American groups are a ‘force for good’ and makes it clear that he is frustrated at the lack of engagement from such groups. As Horsburgh’s (Reference Horsburgh2018, 657) review of the book highlights, not only does this fail to understand structural inequality, but furthermore Reich is not in a position to define what constitutes harm for indigenous groups. Anyone doubting the political salience of these points needs only to examine the recent debates around the US Senator Elizabeth Warren’s DNA test and Native American ancestry (TallBear Reference TallBear2018; cf. TallBear Reference TallBear2013).

We thus welcome and support the developing critiques of archaeologists and geneticists that have flagged up key theoretical, political and ethical issues for working with aDNA (e.g. Frieman and Hofmann Reference Frieman and Hofmann2019; Furholt Reference Furholt2018; Hakenbeck Reference Hakenbeck2019; Horsburgh Reference Horsburgh2018; Prendergast and Sawchuk Reference Prendergast and Sawchuk2018; Vander Linden 2016; Reference Vander Linden2018, as well as the reflexive commentary of aDNA specialists themselves); however, as we noted in our introduction, we seek to take a different tack. Rather than further revisit the problems of culture history we suggest that we can also begin to deconstruct the philosophical approach behind aDNA research and offer an alternative.

Binary molecules, binary models, binary answers

Research into aDNA, like so much of archaeology, rests upon a series of binary oppositions (Harris and Cipolla Reference Harris and Cipolla2017; Jones Reference Jones2002; Thomas Reference Thomas2004). The opposition of cultural group and genetic signature maps neatly onto the classic opposition between culture and nature. In models of these kinds, nature is presented as universal and best understood through the mechanisms of science, generating singular explanations. Culture, by contrast, is understood effectively to be an add-on, the purview of anthropologists and sociologists where cultural interpretations can be multiple and run counter to the ‘factual logic’ of Western science. This distinction also maps onto contrasts between ontology on the one hand (what the world actually is) and epistemology (what we think about the world), the former always singular and the latter always multiple (Alberti Reference Alberti2016).

This distinction between science as the singular truth and culture as the confusing multiple runs through the recent article by Eisenmann et al. (Reference Eisenmann, Bánffy, van Dommelen, Hofmann, Maran, Lazaridis, Mittnik, McCormick, Krause, Reich and Stockhammer2018). The authors discuss how we might reconcile material culture (read ‘culture’) with genetic data (read ‘nature and science’) – the two stand opposed from the outset. Eisenmann et al. (Reference Eisenmann, Bánffy, van Dommelen, Hofmann, Maran, Lazaridis, Mittnik, McCormick, Krause, Reich and Stockhammer2018, 6) comment that there is no ‘universal explanation for what stands behind an archaeological culture’. They also place material culture firmly in the secondary position when it comes to generating knowledge about the past – we must rely on material culture, we are told, when there are no eyewitnesses, sociologists or written sources that can tell us about group identity (ibid., 6). As aDNA specialists remark, aDNA offers ‘a solid genetic framework against which archaeological and linguistic models can be tested’ (Brandt et al. Reference Brandt, Szécsényi-Nagy, Roth, Alt and Haak2015, 87; cf. Vander Linden Reference Vander Linden2016, 720). David Reich (Reference Reich2018, xx) has stated that ‘human genome variation has surpassed the traditional toolkit of archaeology’, and aDNA now offers ‘constraints’ to other forms of interpretation (Olalde et al. Reference Olalde, Brace, Allentoft, Armit, Kristiansen, Booth, Rohland, Mallick, Szécsényi-Nagy, Mittnik, Altena, Lipson, Lazaridis, Harper, Patterson, Broomandkhoshbacht, Diekmann, Faltyskova, Fernandes, Ferry, Harney, de Knijff, Michel, Oppenheimer, Stewardson, Barclay, Alt, Liesau, Ríos, Blasco, Miguel, García, Fernández, Bánffy, Bernabò-Brea, Billoin, Bonsall, Bonsall, Allen, Büster, Carver, Navarro, Craig, Cook, Cunliffe, Denaire, Dinwiddy, Dodwell, Ernée, Evans, Kuchařík, Farré, Fowler, Gazenbeek, Pena, Haber-Uriarte, Haduch, Hey, Jowett, Knowles, Massy, Pfrengle, Lefranc, Lemercier, Lefebvre, Martínez, Olmo, Ramírez, Maurandi, Majó, McKinley, McSweeney, Mende, Modi, Kulcsár, Kiss, Czene, Patay, Endrődi, Köhler, Hajdu, Szeniczey, Dani, Bernert, Hoole, Cheronet, Keating, Velemínský, Dobeš, Candilio, Brown, Fernández, Herrero-Corral, Tusa, Carnieri, Lentini, Valenti, Zanini, Waddington, Delibes, Guerra-Doce, Neil, Brittain, Luke, Mortimer, Desideri, Besse, Brücken, Furmanek, Hałuszko, Mackiewicz, Rapiński, Leach, Soriano, Lillios, Cardoso, Pearson, Włodarczak, Price, Prieto, Rey, Risch, Rojo Guerra, Schmitt, Serralongue, Silva, Smrčka, Vergnaud, Zilhão, Caramelli, Higham, Thomas, Kennett, Fokkens, Heyd, Sheridan, Sjögren, Stockhammer, Krause, Pinhasi, Haak, Barnes, Lalueza-Fox and Reich2018, 194). It seems that there is not only a divide between nature and culture but also an internal separation between which explanations of culture are seen as superior.

