The Neglected Networks of Material Agency: Artefacts, Pictures and Texts

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Synopsis

I argue, following Actor-Network-Theory (ANT), that agency is a process distributed across collectives of humans and nonhumans. These collectives can be considered in terms of networks, composed of heterogeneous nodes and links. Yet despite its name, Actor-Network-Theory has paid relatively little attention to the spatial and organisational structures of these human-nonhuman networks and their effects upon network ‘behaviour’ or dynamics. I draw upon some new network concepts in an attempt to fill this gap, and demonstrate my approach using an archaeological case study, one that explores the differential role of artefact, picture and text in actor networks. One reason for choosing such a case study is that archaeological approaches to agency remain anthropocentric, despite the material basis of the discipline, and have not as yet made much systematic use of ANT. Not only can archaeology benefit from ANT in tackling agency (particularly when supplemented with network concepts), but it can contribute to wider debates on agency thanks to its material basis.

Trimarans and Guns

In 2004–5 Ellen Macarthur stunned the world by sailing round-the-world at the age of only 28. This was a solo voyage completed in a mere 71 days. Phrased in this way, the virtuoso yachtswoman is clearly the ‘agent’ in control of this endeavour, with the yacht as her tool, responding to her skilful choices and actions. However, the 75 foot B&Q trimaran is no straightforward tool, but an ultra hi-tech piece of seafaring equipment; so much so that it can sometimes give the impression of sailing itself, with Macarthur as little more than a privileged passenger. From this perspective, perhaps the trimaran is the agent, doing all the sailing. And it is very easy to fall into a debate over what or who the real agent is: yacht or yachtswoman.

A more familiar and polarised debate of this kind is that between the pro-gun and anti-gun lobbies. While the anti-gun lobby argues that guns kill people, the pro-gun lobby maintains that people kill people, with the gun as nothing more than a neutral tool. The dualism here is between materialist and sociological explanations – the former portrays the gun as the responsible agent, the latter puts all responsibility in the hands of human agents.

Is there any way out of these dualistic debates? Is there a solution which admits that both human and gun may to some extent act, or that both Macarthur and the trimaran played a part in the round-the-world project? One solution that seems to work at first is to say that, yes, tools of this kind can have agency of a kind, a special kind of agency that is secondary. A gun has agency because human designers have intentionally delegated it with agency in their absence. Gell (1998) has argued as much for landmines and various other kinds of artefact, in particular artworks. It is agency by association; the argument is that agency has to emanate from humans. However, we should note that in this kind of secondary agency, in which primary human agency is delegated or transposed onto materials, the primary agency or ‘authorship’ may be obscured; and this property may come to be deliberately exploited.

There is, though, another answer. If we go back to the yachting example, it is interesting that Macarthur invariably speaks not of ‘I’ but of ‘we’. The impression is not one of a virtuoso yachtswoman acting with considerable help from her hi-tech tools, but something much more like an equal partnership between woman and machine. Perhaps we can understand what is going on in this case with a little help from Latour, who comments not on this directly but on the question of agency with the man-gun example (Latour 1996). His approach is what is often termed ‘symmetrical’ – one cannot assume primacy for either man or gun. Rather, the two are mutually constituted, each being transformed by the other in their conjunction. Trying to decide whether one or the other is the agent makes little sense from this perspective. Think not of agents as entities, but of agency as a process. More specifically, agency is a process unfolding in given situations or activity frameworks and for this one can turn to the work of Kirsh (1995) and Goodwin (1994) in distributed cognition.

However, with the cases mentioned above, the situation or activity framework is relatively straightforward: there is a clearly definable task at hand, which is to be achieved by one human in conjunction with one tool. This is, in fact, often the kind of scenario selected for analysis by those working in distributed cognition: Kirsh and Maglio’s Tetris, for example (Kirsh and Maglio 1994). The single artefact can, though, be deceptive. In the case of the Macarthur trimaran project, a significant web of human and technological support lies behind it, with ‘Team Ellen’ (‘Offshore Challenges Sailing Team’) being 35 strong and covering communications technology, boat design and performance and even marketing and sponsorship aspects. The agency in this project cannot be confined to the situations in which Ellen and her trimaran find themselves while at sea; it spills out across these widely distributed networks.

