In her discussion article ‘Fluid consistencies. Material relationality in human engagements with water’, Veronica Strang argues that ‘water provides a useful focus for thinking about relationships between things and persons and between material properties and meanings’ (Strang Reference Strang2014, 133, emphasis added). Water permeates organic things and flows and connects in a multi-scalar way. Therefore her article emphasizes ‘how material and social processes combine to provide both fluidity and consistency at every level of human–non-human engagement’ (ibid., 133). Ontologically speaking, the emphasis on fluid relations and processes makes her a relationist, i.e. objects emerge from their internal or external relations (Harman Reference Harman2009). I will summarize my reactions to her article in three major points.

(1) The most important one is her use of fluid water as a metaphor for relations, processes and change. This is not uncommon in Western thought since Heraclitus and his ‘eternal becoming’. I differ from both Strang (Reference Strang2014) and Edgeworth (Reference Edgeworth2011) in their suggestion that it is good to think in terms of the logic of flows (and water). Such perspectives have become popular in several contemporary archaeological theoretical positions. In neuroarchaeology it is suggested that ‘the cognitive life of things is about things in motion; it is about hybridity, fluidity and genuinely interactive relationships between brains, bodies and things’ (Malafouris and Renfrew Reference Malafouris, Renfrew, Malafouris and Renfrew2010, 9, emphasis removed). Similarly, Hamilakis (Reference Hamilakis2013, 4) discusses ‘flows of substances, sensorial stimuli, memories, affective interactions, and ideas’. The current upshot for assemblage theory in archaeology also emphasizes fluidity and flow (Harris Reference Harris2014; Lucas Reference Lucas2012; Normark Reference Normark2006; Reference Normark2008; Reference Normark2009; Reference Normark2010). A few years ago I wrote that ancient Maya causeway assemblages ‘have been tied together through the production and repetition of several flows’ (Normark Reference Normark2010, 152). Since then I have encountered some potential problems with the ‘ontologies of flow’ proposed by Bergson (Reference Bergson2004), Deleuze (Reference Deleuze1994), DeLanda (Reference DeLanda2002), and Bennett (Reference Bennett2010).

Do the metaphors of fluidity, flow, flux and liquidity describe consistency, continuity and relations? Are fluids more continuous and consistent than non-fluids? Metaphorically I would say that solids are more continuous and consistent in time and space but the relations between actual solids are not obvious in the present where they appear to be discontinuous from one another. We cannot detect a flow at an instantaneous moment (from which it would appear solid). A flow can only be observed in duration across a spatial distance. Hence time is extended and made continuous, and for these reasons time is associated with movement rather than change. A flow is, contrary to a solid, moving in space at an observable speed. It is deemed more active than a solid. Therefore fluidity is metaphorically used to describe continuous relations between solids. In this perspective even solids melt into fluids and fill voids between solids. This is an ‘ontology of infilling’ suitable for archaeology, where the narrative voids surrounding artefacts, strata, ruins, etc. are filled with a fairly continuous narrative, usually a cultural history but also the continuity of human agency. This ontology suggests that processes and flows are more real than objects. A solid object is reduced to an actual frozen instantaneous moment in a virtual flow of change.

The most influential philosopher among present-day ‘philosophies of flux/fluidity’ is Gilles Deleuze (1925–95). In Deleuzeoguattarian terms the metaphor of fluidity is used to describe something that breaks territorialized boundaries. Flows are deterritorializing boundaries (Deleuze and Guattari Reference Deleuze and Guattari1987). However, is not a flow a moving territorializing boundary rather than the crossing of a boundary? On the beach the water's edge moves back and forth with waves and tides. It is still a clearly visible boundary even if it changes more quickly than a solid. You either get your feet wet or you do not if you try to escape the wave's territorializing edge. Hence the fluid/flow of water still has actual boundaries.