These binaries go beyond nature versus culture; the genetic evidence is used to divide people neatly into opposed sexes with no consideration of the complexity of either biology or gender (Butler Reference Butler1993; Robb and Harris Reference Robb and Harris2013). Similarly, the groups detected provide little in the way of nuance when it comes to identity; people are either Anatolian farmers or Western European hunter–gatherers, as we noted above. Geneticists construct their groups using statistical methods that seek to gather together those who share more in common genetically than they do with others – this effectively creates an in-group/out-group situation – here there is no space for the person who both plants seeds and hunts deer. Given the way in which any person’s ancestry includes many people with whom they share no genetic overlap (Reich Reference Reich2018, 37), this approach simplifies ideas of ancestry, descent, and identity to a level with which any social scientist would be deeply uncomfortable (Fujimura et al. Reference Fujimura, Bolnick, Rajagopalan, Kaufman, Lewontin, Duster, Ossorio and Marks2014).

The simple fact is that when you start with binaries you are bound to end up with them. Furthermore, these binary distinctions are not scientific facts but rather inherited categories of thought (Latour Reference Latour1993). For several decades now, philosophers, sociologists, anthropologists, critical thinkers and, indeed, archaeologists have sought to unpick the ways in which dualistic thinking has limited and delineated the possibilities for investigation and interpretation (Deleuze and Parnet Reference Deleuze and Parnet2002; Descola Reference Descola2013; Latour 1993; Reference Latour1999). Within archaeology, dualistic approaches were first denounced in the early 1980s (Hodder Reference Hodder and Hodder1982), though they remained firmly part of the postprocessual mode of thought. From the mid-1990s onwards, and particularly through the work of Andrew Jones (Jones Reference Jones2002) and Julian Thomas (Reference Thomas2004), dualisms have been the explicit focus of archaeological critique. In the last ten years archaeologists from a range of theoretical approaches from symmetrical archaeology (e.g. Olsen et al. Reference Olsen, Shanks, Webmoor and Witmore2012) to New Materialism (e.g. Conneller Reference Conneller2011; Jones Reference Jones2012) have continued to undermine and deconstruct the effects that dualistic thinking has had, and continues to have, on our interpretations. It is notable that in response to this extended critique there has been little in the way of defence of dualisms: scholars either embrace the fact that this thinking is deeply damaging to our attempts to understand the past or simply ignore the argument against dualisms and carry on with business as usual.

