My article 'Arbitrary locations' was reprinted in 2009 as part of Mark-Anthony Falzon's edited volume Multi-sited ethnography: theory, praxis and locality in contemporary research (you can read Falzon's excellent introduction to the volume here). On this occasion, I added a brief post-script to the original article, which updated the argument and reflected on some of its limitations. The definitive version of this text is included in Candea, M. 2009. 'Arbitrary Locations: In Defence of the Bounded Field-site (with a new afterword)'. In Multi-sited ethnography : theory, praxis and locality in contemporary research, (Ed, Falzon, M.-A.). Aldershot: Ashgate. pp. 25-46.
A partial postscript on arbitrariness
Perhaps the best way to ‘refresh’ (as one might a webpage) the argument of the 2007 article reprinted here in an abridged form, is to refer to philosopher Alfred North Whitehead’s distinction between what he terms a nexus and a society. A nexus, writes Shaviro, quoting Whitehead, is
“ ‘a particular fact of togetherness among actual entities’ ([Whitehead 1920/2004:]20); that is to say, it is a mathematical set of occasions, contiguous in space and time, or otherwise adhering to one another. When the elements of a nexus are united, not just by contiguity, but also by a ‘defining characteristic’ that is common to all of them […], then Whitehead calls it a society (34). A society is ‘self-sustaining; in other words… it is its own reason… the real actual things that endure,’ and that we encounter in everyday experience, ‘are all societies’ ([Whitehead] 1933/1967:203-204)” (Shaviro n.d.:2)
This distinction between nexus and society is a very good entry-point into the question of holism (in its simplest – Aristotelian – form, the claim that an entity is more than the sum of its parts). If a nexus describes a set of contiguous parts, the society which corresponds to it is the whole. Whitehead’s distinction gives us a good way of talking about the evident fact that not every nexus doubles up as a society, i.e. not every collection of parts is a whole.
So what does this have to do with multi-sited ethnography (henceforth MSE)? My argument is that the multi-sited imaginary productively challenged the assumption that the local nexus mapped out by the ethnographer was also a society. It challenged in other words a simple holism (let’s call it holism 1) according to which say, ‘the Nuer’ or ‘the Bhopal disaster’ could be examined ethnographically as a whole in isolation from the broader multi-local processes in which its various elements were enmeshed. What initial formulations of MSE did not set out to challenge however was the notion that ethnography should set itself as a task (or at least as a horizon) to map and investigate wholes. Take for instance Smith (2006, 13–5, emphasis added)
‘Although I was based in one town in southeastern France, Aix-en-Provence, this is not an ethnography of [Maltese] settler life in that town, but rather an ethnography of an entire diaspora. This is thus an ethnography of social practices engaged in in rented halls, during day- and week-long trips, and in other temporary spaces, in the interstices of the settled villages and city neighbourhoods that have been traditional loci of study […M]ost of the settlers I spent time with [in Aix] were clear outsiders and isolated even from their closest neighbours. They were connected to other repatriates, but not to Aixois, and I don’t think that I was ever given the telephone number or address of a nonsettler, a “true” French man or woman, to contact.’’ (Smith, 2006, pp. 13-15, emphasis added).
Having dissociated the nexus from the society, early proponents of MSE prompted us to follow the society (entities which were more than the sum of their parts, such as ‘cultural formations’, ‘the world system’, ‘diasporas’ or ‘events’ – let’s call this holism 2), while leaving behind the irrelevant bits of mere geographic contiguity. The focus is on the thickly interconnected entity that is Maltese settler identity, say, rather than on the minutiae of daily cohabitation in a French town where people don’t know each other.
To think of one’s field-site as an arbitrary location, by contrast, is to think of it as a ‘nexus’ which bears no necessary relation to – and thus gives one a vantage point on – the ‘society’ or ‘societies’ under investigation. This is not in and of itself a better way to go about research, any more than multi-sitedness in and of itself is either sufficient or necessary to provide power and relevance to anthropological ethnography. The arbitrary location is just an alternative, which may be more or less productive than others, depending on the subject and aims of the research. It highlights, by contrast, the implicit holism of the multi-sited imaginary. However, whether one chooses to work in arbitrary locations or to map out whole entities, the broader point is that it pays to be explicit about the way one makes such choices.
Re-reading the article in the light of the other chapters in this volume, I find myself compelled to score a few argumentative own-goals. Firstly, the article failed to distinguish between methodological caution and ontological postulation. As Cook, Laidlaw and Mair (this volume) rightly point out, this is analogous to a confusion between keeping the description ‘flat’ and claiming the world actually is flat. The point is not that there are no wholes, but that arbitrary locations give one a vantage point from which to observe (and indeed engage) the processes whereby wholes are made and unmade – just like ‘flat’ description allows one to see the way actors themselves actually deploy scale.