Actor-Network-Theory

It is this kind of phenomenon that is the meat and drink of Actor-Network-Theory,¹ of which Latour is one of the principal protagonists (e.g., Callon 1986; Law 1992; Latour 2005). Actor-Network-Theory, or hereon ANT, was devised as an approach to social phenomena that decentred the human subject, seeking to overcome the assumed ontological primacy of humans by adopting an analytical impartiality (Ashmore et al. 1994, 735). This impartiality allowed the focus to fall on objects (‘nonhumans’) as well as people in social collectives, rather than prejudging what should or should not belong within what we label society (Latour 2005, 72). ANT also shifted the focus onto relations – the ‘semiotic’ connections between the diverse elements in socio-technical ensembles. In science and technology studies, there are now many examples of how ANT has encouraged a focus on the spiralling networks of connection that seem to spread out from what appeared to have been singular, bounded technologies (see numerous papers on ANT resource, hosted by Centre for Science Studies at University of Lancaster and managed by John Law).

If we were to consider briefly the Macarthur-trimaran scenario from an ANT perspective, then the focus would fall very much on both the conjoined human-nonhuman character of the project and the multiple connections that hold it together across many scales. These connections in the Macarthur-trimaran actor network might be considered as semiotic, or syntactical; or, as Law (2000) puts it, in relation to his work on 15th–16th century Portuguese vessels:

Hull, spars, sails, stays, stores, rudder, crew, water, winds, all of these entities (and many others) have to be held in place, so to speak functionally, if we are to be able to point to an object and call it a ship

Latour has described such objects as ‘immutable mobiles’ (Latour 1990; Law 2000). In other words, the semiotic / functional network has to remain immutable in order for them to move around in physical space. Law has used the example of Portuguese vessels to think through the existence of objects in different spaces, which he calls Euclidean and network spaces, although one might also term them physical and relational spaces. He argues that “objects are always performed in a multi-topological manner, and are dependent for their constancy on intersections between different topoi” (Law 2000, 7).

Yet such explicit considerations of the topologies of actor networks are surprisingly rare in ANT (see also Murdoch 1998; Latour 2005, 128–31). Despite the important advances it has kickstarted, I would argue that ANT does not go far enough in its exploration of network structures at either the micro- or macro-levels and the potential impact of these structures on network ‘behaviours’. Furthermore, the different kinds of actors or actants, especially when ‘nonhuman’, have not been detailed; what are the roles played in actor networks by, for example, artefacts, texts and pictures? I now deal with these two categories in turn – networks and actors.

Neglected Networks

Although the term ‘network’ is integral to ‘Actor-Network-Theory’, it has been surprisingly overlooked, at both micro- and macro-levels.² If humans and nonhumans are assembled together in complex collectives, then presumably these must have some kind of network structure, albeit shifting and dynamic. Although by no means numerous, some scholars working with ANT have highlighted the multiple topologies of the social; Mol and Law (1994), for example, focused attention in particular on 'fluid space', a move paralleled by Lee and Brown (1994) in their call for a consideration of ‘smooth space’ or ‘rhizomatic networks’, the latter drawing upon Deleuze and Guattari (1988). This is presumably in response to a perceived bias in ANT towards what Lee and Brown (1994) call ‘striated space’ or ‘arborial networks’. As much as anything else, these efforts may be aimed against the idea that social collectives are hierarchically structured. However, there appears to be an implicit assumption that complex structures must be either chaotic or commanded, heterarchical or hierarchical. This polarity echoes that encountered in the natural sciences between regular networks on the one hand, of the character of a crystal lattice, for example, and random networks on the other (e.g., gases).

Yet social networks are rarely either regular or random, falling instead somewhere in between these two extremes. Such networks can have complex structures and dynamics, requiring particular methodologies for their study. While sociologists have long suspected that social networks have such characteristics, it is only recently that the mathematics of such networks has caught up. The main instigators of this new flurry of activity in complex networks are Duncan Watts and Steve Strogatz, who have published a series of influential articles in Nature and Science, spawning a whole field of network science across physics, biology, economics and sociology, among other disciplines (Barabási 2002; Watts 2003; Watts 2004; Evans 2005; Newman et al. 2006; Lane et al. in press).

