Abstract
A Presence-Only Artifact is an object whose surviving evidential profile is exhausted, at the level of recoverable content, by its own physical existence. It carries no inscription, no maker's mark, no surviving recorded purpose, and no stable descendant tradition capable of explaining what it was for. The object still testifies, but it testifies asymmetrically: it can establish that a structured practice existed, while permanently withholding the specific content of that practice.
This manuscript names the positive architecture behind that asymmetry as concept carriage. A concept carrier is not a message, a code, or a decodable sign system. It is an object whose form stabilizes a relation — among hand, holder, role, attention, restraint, recognition, or procedure — that can be transmitted in practice without being recoverable from the surface alone. Mnemonic use may be one mechanism of such carriage, but it is not the governing category.
The manuscript argues that the Gallo-Roman dodecahedron is the calibration case for this category. The dodecahedron is physically real, multiply attested, unsigned and uninscribed, absent from the documentary record, and resistant to functional convergence after centuries of inquiry. These traits do not merely make it mysterious. They define a particular kind of archaeological closure: closure at the level of function, combined with continuing openness at the level of structure. The argument proceeds by constraint rather than by solution. It eliminates broad classes of proposed function, maps the residual function-space that survives, and explains how an embodied concept carrier can decay from living practice into presence-only testimony. The strongest supported reading of the dodecahedron's generative condition is structural: the object is Rome-dependent rather than Rome-surviving, whose known lifespan sits entirely within the Roman provincial period. A narrower provincial-resistance reading survives as one strongly constrained conjecture inside the residual function-space, marked as such.
Three clarifications govern what follows. First, the transmission model does not assert continuity. The original practice is stabilized in the carrier's form, but what reaches a later receiver is not guaranteed to preserve it. The model is stated in two parts: from the maker's side, Originating practice/concept → Carrier → ?; from the receiver's side, Carrier + Receiver → A → B. B may stand in lineage with the original concept, may be a transformation of it, or may arise as a new concept under encounter — but in every case B arises within a field whose form constraints were imposed by the originating practice, and those constraints rule out arbitrary completion. The carrier does not certify B's conceptual pedigree. It does certify that B was not blurted. Second, the dodecahedron does not materially encode a recoverable message; the relevant fact about it is not what was carried but that carriage occurred. Third, the dodecahedron is not the source of the manuscript's governing question. It is the severe test case for it. The governing question is therefore not simply, "What was the dodecahedron?" It is: when I intervene with understanding, am I completing a process — or ending it?
Orientation Note: What This Manuscript Is
This document is one part of a larger project. The methodological framework that defines Presence-Only Artifacts and the constraint chain used here is given in Constraint and Closure. The comparative testing of candidate objects against the same intake gates is given in the NOTDOD catalog, which currently records approximately seven hundred entries, none of which has survived the full intake and evaluation sequence required by the third Standard Candle (Persistent Non-Resolution) to become a second calibration case. The normative argument — why epistemic foreclosure on objects of this kind constitutes an ethical question rather than only an interpretive preference — is given in Understanding as Ethical Intervention. Within that project, the Gallo-Roman dodecahedron functions as the current clean material instantiation of the third Standard Candle. It is not introduced here as a puzzle awaiting solution. It is introduced as a calibrator: the cleanest available example of what a particular kind of evidentiary closure looks like in a real object. This manuscript is about concept carriage and constrained emergence. It is not an argument about ancient bronze that happens to use a framework. The reading developed in Section 13 — that the dodecahedron is Rome-dependent rather than Rome-surviving — is the best-fitting structural account inside the residual function-space. It is not the manuscript's claim about what the object was. The narrower provincial-resistance conjecture survives only as one subtype inside that space, and is marked as such. A reader who reads this manuscript only as an argument about ancient bronze has not yet engaged the larger argument it serves.
A companion document, B Was Not Lost, develops the specific argument that the governing question — B in the framework — was not lost with the transmission chain but is reachable from A's own instantiation. The argument identifies derivation from instantiation as a fourth relationship between a modern B and the original concept, distinct from the three relationships named in Section 6. B Was Not Lost should be read alongside this manuscript and Understanding as Ethical Intervention.
1. The Framework Question
The Presence-Only Artifact classification begins with a narrower question than the ordinary archaeological question of use. Instead of asking, "What was this object for?" it asks which classes of embodied object can decay into presence-only status under known human transmission failure modes. That shift matters. A Presence-Only Artifact is not an object born meaningless. It is an object that once participated in a living system of recognition, use, position, inheritance, or embodied practice. It becomes presence-only when the system that made the object intelligible dies while the object itself remains physically recognizable.