As an example of how unhelpful binaries are, let us turn to DNA itself. DNA is often understood as nature embodied. What could be a more pure form of nature than our DNA? It is something that can only be revealed to us through science. Shaped by evolution, subject to the laws of biology, DNA, and the genetic information it contains, clearly fall squarely on the nature side of the nature–culture divide. On the other hand, though, consider this: your DNA is the product of social and historical processes of both the short and the very long term. Ideas of class, race and nationality, all modern impositions, have shaped who people have reproduced with, where their descendants have moved to, and therefore people’s DNA signatures (Roseman Reference Roseman2014). Over a longer time frame, many more historical processes come into play, including those that aDNA research purports to reveal, such as the spread of farming and the arrival of metalwork. Beyond this, however, DNA is a chemical that responds to the world in which it is enmeshed; as Ingold (Reference Ingold2000) famously pointed out, the old biological maxim of genotype plus environment equals phenotype mistakenly imposes the idea that there is anywhere in existence a genetic sequence absent from its environment (cf. Pigluicci Reference Pigluicci2010). The human genome has not evolved, and does not exist, separately from what we might term ‘culture’; the two are interwoven from the start. From our use of stone tools, to the ability of some of us to digest milk, via countless other mutations, alterations and transformations, human genomes are as thoroughly social as one could possibly imagine.

Why, then, continue to divide the world up into these binary categories? As Latour (Reference Latour1993) has shown, these oppositions are not merely present in the world; they are actively produced through a process he calls purification. People go to great lengths to make nature and culture separate from one another, and when such oppositions are threatened they can react violently. The reason why modern medical biotechnologies (which seem to endanger the sanctity of the human body) and the ever more apparent movement of people across national borders threaten people’s understanding of the world is because of the way in which these challenge our dualistic oppositions (Robb and Harris Reference Robb and Harris2013, Chapter 8). Patriarchal social structures rest upon the opposition of male and female and the association of the latter with nature and the body. A rejection of dualisms, therefore, is scientifically more accurate, politically necessary and ethically essential in the world we face today. Indeed, this complexity is recognized by many scientists, anthropologists and others working with biological evidence (e.g. Dunn, Reese and Eisenhauer Reference Dunn, Reese and Eisenhauer2019; Haraway Reference Haraway2008; Tsing Reference Tsing2015). Although it is beyond the purview of this paper, any attempt to challenge global warming will rest upon our ability to challenge the nature–culture divide (Latour Reference Latour2018). Time, indeed, for something new.

Beyond binaries: post-humanism

Let us summarize, then, what our new approach to aDNA will require. First, it is clear that we need to situate our understanding of aDNA within an approach that does not divide the world into binary oppositions if we want both an accurate understanding of the past and to prevent the imposition of modernist ways of thinking. Second, we need an approach to identity that does not privilege a particular mode of being human (one all too familiar today), nor one that bounds the body off from the world around it. It is the radical separation of body from environment (another dualism) that reduces the body to the status of nature, and thus DNA to the essence of this nature. Finally, we will need an approach that does not allow one strand of evidence to ride roughshod over others. We cannot adequately understand the past if we continue to privilege specific forms of knowledge. This too would return us to a world of dualisms, and all the problems this entails.

In order to develop this approach, we advocate a framework rooted amongst the complex ontological positions beginning to be advocated across the discipline of archaeology (for a review see Harris and Cipolla Reference Harris and Cipolla2017). Specifically, we suggest that we need to draw on elements of assemblage thinking (Deleuze and Guattari Reference Deleuze and Guattari2004; DeLanda Reference DeLanda2016; Jervis Reference Jervis2018; papers that follow Jones and Hamilakis Reference Jones and Hamilakis2017, inter alia). This provides us with an apparatus for thinking through the past in ways that do not ontologically privilege science or humans, and allows us to move beyond binary thinking; we return to these ideas below. The related and compatible ideas of post-humanism (Braidotti Reference Braidotti2013; Fredengren Reference Fredengren2013; Harris Reference Harris, Pierce, Russell, Maldonado and Cambell2016) emphasize the need for a non-dualistic approach to the world and to embrace and appreciate the diversity of humanity.