Secondly, two different flavours of arbitrariness get entangled in the article. On the one hand there is the all-out playful, ‘what-the-heck’ arbitrariness of, say, a book written without the letter e. On the other, there is the more delimited and technical use of the term, in which the field-site is arbitrary in relation to the object of study. This arbitrariness allows the site to remain a nexus, space that cuts through meaning. In this second sense, of course, the method itself is far from arbitrary in relation to the anthropologist’s theoretical or analytical concerns. I was alerted to my failure to clearly distinguish these two kinds of arbitrariness – the general/whimsical and the technical/limited – by the stated concern of some readers that to describe anthropological fieldsites as ‘arbitrary’ would suggest (particularly to a non-anthropological audience), that ‘anything goes’ and that there are no ‘better’ or ‘worse’ anthropological descriptions or analyses. On the contrary, arbitrariness in the technical, limited sense is anything but facile whimsy: it is a purposive method which involves conscious and motivated processes of selection, an instance of what George Marcus (this volume) calls “a strong norm and accountability for intended, structured partiality and incompleteness in ethnographic research designs”.
Thirdly, this approach is also likely to be perceived as arbitrary from the purview of at least some of the people the anthropologist is working with. This raises in sharpened form the question of engagement, which the article does not address. This being said, arbitrariness in this sense does not in any way preclude collaboration, unless this is imagined restrictively as a direct and total alignement of the anthropologist’s interests upon the interests of the people they work with – it certainly does not preclude collaboration in the far more exciting senses explored by some of the contributors to this volume (Marcus, Matsutake Worlds).
Fourthly, there is a great time-shaped hole in the paper, which stops short of considering the temporal and iterative process whereby boundaries are drawn and redrawn, arbitrariness shifted, sharpened or blurred, in relation to happenings in the field, and in interaction with others (see for instance Gallo, this volume). Building this back in would form part of the answer to the previous concern, as well.
Yet the most important corrective is a matter of style. Too often, the playful/polemical tone of the piece left readers with the impression, despite explicit claims to the contrary, that this is some kind of all-out onslaught against multi-sitedness, its proponents or practitioners. It is not. In fact, this article is situated within as much as it is about the wave of methodological reflexivity generated by proponents of MSE. Multi-sitedness is a positive development for the discipline, although it is not a sine qua non of good ethnography. But there was certainly no intended suggestion that doing fieldwork in one place is in and of itself better than doing it in many (cf. Weissköppel, this volume): the number of sites has nothing to do with their arbitrariness (any more than it has to do, incidentally, with their ‘complexity’).
A final thought: in a penetrating and thought-provoking comment on the paper, Akhil Gupta (pers. comm.) suggested that the notion of ‘partiality’ (following Haraway) would be a preferable alternative to that of ‘arbitrariness’. This is in many ways appealing, not only because it would put my usage in line with that of many contributors to this volume, but also because the argument patently owes a large debt to Marilyn Strathern’s Partial Connections (2004). And yet, while both partial and arbitrary suggest a deferral of completeness, the relation between parts and wholes implied is not quite the same in each case. A partial account postulates and hankers after or runs from its (impossible?) whole; an arbitrary location just cuts holes through wholes: it is little more than a ‘clamp’ which keeps accounts flat, allowing ‘the actors to speak for themselves’ (see Latour 2005 and Krauss, this volume). Furthermore, the appeal of partial is at least in part due to its polysemy (as in incomplete, but also partisan, interested, not impartial), which has been very effectively used since the 1980s to wed a critique of holism to a critique of objectivity. Salutary as this demise of the ‘God’s eye view’ has been, arbitrary belongs to a different project: that of imagining new bases for a ‘post-posivist empiricism’.
M.C., Cambridge 25/08/08
 And (pace Gatt, this volume), ‘societies’ here refers to hybrid entities or matters of concern, not to social relations as divorced from the environment (cf. Candea 2008).
 After all, developing a holistic account may be precisely what is aimed at (e.g. Horst, this volume)
 as Cook, Laidlaw and Mair once again very clearly point out. In my own case one of the driving concerns behind this entire discussion was an interest in the holistic depiction of categories of identity (French, Corsican, etc.). Crucetta is arbitrary in relation to Corsicanness (it does not encompass it, represent it, or bear any necessary relation to it), but not in relation to my argument about or interest in Corsicanness (see Candea forthcoming).
 For instance Akhil Gupta (pers. comm.), and perhaps, by implication, Nadai & Maeder (this volume).
 I borrow the term from Born (n.d.)