We should perhaps not be overly surprised that this field has not intersected with ANT. But social network analysis using more established mathematical techniques has been around in sociology for a long time (Wasserman and Faust 1994; Scott 2000; Carrington et al. 2005). It may be that ANT has not intersected with this branch of sociology because the actors in these networks are invariably human, with little room for the nonhuman. Whatever the reasons, surely an approach that takes all the pros of ANT and integrates it with a stronger ‘network’ perspective, one that can tackle structure and topology more systematically, might give us a better way forward for getting to grips with the ways in which agency is exercised /distributed in ‘collectives’ of humans and nonhumans.

It is not only at the macro-level that ANT is relatively weak with regard to network analysis. At the micro-level too, there is much more that might be done to characterise the variable connections that bring humans and nonhumans together in collectives. Latour (2005, 72) says that things can authorise, allow, afford, encourage, permit, suggest, etc. (and note the link Latour makes with Gibsonian ecological psychology here); and this is perhaps getting towards these micro-connections. But why not try to systematise these relationships more tightly and look at the kinds of connections that might occur – such as directionality, frequency, fidelity and distance (Knappett in press)? And as ANT has claimed to have a semiotic dimension in its attention to connections, why not even look at the semiotic links between entities? However, ANT may have been hamstrung in this by its adherence to a Saussurean semiotics (Watts, this volume);³ whereas using a Peircean ‘semeiotic’ offers far more potential when it comes to understanding significative relationships in material culture (Knappett 2005; Preucel 2006; Watts, this volume). If parameters such as resemblance, contiguity, factorality, causality and convention could be used to analyse network properties, then perhaps we might be able to say something new about network ‘behaviour’ in terms of how agency is distributed across different ‘nodes’, human and nonhuman. However, I do not intend to pursue these points on network structure and behaviour much further in this chapter, as I explore these elsewhere; my principal concern here is with the variable character of material ‘actors’.

Material Actors: Objects and Things

When it comes to the ‘actors’ in ANT, we have perhaps seen more on their differentiation, not just into humans and nonhumans, but actors and actants. Yet arguably ANT has not seen a concerted effort to examine the qualities of different kinds of artefact. One straightforward yet very useful distinction that we might introduce is between ‘objects’ and ‘things’. These two terms are usually employed interchangeably in sociology, anthropology and archaeology, with little thought to their potential differences. In an attempt to underline the relationality of material meaning (and thus very much in line with the aims of ANT), Bill Brown has, in the context of literary criticism and cultural theory, proposed such a distinction (Brown 2003; see also Mitchell 2005, Schwenger 2006). Things, he argues, are ambiguous, undefined. They have a metaphysical presence. It may be difficult to define ‘a thing’ – it may be un-nameable – for example, ‘pass me that green thing over there’ – the green thing is unintelligible in some way. As for objects, they materialise out of the amorphousness of things. Objects are named, understood and transparent. Object and thing thus clearly have a relational component in their definition – what is an object to one beholder might very well be a thing to another. Objecthood and thingness are relational registers.

Interestingly, a distinction between object and thing as two kinds of register has also been drawn recently by Gosden (2004). He defines objects as items that are alienable, quantifiable and disembedded from social relations. Things, on the other hand, are inalienable, possessing unquantifiable qualities and are embedded in social relations. Things exist in assemblages, in artefact communities, from which they are difficult to extract without losing much of their meaning. Things cannot be singled out, objects can.⁴ There do appear to be significant overlaps between the formulations of Gosden and of Brown, and these can probably be attributed to a shared genealogy that can be traced back to Heidegger (see Harman 2005, 2007, on Heidegger’s conception of object and thing).

As an aside, we should be aware of the dangers in defining objects or things in purely relational terms. This kind of approach is termed by Brown (2001, 7) “a new materialism that takes objects for granted”, and is essentially what arises out of Appadurai’s The Social Life of Things, and more fully developed in the wide-ranging and influential work of Danny Miller and colleagues on the consumption of material culture (Appadurai 1986; Miller 1987; 2005). That is to say, the emphasis is very firmly on the subject-object relation, with little attention afforded to the materiality of the object/thing itself (Watts 2007). This is reminiscent of the Latour – Lemonnier debate over the man-with-gun (i.e., Latour looks principally at relations, Lemonnier demands that the gun-ness of the gun receives attention too). The way in which ‘materiality’ approaches within material culture studies have overlooked the material properties of things/objects is also the subject of a powerful critique by Ingold (Ingold 2007).