Whilst a full description of the multiple post-humanist approaches is beyond the scope of this paper, they share a rejection of the central tenets of humanism: that human beings hold unique ontological status. Furthermore, post-humanist approaches argue that only a certain subgroup of humans have ever been granted full membership of the human category (Braidotti Reference Braidotti2013), that the world does not begin with prefigured entities (e.g. people and things) but rather emerges from relations, and that following from this we cannot artificially separate the world into neatly bounded categories such as nature and culture. For post-humanists, people and things emerge from a world of relations in which they are always already enmeshed. As we saw with the human genome above, no human being is born outside relations, and there is no relationship that can be easily categorized as either natural or cultural. Indeed, this is demonstrated by the work of ecologists and biologists in a number of different fields (see, for example, Cardinale et al. Reference Cardinale, Duffy, Gonzalez, Hooper, Perrings, Venail, Narwani, Mace, Tilman, Wardle, Kinzig, Daily, Loreau, Grace, Larigauderie, Srivastava and Naeem2012; Dunn, Reese and Eisenhauer Reference Dunn, Reese and Eisenhauer2019).

This radically relational approach to the world opens up new ways of thinking about the interrelationship of humans and animals (Haraway Reference Haraway2008), humans and things (Braidotti Reference Braidotti2013), humans and flora (Ingold Reference Ingold2000), and humans and fungi (Tsing Reference Tsing2015). More than this, even, this approach to relations does not require us to include human beings at all, so it is not merely an approach that investigates relations between humans + X but rather forces us to jump into the middle of things, a web of relations which connect and produce the world. A flat ontology, which many post-humanists adopt, posits that all elements of the world emerge from relationships, including those that do not involve human beings. Such a flat ontology does not propose that all elements of the world are the same, but that we cannot understand the world by ontologically elevating the human category from the outset. Humans are undoubtedly different from buttercups, but buttercups are different from oak trees (cf. Dawney, Harris and Sørensen Reference Dawney, Harris and Sørensen2017, 122). It makes no sense to elevate one of these things to a unique ontological plane if we wish to understand how the world comes into existence. Furthermore, taking a post-humanist approach demands of us an explicitly political engagement with both past and present (Braidotti Reference Braidotti2013). When the material being discussed and the interpretations produced are so potentially powerful politically, nothing less than a committed and engaged political stance will suffice. Simply burying our heads in the sand and proclaiming the protection of scientific objectivity denies the ever real presence of political power, influence, control and histories of complex colonial and post-colonial engagements. This is as true of studies of European identity, past and present, as it is in the contested world of Indigenous politics.

Such an approach has obvious implications for the human body. No longer seen as natural, bodies emerge from relations between a wide variety of entities. Our human body is not just the product of biology, but is shaped by the chairs we sit in, the foods we eat, the world we explore, the air we breathe, the plants that are in our offices, the other people we interact with, the microbes on our skin, and the animals we live alongside. This shaping is in no way simply cultural; it affects our skeletons, our muscles, our brains and indeed our DNA. As noted above, this shaping is both short-term (what a biologist would term acclimatization), medium-term (the plasticity of our body in response to action) and at an evolutionary timescale (termed adaptation) (Sofaer Reference Sofaer2006). We are the product not just of our human genealogies but of a history of interactions with things from choppers in Olduvai Gorge, via Terra Sigillata in Rome, to iPhones today. There is no point at which our genome can be separated from our history, there is no point at which our genome is ever natural, and there is no point in this sequence when our genome stopped changing (cf. Ingold Reference Ingold2000). Not just humans either; the diseases that shape our bodies and population dynamics have their own complex genetic histories too (Bos et al. Reference Bos, Herbig, Sahl, Waglechner, Fourment, Forrest, Klunk, Schuenemann, Poinar, Kuch, Golding, Dutour, Keim, Wagner, Holmes, Krause and Poinar2016; Schuenemann et al. Reference Schuenemann, Avanzi, Krause-Kyora, Seitz, Herbig, Inskip, Bonazzi, Reiter, Urban, Dangvard Pedersen, Taylor, Singh, Stewart, Velemínský, Likovsky, Marcsik, Molnár, Pálfi, Mariotti, Riga, Belcastro, Boldsen, Nebel, Mays, Donoghue, Zakrzewski, Benjak, Nieselt, Cole and Krause2018). This is to say nothing of the complex world of epigenetics (cf. Niewöhner Reference Niewöhner2011).