Presence-only status is therefore developmental. It is the terminal condition of a failed transmission chain. The classification is also recognized epistemologically, through encounter: an analyst identifies an object as presence-only by meeting it and finding the recovery channels exhausted. The object does not ontologically become a Presence-Only Artifact only when seen. Its presence-only condition is a fact about its surviving evidential profile, and the encounter discovers that fact rather than creating it. The Gallo-Roman dodecahedron is the calibration case because it satisfies the conditions of this developmental journey with unusual completeness. It is a highly distinctive object class, known in enough examples to rule out accident or singular eccentricity, yet unaccompanied by any surviving account of its purpose. Its form is precise enough to require explanation, but not self-explaining enough to supply one.
The normative argument behind the governing question — why epistemic foreclosure of this kind constitutes an ethical issue rather than merely an interpretive preference — is developed in full in Understanding as Ethical Intervention. That manuscript earns the question through three irreducible calibration conditions (Existence/Standpoint, Irreversibility/Death, and Persistent Non-Resolution), an applicability constraint specifying exactly when the Question of Action is live, and a distinction between completion and termination that grounds the ethical difference between disciplined openness and premature closure. A reader who encounters the governing question here as a structural anchor is directed to that manuscript for the underlying argument. This manuscript takes the argument as established and treats the Gallo-Roman dodecahedron as its severe test case.
2. Why the Category Is Necessary
Archaeological interpretation often assumes that every manufactured object is, in principle, recoverable by sufficient evidence. When an object resists explanation, the field generates hypotheses and continues generating them indefinitely. Non-resolution is treated as temporary: the right find, the right experiment, or the right analogy is presumed to be waiting somewhere.
A Presence-Only Artifact challenges that assumption. It names cases in which the ordinary verification channels have failed together. Texts do not mention the object. Images do not depict it in use. Find contexts do not converge. Functional descendants do not preserve the procedure. Wear evidence, if present, constrains handling but does not name purpose. The result is not merely ignorance; it is a structured evidentiary condition.
The category is necessary because "unknown object" is too broad. Many unknown objects are merely under-studied, poorly preserved, fragmentary, or awaiting better context. A Presence-Only Artifact is different. It is intact enough, repeated enough, and structured enough to show that something was made deliberately, while the channels that would translate that structure into function are absent or exhausted.
3. Definition and Intake Criteria
For an object to qualify as a Presence-Only Artifact, it must satisfy a constraint chain. The constraints are eliminative. Each removes a class of ordinary explanations; the classification becomes available only when all constraints hold together. Presence-only therefore describes a post-use evidentiary state, not a simple original design category. The object was once completed by a human practice, role, sequence, or context. It becomes presence-only when the surviving artifact no longer preserves enough of that completing ecology to reconstruct which intention, use, or phase made it legible.
Physical reality and manufacture: the object is not legendary, theoretical, natural, or merely decorative debris. Someone made it on purpose. Multiple attestation: enough examples exist to rule out accident, isolated whim, forgery, or purely local idiosyncrasy. Categorical distinctness: the object is recognizable as a class without requiring inscription or external explanation. Unsigned and uninscribed corpus: across known examples, the object lacks text, numerals, maker's marks, or stable symbolic labeling that would carry the function directly. Documentary absence: the literate culture surrounding the object left no known account of it, despite otherwise extensive documentation of the relevant world. Resistance to convergence: many proposed functions remain mutually incompatible, and no proposal has achieved durable consensus. Failure of ordinary recovery channels: contexts, depictions, descendants, and use-wear constrain the object without resolving it.
No single criterion is sufficient. An unsigned tool with a known function is simply an unsigned tool. A mysterious unique object may be an anomaly. An object absent from texts but obvious in use is merely undocumented. Presence-only status requires the combined profile: durable object, recognizable form, lost specific meaning, and no surviving verification pathway adequate to recover that meaning.
One further property follows from the others, and will return as a working diagnostic throughout the manuscript. A Presence-Only Artifact is over-completable. Once its original interpretive ecology has failed, it accepts many secondary assignments — candle holder, gauge, knitting form, gaming piece, ritual object, and so on — without uniquely confirming any. The proliferation of plausible-sounding completions is not noise around the truth; it is a structural property of objects whose completing rule has been lost.
The intake gates also constrain conceptual emergence. They do not merely sort objects; they prevent arbitrary completions from being treated as live readings of the object. A completion that the constraint chain eliminates is not a weak candidate inside the residual space. It is removed before the residual space is entered at all. This point is developed in Sections 5, 7, and 23.
