The Antiquity keynote lecture at TAG-on-Sea 2013 was held at the Talbot Campus, Bournemouth University, on Monday 16 Dec 2013.
For many archaeologists, the phrase ‘archaeology and science’ means the application of modern scientific knowledge and instruments to artefacts and deposits. In this construal, archaeology stands in a dependent relationship to other sciences. It is argued here that other sciences—and the history of science—also depend on the products of archaeology. This paper illustrates the many inter-relationships among archaeology and other sciences through brief examples and a case study. Three sets of relationships are of special interest here: 1) the substantive and conceptual contributions that archaeology makes to the physical, biological and social sciences; 2) archaeological research on the remains of scientific activities, which helps to illuminate the materiality of early modern and modern science; and 3) investigations of both the processes of science and its knowledge products. Illustrating the third set of relationships, the case study shows how archaeologists can fashion models of discovery processes that cross-cut different people, investigators, time periods, social groups and polities by generalising about the people-artefact interactions that lead to discovery claims.
For many archaeologists, the phrase ‘archaeology and science’ means applying modern scientific knowledge and instruments to artefacts and deposits. That is, ‘archaeology and science’ is equated with archaeometry, a one-way relationship that emphasises archaeological dependence on other disciplines. Unfortunately, the archaeometry meaning excludes significant two-way relationships between archaeology and other sciences. Three kinds of relationships are addressed here: 1) the substantive and conceptual contributions that archaeology makes to the physical, biological and social sciences; 2) research on the remains of scientific activities, whether ancient or modern; and 3) fashioning new generalisations, relevant to science studies, about the processes and knowledge products of science. Relationships 2 and 3 comprise what I call ‘the archaeology of science’ (Schiffer 2013). This paper illustrates these relationships with brief examples and a case study on discovery processes.
The mutual dependence of archaeology and the earth sciences has a long and storied history, especially in Palaeolithic and Palaeo-Indian studies. In some cases stratification of cultural and environmental deposits furnishes important information to geoscientists. A case in point is the occurrence of independently dated late Palaeo-Indian artefacts on river terraces in the Lower Mississippi Valley, which contributed to geographer Roger Saucier’s extensive revision of the alluvial chronology. As Saucier (1974: 1) noted, his work “relies heavily upon observations and conclusions resulting from archeological investigations”. In research on landscape evolution during the late Holocene, archaeological deposits furnish crucial evidence, especially in the absence of historical documents (e.g. Denevan 1992).
As for palaeomagnetism, measurements on fired-clay features have documented high-frequency movements of the magnetic north pole for time periods and regions lacking historical information (Wolfman 1984: 390). Archaeological evidence also profoundly enhanced radiocarbon dating. As you may recall, it was once assumed that the atmospheric ratio of 14C to 12C remained constant over time (Libby 1955: 8). However, radiocarbon dating of known-age archaeological specimens undermined this assumption and led eventually to the development of tree-ring-based calibration curves (Taylor 1987: 19–21). And, lest we forget, modern tree-ring dating began as a collaboration between astronomer A.E. Douglass and archaeologists of the American Southwest who supplied wood specimens from historic and prehistoric ruins (Nash 1999).
Chemistry and materials science have also benefited from analyses of archaeological specimens. A recent example is Maya Blue, a handsome paint found on murals, pottery and other objects in Maya ceremonial contexts. The blue pigment is an unusual material new to science: molecules of indigo, an organic substance, are so firmly bound to the crystal lattice of the clay mineral (palygorskite) that they resist the attacks of strong acids and bases. Chemists have labeled Maya Blue a “nanostructured polyfunctional hybrid organic-inorganic material” and called for a new chemical paradigm (Doménech et al. 2009: 2371). I note that archaeological ceramicist Anna O. Shepard (1962) first proposed that the pigment in Maya Blue was an organic substance.