There is usually a dichotomy between the fluid/flow and the solid/static. Why is not gas, such as oxygen or nitrogen, used as a metaphor for change and process instead of a liquid? After all, like liquid water, these gases also permeate organic things and connect in a multi-scalar way. The unimportance of gas as a metaphor for change in Western thought is probably due to a long philosophical tradition going back to at least Heraclitus where air was ‘invisible’ but felt through wind, breath etc. This third ‘state of matter’ (gas) complicates a dichotomy, not to mention the fourth ‘state of matter’ (plasma). Dichotomies need two extreme positions, not four.

Sutherland (Reference Sutherland2014, 172, italics original) argues that ‘all philosophers of flux, operating in the tradition (allegedly) inaugurated by Heraclitus, are premised upon thinking the absolute qua becoming. Becoming is the necessary being upon which all causality is grounded’. Hence any arguments regarding continuous flux rely on one necessary and unchanging condition. Even becoming must remain a being. Becoming itself cannot become. If there is something Meillassoux (Reference Meillassoux2008) tries to teach us it is that there are no necessary entities. Only contingency is necessary, in his view. However, Sutherland points out that, for the very same reason as above, not even Meillassoux escapes the problem of a necessary being with his hyperchaos (absolute contingency) that supposedly precedes both being and becoming. This is because he still posits one necessary condition (contingency).

Despite the long tradition of fluidity in Western philosophy, references to such philosophers are few in Strang's text. Hegel's dialectical process and flows are mentioned (Strang Reference Strang2014, 137). However, Strang makes more use of Latour and actor-network theory (ANT), which she argues expresses ‘the fluidity of relationships between humankind and non-human species and things’ (ibid., 135). She suggests that ‘notions of generative seas or “flux” are appealing, resonating with Latour's (Reference Latour2005) view that agency emerges neither from people nor from things but from their combination’ (Strang Reference Strang2014, 139). Why does Latour's view resonate with something fluid? Combining a person and gun (a popular Latourian example) is not fluidity, it is only a relation. The fluidity metaphor does not easily fit Latour's ANT.

Elsewhere in the text, not in connection with Latour, Strang argues for ‘a shifting continuum of agency’ (Strang Reference Strang2014, 142). The Latourian network is not a continuum of agency. There are nodes and gaps in the network. Gaps are not connected and in that sense there is no such thing as a continuum, only selectively connected nodes. In other words, agency is discontinuous (if agency exists at all).

The main reason why we should not conflate (Deleuze’s) fluidity with Latour's network can be found in Harman's (Reference Harman, Bryant, Srnicek and Harman2011b, 6) distinction between underminers and overminers of objects. Undermining is a form of reductionism that destroys objects in favour of something more fundamental, either atoms or a deeper force. Philosophers of flux reduce objects to deeper immanent processes of becoming, like Deleuze's virtuality and Bergson's duration. In overmining, the object becomes a fiction of relations and qualities (Harman Reference Harman, Bryant, Srnicek and Harman2011a, 38). This ‘upward’ reduction sees objects as part of something greater, like Latour's network. Hence Latour's network is the ‘upside-down’ version of Deleuze's fluidity.

In the comment section of Strang's article, Benjamin Alberti (Reference Alberti2014, 162, emphasis added) refers to Karen Barad when he writes that

Like Deleuze's virtuality, Barad's (Reference Barad2007) ontology of intra-action undermines objects by making them part of a deeper unity and relations. The objects are reduced to static illusionary reifications (Morton Reference Morton, Gaard, Opperman and Estok2013, 60 f.). True, a flow of water always occurs in relation to something else but this does not affect the ‘essence’ of the water molecule. To Harman (Reference Harman2011b), a real object interprets/translates another real object as a sensual object. It interacts, i.e. forms a relation, with the sensual object but never with the real object. Harman insists there is a withdrawn essence of each real object that is not reducible to a deeper force or external relation. Relations cannot precede objects since relations occur between objects. Unfortunately, Harman can never describe the real object-in-itself because it is, conveniently, withdrawn from all access.

