What if a method didn’t come down from our theories, but was drawn from the material organisation of the thing we study? That question is the whole of Material Derivation — an attempt to let a phenomenon specify the method through which it is known. The phenomenon here is a seashell: its accretion at the mantle margin, its cementation into limestone, the vacancy chains through which hermit crabs pass it on, and the skin that keeps it from dissolving into the sea it precipitated from. From each of these I derive a rule for how such an object has to be studied — and then put those rules to work in a studio, where an article was cut into hexagonal tiles and reassembled, collectively, into knowledge no single hand could hold. I gave this as a talk at EASA 2026, in the Studio Anthropology panel convened by Ignacio Farías (Humboldt University of Berlin) and Andrew Gilbert (Burg Giebichenstein University of Art and Design, Halle). Below are the full transcript and the slides. It grew out of Liminal Seashells, written with Nan Makarova, and it sits where I tend to work — between art and science, and close to the material.
I work at the meeting of art and science, and my concern is materiality, studied materially, from what the matter itself does rather than from the theories we would otherwise bring to it. So let me begin with an object, not an argument. Take a seashell (i go around and give people seashells)
Hold a seashell to your ear. The folklore offers two answers to what you hear — the ocean, or the blood in your own head, the romantic and the anatomical. Both are wrongish. What the cavity returns is the ambient sound of this room, amplified- a Helmholtz resonator selecting frequencies already present. I raise this not as a metaphor but as a precise epistemic figure. The shell does not reflect a pre-existing ocean back to you; it diffracts, producing from what is already there a patterned difference. Reflection is the trope of representationalism, the assumption that knowledge mirrors a world held at a distance.
Ethnographic knowledge operates in the same register. It neither mirrors worlds nor invents them; it is an apparatus that amplifies and patterns relations already underway. If that is what our knowledge does, the design of the apparatus — the method — is not a neutral preliminary but the consequential act. This is the lesson of Law and of Mol’s ontological politics: method participates in enacting it. The question that lesson leaves open is the one this project takes up. If a method enacts, on what basis should a method be designed? The reflexive answer locates that basis in the knower’s accountable position. The alternative proposed here is materialist: that a method can be designed from the material organisation of the phenomenon it studies.
The proposal came from an empirical impasse. An earlier study, Liminal Seashells conducted together with Nan M, attempted to write the shell across its registers — the mantle’s biochemistry of secretion; the sedimentation and cementation that bind dead shells into limestone; the industrial extraction of that limestone into concrete; the folklore of the shell as a haunted, speaking thing; the timescape in which a mollusc’s season and a rock’s ten million years coincide. A dozen registers, and no way to hold them together. The difficulty was not excess but organisation. Written as a single continuous argument, the account imposes a coherence the shell does not possess; written as a list, it severs connections between registers that are nonetheless real. Strathern’s term is exact: these registers are partially connected, linked without summing to a whole, less than one and more than many. What was missing was a procedure adequate to partial connection.
The lack is general. Multimodal anthropology, studio anthropology, participatory methods, and postqualitative inquiry have each diversified content, authorship, or output — but none relocate where the method’s form originates. In every case, that form still descends from the anthropologist’s epistemological commitments. That is the assumption at issue.
A model of coherence that does not totalise is required, and its clearest statement is not anthropological but pictorial: Gleizes and Metzinger’s insight, on Cubism, that an object has as many forms as there are planes of meaning. The conventional reading — many views of one stable object — inverts the claim. Multiple viewpoints presuppose a single object held whole behind its appearances, with the viewer sovereign over it; that is the representationalism already in question. What the painters assert is that the object has as many forms as there are planes of meaning, and that its essence is the plurality of minds engaging it. Form is distributed across the planes, and is not recovered by assembling them. The proposition converges with two that anthropology holds, without collapsing into either. It meets Haraway’s situated knowledges — no view from nowhere, objectivity as the accountability of partial perspectives — but adds what that formulation leaves implicit: the partiality is in the object’s form, not only in the knower. It meets Strathern’s partial connections, but treats them as a technique of composition rather than only a condition of analysis; Cubism tries to build a coherent artefact from it. We diagnose distribution; we have not learned to compose with it. The shell demands exactly that composition, since its meaning is distributed across the crab, the tide, the quarry, and the collector — as many forms as it has relations.
The decision taken here is that where an object’s agency is distributed and only relationally coherent, a method concentrating knowledge in a single synoptic vantage does not simplify the object but misdescribes it by construction. Distributed knowledge is required because distributed agency is the case. Formalising that inference is the work I’m calling Material Derivation.
Methodological principles need not originate in the researcher’s epistemological commitments. They can be derived from the organisational principles of the phenomenon under investigation. It relocates the origin of method from the discipline’s inherited repertoire to the material organisation of the object. Unlike Ingold’s correspondence, which distrusts formal procedure, Material Derivation seeks a procedure that remains transmissible while still being derived from material organisation rather than imposed upon it.