Transformations: Artefact/ Image/ Text

How do things become objects or vice versa? Well, the process may not involve image and text at all but may occur when ‘smooth coping’ is disrupted. Here we may bring in Heidegger, as discussed by Wheeler (2005). Normally speaking, we encounter equipment as things – a hammer, for example, is ‘ready-to-hand’, part of a total actor network. If it breaks, however, the actor network is disrupted and the ‘immutability’ challenged (as it would be if Macarthur’s trimaran keel snapped). At this point, the hammer (or keel) is no longer ready-to-hand but is ‘present-at-hand’ (Wheeler 2005, 128–144). This is when its material properties may become more transparent, consciously recognised and named. It has, in other words, shifted from thing to object; but if quickly fixed can return seamlessly to thingness.

But what I particularly want to look at here are the ways in which images and texts might alter the status of artefacts. Gosden talks of the process of display in museums as one means of singling out or objectifying artefacts (Gosden 2004). A similar kind of process is discussed by Mitchell, who observes the process whereby found objects (or more properly, things) are turned into artworks. He describes this as a process of making an image of the object/thing; one example given is Jeff Koons’ work New Hoover Deluxe Shampoo Polishers (Mitchell 2005, Fig. 33). The ordinary thing is transfigured, yet its ordinary status is never quite forgotten; one could argue it continues to haunt the image. This could describe what happens in museum display too – an image is made of the thing through display and in the process becomes objectified. But the artefact’s thingness never quite goes away.

If an artefact can be transformed from thing into object through imagining/ imaging, then might words have the same effect? Peter Schwenger certainly argues as much, pointing to the ways in which words can make objects of (artefactual) things. The act of naming something with a word makes it into an object and nullifies its ‘thingness’ – what Schwenger calls, rather dramatically, ‘the murder of the thing’ (Schwenger 2001; 2006). Naming a bowl a bowl or a dog a dog establishes a lordship over it and denies its thingness. Thingness is somehow beyond representation and is thus unavoidably transformed in the act of objectification. The word denies thingness in much the same way that the image does. In both cases, a process of categorisation means that we close ourselves off to things – cognition overrules the senses (cf. Brown 2003).

Schwenger is careful to point out, however, that (artefactual?) things are transformed in this process rather than annihilated. Murder may bring an end to the physical thingness but not necessarily the metaphysical. The spectre of the thing may live on:

If there is a murder of the thing by the word, then, this does not definitively annihilate that thing; it only transposes it to the scene of an interminable haunting of language. (Schwenger 2001, 113)

The original thing, unrepresented, is always there trailing the representation like a shadow, whether that representation be an image or a word. But what is of added interest is that both image and word do very often have a material, artefactual existence themselves, in picture and text respectively. But when the word is made text, does it continue to exist solely as an object, or might it also ‘lapse’ back into thingness? In that, objecthood and thingness are registers which might equally ‘afflict’ artefact, picture and text; then surely a textual artefact can become a thing. Indeed, Bill Brown argues that the text is striving to become a certain kind of thing, rather than a representation of things.

There is, potentially, a fascinating temporality at work here – whereby an (artefactual) object can, through display, naming, or imagining /imaging, be brought out of its latent thingness, but that once there it requires the play of a set of forces to keep it suspended in objecthood and prevent it from being drawn back into the soup of thingness. Furthermore, the naming or imaging of the thing as object may itself have an artefactual dimension, with image becoming picture, much as word becomes text. And then picture and text do not remain as objectifiers, but themselves are brought into thingness. And then those doing the naming may themselves be named: a recursive, reiterative process.

This approach is very much consistent with the aims of ANT, in its twin focus on associations and on materiality. By looking in a little more detail at some of the different registers materiality can take – that is, objecthood and thingness – I hope to develop a deeper understanding of the overall character of actor networks as human-nonhuman collectives. In particular, I focus on the transformations that can occur in these networks as artefacts are imagised or verbalised, and as images and words are in turn artefactualised as pictures and texts respectively. These changes are almost incessant in social collectives of various kinds and they alter the texture of the network and hence not only the overall dynamics, but also the particular agency of individual objects and things, as these are relationally derived.