There is also an impressive roster of archaeological contributions to the biological sciences, ranging from biogeography to genetics. As we know, the archaeological record uniquely furnishes evidence on the beginnings of plant and animal domestication throughout the world. In turn, interactions of people with plants and animals have provided new insights into processes of co-evolution (e.g. Rindos 1984), which contribute to archaeology and biology. Archaeologists also assess the effects on native animal populations of human predation, habitat losses and competition from domesticates and other invasive species—issues that are of interest to ecologists and others (e.g. Lyman 2010, 2011a). A well-known example is documentation of the mass extinction of birds in Polynesia (e.g. Steadman 1995), including the impressive moa of New Zealand, which followed prehistoric colonisation. Floral and faunal evidence also enables palaeoecological and biogeographical reconstructions (e.g. Lyman 2011b). Thanks to human remains unearthed by historical archaeologists, genomes of ancient pathogens, such as the bacteria that cause leprosy and tuberculosis, are being sequenced, leading to reconstructions of their evolutionary histories (e.g. Gibbons 2013), possibly paving the way to improved therapies. And, lest we forget, archaeologists in recent decades have made many contributions to vertebrate taphonomy (e.g. Lyman 1994) and forensics (e.g. Hunter & Cox 2005).
Although familiar with many of these examples, we may fail to grasp the larger pattern: there is a two-way flow of matter and information between archaeology and other sciences. Through archaeometry we obtain useful tools from the physical and biological sciences; they, in turn, receive archaeological specimens, analyses, insights, concepts and inferences that further their research agendas.
As for the social sciences, we lament not generating the kinds of theories that processualists once promised. Nonetheless, the so-called ‘material culture turn’ in the social sciences and humanities is benefiting greatly from archaeological publications (e.g. Hodder 1982; see also Knappet & Malafouris 2008; Hicks & Beaudry 2010; Olsen 2010; Olsen et al. 2012; Graves-Brown et al. 2013) as well as from the works of turncoat archaeologists who pass as cultural anthropologists (e.g. Miller 1998). Likewise, archaeological theories of technological change (e.g. Schiffer 2011) and communication (Schiffer & Miller 1999) have begun to inform discussions in several social science disciplines. In addition, every prehistory textbook contains examples of societies that lack modern counterparts, and these societies have to be considered when one builds theories of societal evolution. And, of course, the humanities owe much to archaeology, beginning with what Glyn Daniel (1964 ) called the ‘idea of prehistory’, which today infuses the worldviews of educated citizens.
Let us now turn to ‘the archaeology of science’, which is defined as “archaeological research into the processes and products of science, which includes the scientific activities of any person, organization, or society as well as the comparative study of such activities” (Schiffer 2013: 13). What makes the archaeology of science ‘archaeological’ is the framing of research questions in relation to people-artefact interactions at any scale of activity. Such research is carried out in prehistoric, historic and modern societies.
As a foundation for surveying the broad sweep of the archaeology of science, it is necessary to define science behaviourally. Although many scholars define science as knowledge-producing activities that make use of the ‘scientific method’, this definition is problematic. If the history and philosophy of science have taught us anything, it is that there is no single scientific method. As processes for creating potentially useful knowledge, science consists of varied activities and of activity sequences. There can be no single scientific method because people create many kinds of knowledge, in many ways, for many uses, in diverse societal contexts. “A person may begin with an observation, puzzle, anomaly, problem, hypothesis, question, theory, dream, model, analogy, or new artifact, and can reach an outcome inductively, deductively, abductively, or even by nonlogical means” (Schiffer 2013: 3–4). That is all that needs to be said about the processes of science until the last section of this paper, which examines one kind of discovery process that does operate in a more or less logical manner.