This does not rule out design intention. An object may be designed for primary use, secondary reuse, future recognition, restricted handling, symbolic preservation, or more than one phase of life. A modern jelly jar can begin as packaging, become a household drinking glass, and survive decades after the jelly, label, brand, and purchase context have vanished. The later glass is real; the original context is no longer recoverable from the object alone. Presence-only is not the absence of design. It is the survival of design after its completing context has failed. Objects often have biographies rather than single functions, and the dodecahedron may have passed through several meaningful states before arriving at the condition in which we encounter it.
4. The Gallo-Roman Dodecahedron as Calibration Case
The Gallo-Roman dodecahedron is a small hollow copper-alloy object, usually bronze, formed as a regular dodecahedron: twelve pentagonal faces, a circular aperture in each face, and knobs at the vertices. The apertures vary in diameter across faces and across specimens, and many examples have concentric ring decoration around the apertures. Within the recognized type, twelve aperture-bearing faces appear to be defining rather than optional; damaged examples and fragments should still be treated cautiously. The objects generally date to the Roman imperial period, especially the second through fourth centuries of the common era, and are found principally in the northwestern provinces of the Roman world rather than the Mediterranean heartland. The current public count is usually given as over 120 or approximately 130 examples, with the Norton Disney example bringing the British total to roughly 33. Counts vary because new finds, fragments, private holdings, and catalog criteria change the corpus. For the framework, the important fact is not the exact number but the order of magnitude: the class is too numerous to dismiss as singular, yet far too sparse for an ordinary household or mass-use object.
Against the intake criteria, the dodecahedron is unusually strong. It is physically real and manufactured, not legendary. It is multiply attested across a broad region. Its form is unmistakably categorical: a hollow twelve-faced object with vertex knobs and differentiated apertures. It is not known from contemporary explanatory texts or depictions. It remains resistant to convergence, with proposed functions including measuring instrument, range-finder, gauge, candle holder, religious object, gaming piece, toy, knitting or cord-making form, astronomical device, standard for pipe sizes, and status or ritual object.
The calibration claim is not that the dodecahedron is the only possible Presence-Only Artifact. It is that it is the cleanest known case. It allows the framework to define the category by constraint, then ask what other objects could survive the same filter.
5. The Falsification Catalog: F minus F-minus
Before F-minus operates, the candidate must satisfy the Constraint and Closure framework's intake constraints. These are prior gates, not part of the F-minus analysis itself. The dodecahedron satisfies all four. It passes Constraint 0: it is a physically real, bounded, manufactured artifact. It passes Constraint 1: it is portable and self-contained, evaluable as a discrete unit. It passes Constraint 2: it bears no inscription, text, script, or encoding system; it is non-communicative in the relevant sense. And it passes Constraint 6: it is a single, recognizable object-type rather than a plural assemblage or hoard. These four gates are what the NOTDOD catalog uses to exclude the communicative artifacts, non-objects, bucket-class assemblages, and symbolic-representational objects that constitute most of that catalog's entries. The F-minus analysis that follows operates within the space cleared by passing those gates.
Let F denote the full function-space of possible uses for a cast bronze object in the northwestern Roman provinces. F includes tools, instruments, ornaments, ritual objects, administrative devices, gaming pieces, household items, professional implements, personal effects, and many hybrid categories. The framework's primary operation is not to guess within F, but to remove the classes that the corpus falsifies. The eliminated set, F-minus, is produced by independent constraint axes: Aperture variation across specimens eliminates functions requiring cross-instrument calibration, standardized measurement, or interchangeable gauge values. Low find density relative to provincial population eliminates ordinary domestic implements, common craft tools, popular games, toys, and mass ornaments. Contextual non-convergence weakens shrine, workshop, military-kit, burial-wealth, tavern, bath, and leisure-context readings unless a future secure pattern appears. Inscription absence eliminates functions whose mechanism depends on written text, numerals, named office, tally marks, certified identity, or readable message content. Geographic specificity weakens imperial-issue, pan-Roman, or Mediterranean administrative explanations. Chronological specificity weakens both long-tradition Celtic heritage-object readings and purely late Roman administrative readings if no precursors or descendants can be shown. Resistance to consensus after repeated hypothesis generation eliminates "ordinary but forgotten" as an easy default.
The defensible claim is that the independent axes eliminate many broad ordinary-use classes and leave a smaller, bounded residual space.