This interdisciplinary move is a difficult one, as it takes us into visual culture and literary theory; but this foray is worthwhile for what it can provide with regard to ‘material agency’. And fortunately there are scholars in these areas now exploring the relationship between texts and artefacts (e.g., Brown 2003) and images and artefacts (e.g., Mitchell 2005; also Gell 1998; Renfrew 2003; Renfrew et al. 2004; Gosden 2004). What I would like to do next is introduce an archaeological case study in which the interactions between these three categories – artefact, image and text – can be traced over space and time. These three – let us say material, visual and textual culture – are rarely considered together within a single methodological framework, as indeed is illustrated in our case study.

Artefact, Picture and Text in the Aegean Bronze Age

A deep-rooted separation of material, image and word – of artefact, picture and text – is pervasive in Aegean Bronze Age archaeology. Texts (largely in the form of clay documents) tend to be treated by specialists, often with more epigraphical than archaeological training. Cretan Hieroglyphic, Linear A and Linear B, only the last of which is actually deciphered, requires philological skills well beyond most prehistoric archaeologists. ‘Pictures’ of various kinds, on wall paintings, seals and pottery, are also the subject of a particular iconographical/ art historical approach which often fails to tie in effectively with archaeological approaches to artefacts. Despite these methodological barriers to a conjoined study of artefact, picture and text, they need to be confronted here if we are to gain some impression of their status as things and/or objects.

The Aegean Bronze Age is actually a highly suitable domain of enquiry because of the wide range of artefacts, pictures and texts in various media. While artefacts and images are of course common throughout the Aegean Bronze Age, texts are not, with the first writing on clay documents not appearing until the early part of the 2nd millennium BC on Crete. Cretan Hieroglyphic and Linear A are the two earliest scripts known, the latter gradually supplanting the former. Linear A then disappears at the expense of Linear B, probably some time in the 15th century BC. Linear B, an early form of Greek, is the only one of these scripts to have been deciphered.

These three scripts are used in various ways in relation to artefacts and pictures. We will first consider the relationship between text and picture, before moving on to that between text and artefact.

Text and Picture

Although Aegean Bronze Age scripts do have many signs which appear pictographic, all three scripts are principally of a syllabic character: Cretan Hieroglyphic has around 90 syllabograms, Linear A has 75 syllabic signs and Linear B 87 (Olivier 1986, 378–9). Nevertheless, a certain number of signs are recognised by scholars as being logograms (or ‘ideograms’) rather than syllabograms, and many of these are pictorial – so the logogram for chariot in Linear B is a schematic iconic depiction of a chariot. In Linear A, for example, there are 19 signs which are ideograms (or more correctly, logograms) of pottery shapes, and all of these are pictorial (Godart and Olivier 1976–1985; see Fig. 8.1 ). Other pictorial Linear A ideograms exist for wine and olive oil, as they do (amongst others) in Linear B. In Cretan Hieroglyphic there are currently 33 identified logograms, and in some of these cases syllabic signs double up as logograms, a phenomenon also seen in Linear A and B (Olivier and Godart 1996, 16).

Fig. 8.1

Some Linear A logograms of pottery shapes (from Godart and Olivier 1976–1985, vol. 5, LII)

There is clearly a rich interplay at work here between artefact, picture and text, extremely valuable for our current focus on how they interrelate in the transformation of things into objects. Pictographic representation in texts, while perhaps murdering the thing through the word, does maintain a shadow of the ‘murdered’ thing. In the transformation of the representation from pictograph to ideograph, the image may, of course, be read less and less as an icon and more and more as a symbol, with the shadow hence gradually fading. And here I think we can very usefully go a step further by following Robertson (2004) in distinguishing between direct and indirect representation. Indirect representation is when an icon that looks like the object represented is used to reference the word. The representation is thus indirect because it proceeds from iconic sign to object to word. Direct representation, on the other hand, is when the textual signs correspond to syllables or phonemes – the link between the sign and the spoken sound is direct.