Clearly, a behavioural definition of science prescribes no particular scientific method. Rather, it focuses on the products of scientific activities, is applicable to all societies, and helps us to understand why science is a human universal. Although scientific activities often create new artefacts, the product of interest here is useful knowledge, which I define as follows: “scientific knowledge makes possible, though it does not guarantee in every instance, effective human interactions with artifacts and with living and nonliving phenomena of the natural environment” (Schiffer 2013: 4, emphasis in original). All groups interact with material phenomena, and so must create the knowledge that enables their members to predict the outcome of performances and interactions. Thus, scientific knowledge is found in all modern and ancient societies. The major kinds of knowledge entailed by this definition are descriptions, categories and classifications, empirical generalisations, experimental laws, theories, models and recipes. These categories are familiar to science scholars—with the possible exception of recipes (Schiffer 2013: 32–35). Because recipes are a major product of the archaeology of science, they deserve some attention here.
Recipes describe complex behavioural processes, those having a high level of organisation and an extended engagement with the material world; everything from making catfish stew to building a nuclear-powered submarine. Richard Krause (1985: 29–31), in an ethnoarchaeological study of several African potters, first called attention to the importance of recipes for modelling manufacture processes, but recipes also apply to other complex activity sequences and render their outcomes somewhat predictable.
A recipe consists of three major components (adapted from Schiffer & Skibo 1987): 1) interactors, which is a list of the quantities and relevant properties and performance characteristics of all people, animals, materials, tools and facilities, and environmental phenomena; 2) interactions, which specifies the matter-energy transactions among interactors, organised in sequence(s), including alternative sequences that are equally effective or that may be needed in response to changed circumstances; and 3) a statement of the expected outcome, which is an emergent empirical phenomenon.
Thus, a recipe describes sequences of interactions and activities that, when carried out competently, yield an outcome such as an artefact’s procurement, manufacture, maintenance, reuse or cultural deposition. Individual empirical generalisations or experimental laws cannot describe these complex processes; only a recipe’s intricate conceptual structure can do the job. That is why recipes are a fundamental and universal kind of scientific knowledge.
Why do recipes usually yield the expected product? The answer is that beneath the sequence(s) of visible interactions and activities lies an invisible realm consisting of empirical generalisations and experimental laws. Thus, a recipe’s interactions are in accord with—indeed, depend upon the validity of—other nuggets of scientific knowledge. Precisely these underlying generalisations enable a recipe, when put into practice skilfully, to create something entirely new to human experience—and new to the universe.
Recipes, and recipes alone, permit people to create—and researchers to explain—outcomes. An important implication is that, by knowing only the interactors, we would be unable to anticipate or explain the outcome, for the latter is determined by the recipe’s interaction sequences.
The main function of scientific knowledge, as defined behaviourally, is to empower people through prediction to engage competently with the material world. Recipes meet this criterion especially well because they make possible the forward motion of behavioural chains, from processes of raw material procurement to cultural deposition.
I should emphasise that the archaeology of science is not new, for we have long modelled the scientific knowledge implicit in traditional technologies, particularly recipes but also empirical generalisations and experimental laws. In doing so, we employ three major research strategies, singly and in combination: experimental archaeology, ethnoarchaeology and archaeometry.
Through replication experiments, which usually involve much trial and error, we learn to fashion a recipe that resulted in a specific product, such as a Folsom point or Puebloan seed jar. In addition, we may design an experiment that yields new experimental laws, such as those describing how surface treatments of traditional cooking pots influence their heating effectiveness and other performance characteristics (e.g. Schiffer 1990). Experimental archaeologists also build theories, an example being our provisional theory of ceramic abrasion (Schiffer & Skibo 1989). The findings of experimental archaeology are codified, for example, in dozens of books on traditional technologies (e.g. Rye 1981; Whittaker 1994; Adams 2002).
Ethnoarchaeologists observe sequences of activities that can be represented as recipes. They record categories and classifications, and may also elicit the artisan’s theories and models that explain successes and failures in achieving the expected outcome. An ethnoarchaeologist may also construct empirical generalisations and do comparative studies that lead to experimental laws, such as the relationships between ceramic form and techno-function (e.g. Henrickson & McDonald 1983; Skibo 2013).