The human genome thus shifts over these multiple scales. Whilst we are used to the idea that genomes alter over the long term, from our very conception our DNA sequences are changing with potentially transgenerational consequences (Bjornsson et al. Reference Bjornsson, Sigurdsson, Fallin, Irizarry, Aspelund, Vui, Yu, Rongione, Ekström, Harris, Launer, Eiriksdottir, Leppert, Sapienza, Gudnason and Feinberg2008; Jackson and Bartek Reference Jackson and Bartek2009). Cell cycles produce occasional but regular mutations (Drake et al. Reference Drake, Charlesworth, Charlesworth and Crow1998), changing the structure and make-up of our DNA sequences. Sunlight degrades the DNA in our skin; cigarette smoke alters the molecules in our lungs. Within any human body there are countless organisms with their own specific genetic histories – there are bacteria that dwell in our guts without which we could not digest our food; there are microbes on our skin that help us fight infections; our interactions with animals leave their DNA signature within us, as the work of Donna Haraway (Reference Haraway2008) so evocatively describes; the cells of other humans with whom we come into contact leave traces on the surface of our skin. The science of DNA fingerprinting, crucial to criminal procedures in the 21st century, rests upon the fact that bits of our DNA are left behind wherever we go. Bodies leak, and they are permeable.

DNA is thus an example of post-humanism par excellence; it is as far from the definition of individual human identity as one can imagine. Rather than revealing the pure essence of who we are, it discloses our deeply relational post-human histories, histories that include not just people but non-humans too. To be clear, we do not mean by this that the scientific analysis of aDNA specialists is somehow mistaken, or has identified the wrong DNA; rather, we take issue with the way in which these data are being thought about from the outset. Seeing DNA as the essence of our biology and identity is not only false (cf. Pigluicci Reference Pigluicci2010), but also serves as the basis for interpretations which continue to operate with a nature–culture divide wrought through their heart, and interpretations which continue to privilege science above all other forms of knowledge. It is not the case that we need to do the aDNA research again; it is just the case that we need to think about the data differently by starting from a different philosophical position. The irony is that the process of extracting aDNA demands cleanliness and purity and it seems that the interpretation of the results seeks this in the past too. The cleanliness of the lab is matched by the cleanliness of the science, yet both past identity and indeed DNA are far more complex than that. Whilst purity in the present may be necessary it can only be an abstraction in the past.

If post-human theory helps us to take a non-binary approach that radically decentres the human body as a bounded object, what might it do for our broader understanding of the way in which we integrate aDNA into our understandings of the past? To answer this, we need to turn to the assemblage thinking mentioned above (Jervis Reference Jervis2018). Assemblage thinking, developed from readings of Deleuze and Guattari, emphasizes that the world is not made up of bounded, fixed entities, but rather of temporary heterogeneous gatherings (termed assemblages) that are always in the process of becoming. Such an approach forces archaeologists to account for the processes by which assemblages come into being, and the forces that sustain them or cause them to fall apart. Assemblage thinking makes no distinction between matter and ideas, nature and culture or any other form of binary opposition. It has the great advantage of being both a theoretical approach that we can apply in the past, and a means of thinking through our practices in the present (Fowler Reference Fowler2013; Lucas Reference Lucas2012). This means we can use it not only to think about the assemblage of past identities (e.g. Harris Reference Harris, Pierce, Russell, Maldonado and Cambell2016), but also to reconceptualize how we, as archaeologists, assemble our data into narratives about that past.

Approaching our data as an assemblage has three consequences. First, it refuses to place ‘science’ and ‘interpretation’ in differing ontological realms, with the former privileged over the latter (Harris Reference Harris, Whittle and Bickle2014). Assemblage thinking famously begins with a flat ontology (DeLanda Reference DeLanda2002). This means that we cannot take a position where our ‘subjective’ interpretations of ‘culture’ are tested against our ‘objective’ scientific facts, but instead requires us to attend to both these elements of the assemblage equally. Second, assemblage thinking emphasizes the relational, and multiple, nature of our evidence and of the world itself. Rather than demanding that we produce a singular story, where each element of the evidence matches the others, we can instead embrace a more complex and messy version where different lines of evidence tell us different things. Rather than demanding singularity, instead we have multiplicity; not one thing or the other, rather one thing and the other (cf. Deleuze and Parnet Reference Deleuze and Parnet2002, 57–59). Third, the consequence of this is that aDNA is one aspect of our data amongst many others. Whilst it is undoubtedly informative and important it should hold no special status.