In my view, the past was not a continuous flow of events, agency and activity. It involved ‘chunks’ and ‘voids’, and that is not well captured by a model of fluidity or liquid water because it fills voids with something other than observable objects, usually a transcendent umbrella term like ‘society’ or ‘culture’ (Normark Reference Normark, Fahlander and Oestigaard2004). The archaeological ‘record’ itself reveals a ‘chunkier’ reality of the world that is discontinuous.

(2) Another issue I wish to address in Strang's article is the use of the term ‘anthropocentrism’. Although Strang suggests (2014, 135) that we should shift ‘to less anthropocentric visions of human–environmental interactions’, she maintains a strong anthropocentric perspective in the article. In her words (ibid., 140),

This is truly anthropocentric because who else has wealth, concepts, core meanings, culture and history but humans?

Anthropocentrism is also evident in the maintenance of the human/subject–environment/object distinction. One term is more specific (human) and the other is more general (environment). This reflects an arborescent/hierarchical thinking where the general is in a hierarchical position above the specific (Deleuze and Guattari Reference Deleuze and Guattari1987). Hence Strang focuses on the specific human relations with water in general rather than viewing water from a ‘specific’ hydrocentric and rhizomatic perspective.

Morton (Reference Morton2010, 76, emphasis removed) argues that ‘hunting for anthropocentrism is anthropocentrism’. By focusing on the problems with anthropocentrism we remain attached to an anthropocentric view. However, centrism is unavoidable. We are humans and we cannot know what it is to be something else (Bogost Reference Bogost2012). Let us instead embrace not only the anthropocentric perspective but also other ‘centric’ perspectives.

As humans we are also geocentric (although this can very well change in the future). Despite half a century of space travel most humans are still earthbound, with all that that implies. For one thing, related to my first objection, the geocentric perspective influences how we primarily view water as a liquid, a flow or fluidity. The physical state of liquid water is often conflated with its motion. Liquid water flows when it is in motion and it only does because of gravity. In zero gravity liquid water forms a sphere unless it is in contact with another object. Earthbound water in gas form is moving even more than in its liquid state, but it is not flowing. Water in its solid form, such as a glacier, can also move, albeit at a slower speed.

Thinking with water as a flow is therefore primarily dependent on our geocentric context where most water we encounter is in a liquid state even though some earthbound water is locked up in various other entities (such as soils, rocks, organisms etc.). We conceptualize water as a liquid by default. Had we lived on Saturn's moon Titan, where surface temperature is roughly 94 K (–179 °C), we might have had another view on fluidity. There much of the surface is made of solid ice and lakes are made of liquid hydrocarbon. In fact, most of the water in the universe is either in solid form or in gas form. These extraterrestrial examples may not be of relevance to archaeological contexts on Earth but Strang's claim is that we should become less anthropocentric. If so we must emphasize that water is not limited to human or geocentric perspectives.

This geocentrism is also reflected in Davies's comment (2014, 156) that water ‘needs to be understood as a rather unique material form which differs in most respects from more solid materials – water moves, flows, changes; it is, in this regard, an atypical form, worthy of atypical discussion’. The reason why water is this ‘unique’ material is because it has a narrow temperature span between its gaseous and solid forms at the atmospheric pressure at sea level on this planet. Change pressure, gravity etc. and the temperature difference between vapor and ice will change as well. Other entities would also behave like liquid water in other contexts. Liquid iron also moves, flows and changes but we seldom encounter it in any great quantities on this planet.