The operation is a chain, each link entailed by the previous one: the withdrawn object, which never exhausts itself in any relation; material expression, what withdrawal nonetheless offers; encounter, a sensible and mutually constituting meeting rather than neutral observation; method, crystallised as those dispositions harden, provisionally, into principle; and pedagogy, since a method is only a method if it can be taught, and therefore criticised. Withdrawn object, material expression, encounter, method, pedagogy.
The word crystallised carries the speculative character of the interpretation it names, and that is where the method is most exposed. That speculation is disciplined — confined to what the shell describly does, laid out so a reader can locate the step they’d contest, and revisable rather than final. So constrained, speculation is not the abandonment of empirical rigour but the form rigour takes when the object withdraws.
The derivation must be performed on the shell itself, or the principle is empty. At the mantle margin the shell is accreted band by band, each increment precipitated from seawater and recording local conditions, on no central template: knowing must be distributed and local, none permitted to speak for the whole. In cementation, dispersed bioclasts are bound into limestone by dissolution and reprecipitation, then uplifted, eroded, and dispersed again: coherence is a provisional, reversible achievement, never consensus or stable synthesis. In the vacancy chain, living shells pass between hermit crabs in size-ordered sequence, each occupant rewriting the shell’s use without erasing the last: claims must be re-occupiable, taken up and used otherwise. In its growth bands and its slow diagenesis, a mollusc’s season and the limestone’s ten million years occupy one object with no common clock: heterogeneous durations must be held together without synchronisation. And in the periostracum — the organic skin secreted first and exposed last, sealing the shell against the seawater that would otherwise dissolve it — the shell persists only as a boundary held against its own medium: there is no neutral standpoint, and the method is the skin that holds a phenomenon apart from the medium that both makes and unmakes it. These are not metaphors laid over a method; they are ascertainable features of the shell’s material organisation, each a constraint on any method adequate to it. The difference from analogy is precisely the mechanism, which analogy omits and derivation requires.
That specification — collective in authorship, unstable in coherence, relational in constitution, but never merely consensual or plural — has one clear precedent outside the discipline: the renga, the linked-verse form whose coherence is local and whose authorship is collective, each stanza binding to its predecessor and then turning away.
The procedure staged these constraints in sequence. Each participant received five tiles; in groups of five, the twenty-five were assembled one at a time, each placement negotiated into the emerging figure, altering the field the next placement would enter. When a group reached a configuration it could hold, they composed one sentence together — the proposition that resonated most — and speculatively named what it defined. Then one tile was moved, and the whole re-stabilised into a second sentence. Finally, each participant added a tile of their own. Five stages, five of the shell’s own principles, made literal: coherence achieved not given, local crystallisation, interpretation without a sovereign referent, re-occupation, and distributed authorship.
What the apparatus produced can be read from the relational analysis… Three results hold. First, the rule that each tile seek another category was borne out in the chains themselves… “hermit crabs negotiating shell vacancies” set against “while life leaves traces in mineral form,” carried through “traces of past lives” and on into “persisting across multimillion-year geological transformations.” … Second, the speculative naming produced concepts that appear on no tile and in no line of the original article — prosperity, love, reincarnation, a “non-anthropocentric potential community,” a “rethinking of our relationship with nature.” … knowledge was generated in the assembly rather than recovered from it. Third, the single moved tile… the two stabilisations of the same materials were not variants of one figure but distinct coherences…
A method derived from a phenomenon’s organisation induced that organisation in the people who used it.
For the duration of the workshop, the shell’s organisational logic migrated into a temporary social ecology. A method derived from a phenomenon’s organisation induced that organisation in the people who used it. Here the argument rejoins the panel. Studio anthropology has established that making can become thinking — that a concept can be worked with the hands, that a prototype can be an argument. Material Derivation proposes a further step: making can become methodological inference. The artefact is not only a means of thinking a concept but a means of determining how that concept demands to be known; the studio becomes the site at which a phenomenon specifies its own method. This supplies studio anthropology with a criterion it has lacked — a way to distinguish a field device that corresponds to its object from one that merely evokes it. The distinction is between derivation and resonance.
If methods can be derived from the organisation of phenomena, method becomes ecological. Methodology becomes more-than-human. And the object moves from the far end of the epistemic relation to its near end, becoming a participant in the specification of the method through which it is known. The limit of that claim should be stated plainly, since it invites misreading. The shell does not intend, choose, or speak. It possesses a material organisation determinate enough to constrain a method’s form — and, as the workshop showed, determinate enough to reorganise the people who followed it. Anthropology has long known that its methods enact the realities they describe. Material Derivation proposes that we grant the realities we study a disciplined part in enacting the methods. The seashell is only the first place the proposal has been made to hold.