With indirect representation, the textual sign is identifying with the object/thing; it is as if the textual sign is in sympathy with the ‘murdered’ thing it references (the image of the thing retains its memory). This kind of sign establishes a relationship of resemblance with the referent, rather than sacrificing that iconic link in favour of an auditory one. Robertson argues that indirection is typical of the first writing; he also observes that this does create difficulties, as many concepts do not lend themselves to simplistic iconography (e.g., how do you draw the word ‘for’?)

However, a sign looking like it might be an image of something does not mean that it is necessarily functioning as a pictograph. Pictographs can become ideographs – which is to say that the sign ceases to signal its referent through visual resemblance, and instead operates through a habitual and conventional association. Further still, the process of representation can become rather more direct, with a visual sign becoming a syllabogram, that is, taking on a phonetic value. Robertson suggests that this might often happen through ‘acrophony’, that is, by taking the first syllable of the word represented by a logogram. So if olive oil is ‘elaiwon’ and is represented by what looks like an olive branch, then at some stage in script development the olive branch may come to stand for the first syllable of the word ‘elaiwon’, for example, ‘el’.

So, what is the situation with respect to the Aegean scripts of Cretan Hieroglyphic, Linear A and Linear B? We have already mentioned that while many of the signs may appear to be pictorial, most are syllabic signs. Applying Robertson’s terms, each script appears to function through direct representation. Here one might raise another point of interest from Robertson: that a general process of change seems to be at work in writing systems, such that a progression from icon to symbol (from indirect to direct) is seen, but rarely vice versa. Is there any indication of a progression from indirect to direct representation in the Aegean Bronze Age scripts, as Robertson would expect? The simple answer seems to be ‘no’, unless we are simply missing the earlier script or scripts out of which Cretan Hieroglyphic and Linear A evolved. This is not impossible, as the origins of each script are unclear. One observation we might add, however, is that there does seem to be a certain degree of indirect representation in Cretan Hieroglyphic. If we look particularly at Hieroglyphic seals, Hieroglyphic signs are used in conjunction with decorative motifs on very small sealstones (often only 1–2 cm across; see Fig. 8.2 ).

Fig. 8.2

Sealstones with Cretan Hieroglyphic (from Krzyszkowska 2005)

This raises the question as to whether these signs were really meant to be ‘read’ as signs, or whether their pictorial qualities were in such instances more significant (see Krzyszkowska 2005, 95–8). We might consider this some indication of an early use of indirect representation, one that fades over time. When considered in the light of Schwenger’s comments above, an evolution from indirect to direct representation can almost be seen as a gradual dispelling of the ‘shadow’ of the object. Yet this process never quite seems complete as long as the syllabograms maintain an iconic, imagistic aspect.⁵

Text and Artefact

While above we have given some thought to the various ways in which artefacts may be represented pictographically and textually, we might also consider the relationship between text and artefact much more directly. The question we can ask is this: how closely associated are texts and the artefacts they reference? Especially in Cretan Hieroglyphic and Linear A, textual documents can be very closely associated with the commodities they name – in the form of what are called ‘direct object sealings’ (Weingarten 1986; Schoep 1999; Krzyszkowska 2005, 99–101). These sealings consist of lumps of clay pressed over the mouths of jars and pithoi, or over pegs securing chests or doors, and then stamped with seals (Fig. 8.3 ). Thinking in Schwenger’s terms, naming and labeling may ‘murder’ the thing and commodify it, but the murdering text stays close to the body just to be sure. Of course, it risks being drawn back into the soup of thingness itself.

Fig. 8.3

Reverse side of a direct object sealing from Malia (from Krzyszkowska 2005)

Another category consists not of sealings associated with artefacts, but of artefacts on which script has been directly inscribed or incised. This phenomenon is encountered in all three scripts. In Cretan Hieroglyphic there are very few examples, with the majority occurring on small juglets known as ‘Chamaizi pots’ (Olivier and Godart 1996, 294–311). These juglets have between two and five signs incised or painted on their bodies, usually at the maximum diameter (Fig. 8.4 ). As the script remains undeciphered, the meaning of these signs is unclear.