Falsification constrains more than use hypotheses. It also constrains later concept-emergence claims. A function eliminated here — candle-holder, standardized gauge — cannot return as a candidate meaning under any reframing, including a reframing in carrier vocabulary. Once a class is in F-minus, it is not available as a reading of the object's surviving structure. This is the point on which Sections 7 and 23 depend.
The comparative catalog by which this residual-space claim is tested is the NOTDOD corpus. NOTDOD currently records approximately seven hundred candidate objects, drawn from across periods, regions, and material classes, that have been considered as possible second calibration cases. None has yet survived the full intake and evaluation sequence required by the third Standard Candle. The exclusions are not editorial preferences; they are the same intake gates applied without favor.
The strongest formulation is therefore provisional but firm: within the tested NOTDOD corpus, no other candidate has yet survived the full intake and evaluation sequence required by the third Standard Candle. A future find or newly analyzed object could become a challenger, but until a candidate passes the same gates, the Gallo-Roman dodecahedron remains the only clean calibration case.
6. Residual Function-Space R and the Vocabulary of Concept Carriage
F minus F-minus equals R: the residual function-space. R is not the answer. It is the bounded space of candidate functions that survive the constraints. Whatever remains in R must operate within each specimen rather than across standardized specimens; must depend on the holder-object relationship rather than on surface inscription; must plausibly belong to a sparse recipient class; must leave little or no documentary residue; and must make the complete form operationally necessary rather than decorative.
The strongest general family inside R is not the mnemonic-institutional model by itself, but the concept-carrier model. A concept carrier is an object whose form stabilizes a concept, relation, discipline, role, recognition, or practice without encoding a recoverable message on its surface. It carries by requiring the right kind of encounter. In the dodecahedron's case, the carrier architecture is plausibly haptic, oral, embodied, sparse, durable, and socially bounded. Mnemonic encoding remains available inside this family, but it is a mechanism, not the category. Candidate families inside R therefore include concept-carrier architectures using mnemonic sequencing, haptic recognition, role authentication, decision procedures using aperture selection, coordination instruments for inter-node operations whose procedures died with practitioners, and ritual-procedural objects whose theology, discipline, or social content was carried orally. These families should be compared by how well they account for the whole object: bronze durability, hand scale, aperture differentiation, knobs, hollow body, unsigned corpus, sparse distribution, and persistent non-resolution.
The transmission model
Earlier drafts of this manuscript implied a continuity model: an original concept passes through the carrier and is received, after loss, as a recovered equivalent — the same concept, dimmed. This manuscript does not assert that continuity. The carrier does not guarantee descent. The model is therefore stated in two parts.
From the maker's side: Originating practice/concept → Carrier → ?. The original practice is stabilized in the carrier's form, and what reaches a later receiver is not guaranteed to preserve it.
From the receiver's side: Carrier + Receiver → A → B. A receiver who encounters the carrier is delivered into A — the surface question, What is it? — and what arises when A is held open under encounter is B. B may preserve the original conceptual structure, may descend from it in altered form, or may arise independently under encounter. The manuscript does not assert which. It asserts only that the carrier constrains what B can be.
The companion document B Was Not Lost identifies a fourth relationship beyond the three named here: derivation from instantiation. In this relationship B is not recovered through cultural transmission, not arrived at by independent encounter, and not a descent from the original concept — altered or otherwise. It is derived from what A was already demonstrating under sustained scrutiny: the accumulated record of failed closure, each instance of which marks the persistence of the field in which B remains reachable. That document develops this relationship and its implication that B remains reachable from A's own instantiation rather than lost with the transmission chain.
Reconstruction, completion, strongly constrained conjecture
The framework distinguishes three positions an analyst can take with respect to a candidate function inside R, and the manuscript uses this vocabulary consistently from this point onward. A reconstruction is a function uniquely compelled by the constraint field; it is what the framework reserves for cases in which the object's structure forbids the alternatives. A completion is a function merely consistent with the object — physically possible, narratively satisfying, but not constrained from elsewhere. Most popular readings of the dodecahedron are completions in this sense, and Section 23 examines why they multiply. Between these sits a third position: a strongly constrained conjecture — a candidate inside R against which the largest number of independent constraints converge, but for which the convergence falls short of unique compulsion. The framework permits strongly constrained conjectures and considers them legitimate intellectual work. It does not permit them to slide into reconstruction.
The standard of unique compulsion that underwrites this vocabulary is itself stipulative. The framework adopts it not because alternative epistemic standards are illegitimate, but because the cost of false closure in this class of objects is asymmetrically high. To complete an object whose completing rule has been lost is to overwrite the very ambiguity the object preserves. The standard is high by design.