In a recent analysis of mixed-methods research in the social sciences, Emma Uprichard and Leila Dawney (Reference Uprichard and Dawney2019) have argued that one of the issues researchers face when integrating differing forms of data is the desire to always make the discrepancies vanish. Uprichard and Dawney (Reference Uprichard and Dawney2019) discuss how the combination of quantitative and qualitative methods is viewed as a key advantage for the study of social phenomena. This can be seen as equivalent to the manner in which archaeology draws upon both scientific and interpretive approaches. It is common practice within the social sciences to appreciate the value of both the richly textured qualitative data and the robust quantitative data that mixed-methods research generates. Uprichard and Dawney (Reference Uprichard and Dawney2019) recognize the necessity for mixed-methods research to capture the complexity, messiness and multiplicity of social data. Equally, however, they are critical of the ways in which scholars seek to flatten difference in order to produce singular and homogeneous accounts from their data. In contrast they urge researchers to embrace the messiness of the data, and to explore contradiction and tension rather than simply to supress and ignore it. In a manner analogous to assemblage theory, they argue that multiple modes of data show us different aspects of the assemblage under study, and diffract and multiply what we can say about our object of research. As they state, ‘although data integration is a sensible goal, we challenge the presupposition that it is necessarily the optimal outcome of mixed methods research’ (ibid., 19, original emphasis). There is much to learn here for archaeologists. First, we need to embrace examples where our data contradict each other. Rather than presuming that one strand of evidence is faulty if it refuses to conform to others, we need to understand that the complexity of the past will always mean that there are contradictory stories to tell. Second, we need to acknowledge, again, that no one form of data, theory or interpretation can necessarily take primacy over others. Third, it also challenges the widespread suggestion that all will be well if archaeologists and aDNA specialists simply work more closely together (e.g. Johannsen et al. Reference Johannsen, Larson, Meltzer and Vander Linden2017) or develop a common vocabulary (Eisenmann et al. Reference Eisenmann, Bánffy, van Dommelen, Hofmann, Maran, Lazaridis, Mittnik, McCormick, Krause, Reich and Stockhammer2018, 2). No doubt there is much to learn on both sides. However, what Uprichard and Dawney teach us is that no amount of collaboration can be guaranteed to iron out difference, contradiction and complexity. More than this, we can recognize that the desire to remove this complexity is part of a wider attempt to construct visions of the past that deny ambiguity. Feminist scholars like Joan Gero (Reference Gero2007; cf. Sørensen Reference Sørensen2016) have been deeply critical of such approaches.

Identity, so often the implicit focus of aDNA studies, is an inherently messy, multiple and contradictory subject. Research into identity frequently shows how we hold multiple and intersectional identities all at the same time. It also highlights how identity is not fixed but is always changing, in process, better considered as an event than a fixed essence (Puar Reference Puar2012). It also shows that different aspects of identity come to be prioritized in different ways at different moments in time. Sometimes it is more important that we are archaeologists, at other times it matters more that we are siblings, parents or friends. One of the problems with the model of identity that aDNA researchers work within is that it prioritizes singular and simplistic understandings of what identity is. The complex nature of identity means that we will always require mixed methods to study it. For the archaeologist this means that aDNA data alone will never be enough; we also need close contextual analysis of material culture, architecture and the human body. The data here are unlikely to align, but rather than that being a problem it instead is revealing of the complex reality of past lives. More than this, however, as feminist philosophers like Rosi Braidotti (Reference Braidotti2013; Reference Braidotti2019) have shown, a key aim of post-humanist approaches is not to focus upon the perceived majority at the expense of others, the minor stories that flow and interweave around dominant narratives that structure our understandings of the past (Braidotti Reference Braidoitti2011, 30–31).