Finally, archaeometry enables the archaeologist and interdisciplinary collaborators to model recipes for manufacture, use and other behavioural processes. In addition, the archaeometrist often furnishes a deeper understanding of these processes by drawing on experimental laws, theories and models from other sciences (e.g. Malainey 2011). My favourite examples come from residue analyses of ceramic containers, which furnish the foundation for reconstructing ancient beer and wine recipes.
In addition to the three traditional strategies for modelling the knowledge-content of science, we employ archaeological survey and excavation for inferring past scientific activities, especially in early modern and modern societies, as indicated by the following brief examples. Archaeologists have recorded the traces of once-secret activities such as nuclear testing in the Nevada desert (Beck 2001, 2002), trials of German V-2 rockets in New Mexico (Eidenbach et al. 1996), and manufacture of British military explosives (Cocroft 2000). A few researchers have excavated in places where well-known scientific figures, including Isaac Newton (Spargo 2005) and Thomas Edison (Gall et al. 2007), created archaeological deposits. Others have investigated the remains of scientific activities in Antarctica (Harrowfield 2004) and in the sixteenth-century Roanoke Colony in present-day North Carolina (Noël Hume 1994). This is but a tiny sample of the exciting directions being taken by historical and industrial archaeologists.
For several reasons, I believe that archaeologists can also generalise about the processes and products of science and thus contribute to science studies. First, science scholars, such as historians, philosophers and sociologists, study a limited number of complex, literate societies, and generalise accordingly. Because their sample of societies, and thus the varieties of science they study, are biased, we may suspect that their generalisations are also biased. Archaeologists, on the other hand, may draw evidence on scientific practices and products from any society, ancient or modern, literate or non-literate, simple or complex. Potentially, then, our generalisations could be more general than those created in other disciplines. Second, unlike the newcomers to material culture studies in other disciplines, we possess a large and sophisticated conceptual toolkit for dealing with the materiality of human life. Because scientific activities, like virtually all others, involve artefacts, we can craft artefact-based generalisations that furnish a different and potentially illuminating perspective on science.
I now turn to a case study that treats discovery processes, which have not been satisfactorily studied in any discipline. The aim is to fashion artefact-based models that cross-cut investigators, time periods, social groups and polities by generalising about the people-artefact interactions that lead to discovery claims.
The archaeology of science presents several behavioural models of discovery processes, including discovery by accident and discovery by technology transfer (Schiffer 2013: 185–98), but here I discuss only the ‘discovery machine’ model. A discovery machine is an apparatus or system whose components are capable of undergoing many substitutions, one at a time, each of which may generate a new effect that might be claimed as a discovery. Let us envision a simple apparatus of two material components interacting only with each other and the investigator. We assume that during the apparatus’ early uses, the investigator observes an emergent effect and claims it as a discovery. To transform the apparatus into a discovery machine, the investigator makes a series of substitutions for one of the components or undertakes a series of new interactions. In this manner, the ever-changing apparatus, still possessing the same basic structure, yields one new effect after another, which may also be claimed as discoveries. Thus, a discovery machine—whether simple or complex—generates a discovery cascade.
One implication of this model is that, after a new discovery machine is reported, other investigators—perceiving a potential for further discoveries— may acquire one, make substitutions and conduct new experiments. A further implication is that some investigators may make the same, somewhat obvious, substitutions and arrive independently at the same effect—all of whom may claim the identical discovery. Thus, Volta’s electric pistol, a glass or metal reaction chamber for combusting gases, rapidly became a discovery machine in pneumatic chemistry as investigators sent sparks through different gases. In several cases, experimenters arrived independently at the discovery that igniting hydrogen in oxygen creates water (Schiffer et al. 2003).
The key to understanding a discovery machine’s development is in appreciating the meaning that investigators assign to an apparatus’ early successes. These discoveries are taken as an auspicious sign that other significant effects are in the offing, merely awaiting fruitful substitutions. Indeed, in a satisfying and sometimes dramatic fashion, a discovery machine raises the probability of making specific discoveries.