Fig. 8.4

Chamaizi pots bearing Cretan Hieroglypic signs (after Poursat 1992)

Linear A inscriptions on pottery are rather more frequent, with around 37 examples (Olivier 1986, 384). Many of these occur on storage jars (‘pithoi’), and are often found incised close to the vessel mouth (Godart and Olivier 1976–85; see Fig. 8.5 ). There is a concentration of these inscriptions in one period in particular, Middle Minoan IIIB, a feature noted by Sir Arthur Evans. Some have been found beyond Crete too, with a recent example coming to light at the Cycladic site of Akrotiri on Thera (Karnava and Nikolakopoulou 2005). As with the Cretan Hieroglyphic examples, the meanings of these inscriptions are unknown; do they relate to the contents of the pot, the producer, or perhaps the intended consumer? What is striking is the very different kind of vessels involved in the two scripts, with the Chamaizi vases of Cretan Hieroglyphic presumably containing very small quantities of something valuable, while the pithoi of Linear A are designed for bulk storage of commodities such as oil, wine or grain.

Fig. 8.5

Linear A inscription on a pithos from Akrotiri, Thera (from Karnava and Nikolakopoulou 2005)

Turning to Linear B, problems of decoding the meaning of such inscriptions no longer apply. Furthermore, there are more examples, with Olivier recording 144 inscriptions painted on pottery vessels (Olivier 1986, 384; although by 1999, Van Alfen mentions 180 examples). As with earlier examples, the inscriptions usually only consist of a handful of signs (see Fig. 8.6 ), and they refer to toponyms or personal names in many cases (Van Alfen 1999). While Van Alfen argues that these inscriptions were indeed meant to be read, and played a role in the palatial bureaucracies, some scholars have argued that they may instead have had a largely decorative function (see Van Alfen 1999, 253, fn. 6). He goes on to argue that the inscribed stirrup jars themselves functioned not only as transport vessels but also as documents, playing the role that sealing nodules played in other settings. Making the artefact into a textual document is a very direct example of the role that texts can play in objectifying various categories of things.

Fig. 8.6

Linear B signs on an inscribed stirrup jar from Malia

It would appear that Aegean scripts and their documents play an objectifying role through their capacity to name and image certain kinds of thing. They do seem to be strongly focused on ‘things’ that lend themselves to iconic naming/imaging, primarily material commodities such as grain, oil, wool, metal etc (i.e., not many abstract concepts such as life/death/god which might be rather more problematic in their representation!). But once objectified and artefactualised do these clay documents not then become things themselves (i.e., parts of assemblages)? We should also consider the extent to which different parts of a community would be able to recognise either the document itself for what it was, or indeed the signs for what they represent. Linear A documents might very well have been objects to some (nameable, transparent) but things to others (un-nameable, incomprehensible).

Discussion

We might benefit from Law’s work on Portuguese vessels as a means of thinking through the interactions between texts, images and artefacts in Aegean Bronze Age contexts. We have to consider network functionality, particularly the overriding function of texts in administration, elites having set up a network, or ‘macro actor’ (Czarniawska and Hernes 2005, 9), for controlling the flow of commodities. This control is achieved, as much as anything else, by making objects of things. And while initially we can see the use of artefacts and images in this process, words do seem to come into the process increasingly. Moreover, images and words themselves then take changing artefactual form, with their distance from the imaged or named artefact increasing/decreasing in network and/or physical space.⁶ The innumerable connections and disjunctures between artefacts, pictures and texts create particular physical and relational topologies that, to my mind, merit much further study than has been possible here in this brief treatment.

Let us now return explicitly to the theme of agency. If we consider the agency of these Aegean Bronze Age elites seeking to oversee the production and distribution of commodities, then we may observe that this agency is very much contingent upon complex networks of interconnection (see also Malafouris in press). These networks may not only have complex multiple topologies, but also may be composed of a wide variety of actants – artefacts, pictures and texts – that are assembled to create the macro actor that is Minoan administration. Therefore, a shift of emphasis is required. This need not be dehumanising or deindividualising or place undue emphasis on impersonal networks, things and objects; but it is certainly non-anthropocentric. One might have expected this kind of perspective to be welcomed in archaeology, given the discipline’s inevitable material bias (at least in its methods). However, approaches from ANT or parallel domains have been only very rarely applied to archaeological scenarios, with a few recent exceptions (Olsen 2003; Witmore 2004; Webmoor and Witmore 2005; Watts 2007, & this volume). Despite its reticence, archaeology is well-placed to make a unique contribution to the widening debate on the character of nonhuman, ‘material’ agency.