(3) My final reflection on Strang's text and its commentaries concerns the terms ‘matter’ and ‘material agency’. These are treated as unproblematic terms. To Strang matter just is, just like Scarborough argues (Reference Scarborough2014, 152, emphasis original) that he does ‘not see water as having a mind of its own or any sentience to deliberately work for or against human interests. Water simply is and humans have directed its course with all the unintended consequences of their efforts’ – despite having a perspective that emphasizes ‘agency’ and ‘becoming’. Strang also treats matter as something that is because all things ‘are dynamically composed of flows of condensed matter’ (Strang Reference Strang2014, 135). We are not told what condensed matter is but I suspect it is something like the ‘preindividual’ (virtual) flow of Deleuze. In a view where everything is a (virtual) flow that actualizes into temporary ‘solid’ entities there are no true distinctions. Thus if an object is described as the abstraction of a deeper flux it is a monism (Deleuze and Barad are monistic philosophers). If there are many fluxes, then each has some unique character and that makes it an object (Harman Reference Harman2011b, 8 ff.).

As pointed out by Deleuze and DeLanda, matter is a concept influenced by a hylomorphic view. Hylomorphism is the idea that matter is inert and ‘dead’, entirely passive. Matter can only be given a form by an external agency. It is therefore odd that the term ‘matter’ and its cognates (material, materiality, material culture etc.) still are maintained in ‘ontologies of flow’ (such as DeLanda's neomaterialism or Bennett's vibrant matter). In these perspectives matter is redefined as active (having ‘agency’) and being self-organizing, but matter as a concept is not questioned.

According to Davies (Reference Davies2014, 154), Strang's essay ‘lacks a concise examination of what forms material agency takes, how it coincides with or differs from human agency’. In my view ‘agency’ is a superfluous term, particularly in relation to the terms ‘human agency’ and ‘material agency’. It is superfluous since agency is something acquired, not there from the beginning. In social science the human being is not always an agent. It acquires agency to not only be a static thing. In material-agency perspectives, agency is therefore also bestowed on ‘matter’ so it can act like people (Ingold Reference Ingold2007, 11). If everything has agency there is no point discussing it as a specific property, just as we do not usually discuss the ontological aspects of mass, colour and temperature, since most entities have these properties.

Agency has properties similar to vitalism, which is another prevalent idea among some philosophers of fluidity (such as Bergson, Deleuze and Bennett) (see Hallward Reference Hallward2006). Bennett traces vitalism back to Immanuel Kant's Bildungstrieb, which is a formative drive that attaches itself to and enlivens matter. It focuses on the self-organizational power inherent in organisms but not in matter itself. Bildungstrieb can only be known indirectly from the organisms it has created. It is an impersonal, ahistorical agency that drives organisms on (Bennett Reference Bennett2010, 65–69).

It was Kant who introduced the subject–object correlate/relation. ‘Correlationism’ is the term used by Meillassoux (Reference Meillassoux2008) to describe the Kantian and post-Kantian positions where subject and object never can be thought of separately; they are always correlated with each other, forming an unbreakable unit. In this view the subject is usually active and the object is passive. Kant attributes living organisms with Bildungstrieb to make them more than hylomorphic matter. Something vital is needed and so is agency needed to be injected into objects to make them ‘active’. In a non-correlationist approach ‘vitalism’ and ‘agency’ should not be needed as terms as all entities are inanimate yet interact in many ways. Hence we do not need to follow Cole's (Reference Cole2013) suggestion to re-examine the medieval ‘call of things’ that Kant and Fichte excluded through the correlationist circle.

In conclusion, things are not always what they seem to be. For millennia it has been tempting to view water, rivers, flows etc. as metaphors for the changing world. Yet this is always a ‘matter’ of scale. Standing on the Moon and observing the Earth with the ‘naked eye’ we would not see flows of water. Zooming in on the single water molecule we do not see flows of water either. Somewhere in between these scales we encounter liquid water, particularly at a human scale and at humans’ tolerable levels of pressure, temperature etc. At that scale we sense water as a flow. This makes our perspectives on water anthropocentric. We see those aspects of water that are open to us but water is more than human relations to water. Water is also hydrocentric (see Normark Reference Normark2014).