Two of the most well-known and productive discovery machines in early modern science are the optical microscope and the refracting telescope, introduced respectively by Robert Hooke and Galileo. These discovery machines have enjoyed centuries of use. Another significant but less well-known discovery machine is the electrolytic cell. Introduced in the early nineteenth century, its basic structure is simple: a glass or ceramic container, two electrodes, a conductive solution called an electrolyte, and a source of low-voltage, high-current DC electricity such as a hefty battery connected to the electrodes. By placing a metal compound and perhaps other chemicals in an electrolytic cell, the investigator could liberate the metal and deposit it on one electrode. The earliest use of this discovery machine was in plating copper from an electrolyte of copper sulphate and water. In turning the electrolytic cell into a discovery machine, investigators substituted compounds of different metals and electrolytes of differing chemical composition. The electrolytic cell was a very fertile discovery machine that, by the early 1840s, had given rise to electroplating, the first electrical industry (Schiffer 2008).
Discovery machines exhibit varying patterns of development. For the microscope and telescope, the developmental distance to create the first apparatus was considerable. Hooke and Galileo had to grind novel lenses and build custom devices to hold them; Hooke also crafted a light source and a stage to hold the specimen. But for later discoveries, the developmental distance was usually trivial, such as merely viewing a new specimen or searching a new portion of the sky. Not so the electrolytic cell, whose initial development for plating copper merely required the assembly of inexpensive, off-the-shelf materials. However, plating some metals, such as gold, necessitated chemistry expertise and many trials with gold compounds and different electrolytes (Schiffer 2008).
In view of these varying developmental patterns, one may ask of any discovery produced by a discovery machine: has scientific credit been fairly apportioned between the inventor and later user(s)? This question may lead to studies that serve as a counterweight to the social and biographical perspectives customarily employed to explain specific discoveries.
A starting point for the artefact-based approach to discovery machines is to identify an apparatus and learn whether its uses led to a cascade of discovery claims. If so, then it can be regarded as a discovery machine. Further research might make for an interesting archaeological project that focused on its development and later uses. Apparatus that might be suitable for case studies include Geissler tubes, X-ray tubes, spectrometers, lasers and gene sequencers. In earlier societies, one might consider fire, cooking pots, pipes for smoking and distillation apparatuses as possible discovery machines.
I contend that there are many relationships between archaeology and science. Archaeometry describes only the one-way flow of information and instruments from other sciences into archaeology. But many relationships are actually two-way and long-standing. Indeed, archaeological research has for much of the past century provided evidence and inferences useful to sciences across the disciplinary spectrum, from earth sciences to biology.
Beyond supplying examples of the latter relationships, this paper identified a large research realm called ‘the archaeology of science’. A traditional part of this diverse realm is the study of scientific activities and knowledge in any time or place through experiments, ethnoarchaeology and archaeometry. Such research in many prehistoric and historic societies has led to the modelling of recipes for behavioural processes such as artefact manufacture and use. Archaeologists also employ survey and excavation to examine the material remains of early modern and modern science, which has, for example, illuminated the materiality of once-secret activities.
We may also apply an artefact-based perspective to widespread scientific processes such as discovery. The aim is to create generalisations—a science of ‘sciencing’, to use Leslie White’s (1949: 3) neologism—that may contribute to the resolution of long-standing problems in science studies. To illustrate this potential, I presented the ‘discovery machine’ model, which helps to explain the productivity of certain lines of research, accounts for some instances of independent discoveries and raises questions about how credit for specific discoveries is allocated.
The notion of the archaeology of science, then, calls attention to both ongoing research as well as future opportunities. As we undertake additional studies, perhaps our artefact-informed inferences and generalisations can contribute to discussions of science taking place across the academy and among shapers of public policy.