Beyond this methodological issue, the commitment to complexity in our data and in our methods, in our embracing of ambiguity and contingency, also has political consequences. It is the privileging of one mode of knowing that has created so many of the problems we see around aDNA. It is not the case that simply working more closely with indigenous groups will mean they ‘understand’ why we ‘need’ to study their DNA (TallBear Reference TallBear2013). If DNA results reveal something surprising about your family background this is only one aspect of who you are, and not necessarily a meaningful one (O’Sullivan et al. Reference O’Sullivan, Posth, Coia, Schuenemann, Price, Wahl, Pinhasi, Zink, Krause and Maixner2018; TallBear Reference TallBear2013). If aDNA says that people in Bronze Age Britain had large amounts of steppe ancestry, this is only one aspect of our understanding of both who these people were, and who they thought they were. DNA is no more truthful (or false) than any other element of the evidence we draw on to think about the past. Rather it is relational, formed in and through relations, studied in relation with archaeologists and scientists in the present, and can contribute to multiple narratives about the past. This call for multiplicity should not, however, be dismissed as mere relativism. Rather, as Deleuze and Guattari (Reference Deleuze and Guattari1994, 130) argue, this is not about the relativity of truth, but rather the truth of relations.

Implications for aDNA research

To summarize our argument, it is clear that we need to adopt a post-humanist assemblage-inflected approach to the past if we are to understand aDNA outside the problematic binary oppositions that currently structure its interpretation in terms of both the approach towards science and the dominance of the nature–culture dualism. Such a move will also let us accept the inherent ambiguity and complexity of the information we can acquire about the past, and will allow our narratives to embrace rather than reject contradiction and difference. Thus rather than wiggle-matching between shifts in aDNA and broader changes in material culture, we would urge scholars to expect these different forms of data to rarely, if ever, align perfectly. These multiple forms of data show us different aspects of the past, and in particular their use to think about the inherently complex and multiple sphere of identity demands that we embrace intersectionality, complexity and diversity. Provocatively, we might suggest that scholars should be especially suspicious where shifts in material culture and society appear to neatly match changes in aDNA, because this may be revealing of the assumptions built into our models given the inevitable complexity of past societies. Processes of change are always messy and multiple, their effects are local and variable, and they eschew simplistic causation (Crellin Reference Crellin2020; Robb and Harris Reference Robb and Harris2013). In contrast to the tendency of some aDNA researchers to want to interpret the data in the most straightforward way possible, we would urge researchers to embrace the search for complexity over simplicity. As Villanea and Schraiber (Reference Villanea and Schraiber2019) demonstrate, multiple models can explain the same patterns of genetic ancestry. What would change in the past look like if we sought the least parsimonious explanation possible?

At the outset of this article we characterized aDNA papers as operating at either the large or the small scale, and have concentrated our argument on the former of these. Before concluding, however, we also want to think briefly about the theoretical frameworks that underpin small-scale studies as well. A number of papers dealing with individual sites and small-scale studies have emerged in the last ten years (e.g. Haak et al. Reference Haak, Brandt, Jong, Meyer, Ganslmeier, Heyd, Hawkesworth, Pike, Meller and Alt2008; Keller et al. Reference Keller, Rott, Hoke, Schwarzberg, Regner-Kamlah, Harbeck and Wahl2015; Knipper et al. Reference Knipper, Mittnik, Massy, Kociumaka, Kucukkalipci, Maus, Wittenborn, Metz, Staskiewicz, Krause and Stockhammer2017; Lee et al. Reference Lee, Rennebery, Harder, Krause-Kyora, Rinne, Müller and von Wurmb-Schwark2014; Le Roy et al. Reference Le Roy, Rivollat, Mendisco, Pemonge, Coutelier, Couture, Tillier, Rottier and Deguilloux2016; O’Sullivan et al. Reference O’Sullivan, Posth, Coia, Schuenemann, Price, Wahl, Pinhasi, Zink, Krause and Maixner2018; Scheib et al. Reference Scheib, Hui, D’Atanasio, Wilder Wohns, Inskip, Rose, Cessford, O’Connell, Robb, Evans, Patten and Kivisild2019). These offer detailed and nuanced engagements with archaeological evidence alongside the aDNA, and in some cases isotopic data as well (e.g. Haak et al. Reference Haak, Brandt, Jong, Meyer, Ganslmeier, Heyd, Hawkesworth, Pike, Meller and Alt2008; Keller et al. Reference Keller, Rott, Hoke, Schwarzberg, Regner-Kamlah, Harbeck and Wahl2015). These papers rest far less on assumptions about archaeological cultures and singular identities than do those explicitly operating at the large scale. The smaller data sets explored appear to leave space for the authors to consider more complex and potentially contradictory lines of evidence. Even where modernist concepts such as the nuclear family are invoked, these authors are careful not to make such a claim universal (Haak et al. Reference Haak, Brandt, Jong, Meyer, Ganslmeier, Heyd, Hawkesworth, Pike, Meller and Alt2008, 18229). In addition, rather than simply invoking migration, these studies explore more complex ideas of exogamy, patri- and matrilocality, and movement on a more local scale. Here we begin to open up access to the historical mechanisms that might produce grand narratives of migration.

Thus our paper could be read as an argument for sticking with these small-scale analyses alone, and a number of commentators have noted how much more easily these approaches fit with standard archaeological interpretations (e.g. Vander Linden Reference Vander Linden2016, 718). This, however, is not our position. As John Robb and Tim Pauketat (Reference Robb, Pauketat, Robb and Pauketat2013) have elegantly argued, it is not the case that we can simply avoid the grand narrative and the politics that surround it. Robb and Pauketat (ibid., 33) argue that we avoid the large scale at our own peril because if we, as archaeologists, do not engage with it, others will – others whom we might see as less informed and less engaged with the data and who, most likely, might have their own political aims. In a world where migration is never far from the newspaper headlines it is not the time to shy away from engaging with migration in the past (Frieman and Hofmann Reference Frieman and Hofmann2019). We suggest that we need to engage with the large scale and the grand narrative but do so in a way that integrates different scales of analysis. It is not the case that there is a correct scale at which to work; rather what we have to do is to combine both the local story and the grand narrative, alongside other scales of analysis.

Here again, assemblage thinking is helpful. Assemblages operate at multiple scales from the atom to the institution (Harris Reference Harris2017; see also Crellin Reference Crellin2017). As Lesley McFadyen (Reference McFadyen, David and Thomas2008, 307) argues, it is not the case that scales of analysis nest together like Russian dolls where either a top-down or bottom-up approach will work. Looking at neither the small scale nor the large scale provides the key. Rather, scale is more complex; DNA sequences could be seen as small-scale but the way in which they are shaped by processes that stretch across time and space means that they traverse scales of analysis in complex ways. Our aDNA interpretations need to do the same; they need to fold together multiple scales of analysis. If we want to talk about the ‘big picture’ or the ‘grand narrative’, that does not mean we should be avoiding the small scale, rather we should be folding that small scale into our analysis. It also means that simply accumulating more and more data to create an ever-larger scale of analysis is not the answer either.

As well as considering the narratives that emerge from aDNA research, we also call for a consideration of the process itself. Ethnographies of archaeology have revealed to us the complex ways in which we move from trowel’s edge to interpretation (Edgeworth Reference Edgeworth2012). Knowledge is produced in the small steps we take as we sample individual bones, as we extract their DNA, analyse the results, categorize them, bring them together with other analyses and piece this together in a narrative. It is clear from the work of both Lucas (Reference Lucas2012) and Fowler (Reference Fowler2013) that archaeologists themselves play key roles in the process of interpretation. As Bruno Latour (Reference Latour1999) has shown, these small steps matter; each act of ‘translation’ (in Latour’s terms) at once both amplifies and reduces some aspects of that which we study. We need to look carefully at these small steps and we need to think about which aspects of the narrative we are amplifying and which we are reducing in the process. It matters how we move from the sequence of one individual to a narrative about the past. Science and technology studies have been effective at revealing how scientific knowledge is produced and we see significant scope for ethnographers in aDNA labs to open up a key reflexive space for consideration.