The vocabulary also clarifies the relationship between the framework's categorical gates and the graded reasoning that takes place between them. The intake criteria of Section 3, the falsification axes of Section 5, and the formal properties of Section 21 are categorical. The work of comparing candidates inside R is graded by nature. Strongly constrained conjectures are ranked against one another by how many independent constraints they satisfy and how tightly they fit. That ranking is comparative and ordinal rather than probabilistic in any formal sense, but it is not binary.
Under this vocabulary, the strongest version of the theory treats the dodecahedron as a concept carrier whose mnemonic, haptic, ritual, or institutional mechanism remains unrecovered. This model explains why the object's form matters, why inscription would be absent, why distribution might be sparse, why trained handling would matter, and why the disappearance of the host practice would produce lasting silence. But it remains a conjecture inside R.
7. Second-Order Constraints: What the Form Makes Available
R is not homogeneous. Once broad function classes are eliminated, the object's form still provides second-order constraints. Each specimen offers several possible feedback channels to its holder, but those channels do not carry equal evidentiary weight: Haptic: apertures of different diameters can be distinguished by touch; knobs, rims, edges, mass, and palm-scale geometry provide grip, rotation, and orientation cues; hole-fit thresholds can create anchors. Visual: aperture size, face orientation, and surface decoration are distinguishable at sight. Kinetic: the object can be turned so any face or axis becomes operationally central, making sequence, return, and reorientation possible. Thermal: bronze is thermally salient — it feels cool, warms in the hand, and holds heat differently from wood, bone, textile, or ceramic — but this channel is not diagnostic without heat alteration, residue, or controlled experimental patterning. Acoustic: bronze examples click or clink during handling, and hollow metal can sound when contacted, but acoustic relevance remains conjectural absent strike-wear, suspension traces, rattling elements, or standardized resonance. Candidates that use multiple disciplined channels are more form-fitted than candidates that use only one.
The aperture pattern sharpens this point. The dodecahedron is not merely a bronze body incidentally pierced by holes; it is a bounded aperture system. The bronze body holds twelve openings in stable geometric relation. A version with twelve equal apertures would make the faces formally present but functionally redundant: one opening repeated twelve times. The known form does something different. Its apertures vary, so the differentiated openings themselves become part of the object's surviving content. The knobs matter within that aperture-system reading. They do not require a single functional explanation, but they keep the aperture-bearing faces from lying flush against a surface, reduce complete face-occlusion when the object is set down, and provide indexable points for grip, rotation, and return. The object therefore presents a set of apertures as available, differentiated, and held together rather than one privileged aperture surrounded by incidental structure.
The carrier criterion
The second-order constraints above can be stated not as a list but as a criterion — a test for whether an object is a concept carrier at all. The criterion has two stages, and the order is essential. The first stage is falsification, already performed in Section 5. A completion that the constraint chain has eliminated — candleholder, standardized gauge — is not a weak candidate inside R. It is not a candidate at all. It never enters the residual space. The carrier criterion does not re-litigate it.
The second stage operates only on the completions that survive falsification. Among those, the distinction is between a blank and a carrier. A blank affords completions that are mutually indifferent to its form: any surviving B is as reachable as any other, and the object's structure does no work in producing any of them. A carrier affords a B-space structured by the object's own form: some B's are reachable, others are foreclosed, and the reachable ones stand in determinate relation to one another. The dodecahedron is a carrier because, for the completions that survive F-minus, its twelve differentiated apertures, knobs, and hollow body are load-bearing in whatever B emerges.
This makes "carrier" partly a grading term — a measure of how far the object's form is load-bearing — rather than a purely categorical one. The form constrains emergence without compelling a single concept.
The disciplined claim is narrower: the surviving object preserves a differentiated aperture set while withholding the rule by which that set was read, handled, sequenced, or completed. In concept-carrier terms, the carrier remains while the carried concept's living discipline has failed.
8. The Blind-Holder Test
The blind-holder test asks whether a proposed function could be performed by a trained holder without sight. This does not mean the historical users were blind. The test isolates whether the object's functional information is materially available through touch and handling, rather than depending on visual markings or external text.
The dodecahedron passes this test as a possible haptic object. Its apertures can be distinguished by fingertip and may cross bodily fit thresholds; its vertex knobs allow orientation; its hollow geometry and opposing openings create internal relationships that can be explored tactually. That matters because a function relying on embodied knowledge could survive without inscription and could vanish without leaving textual residue.
The test does not identify the function. It stratifies R. It favors models in which the object is an active participant in trained handling, rather than a passive symbol or a tool whose working values must be standardized across the corpus.
9. Use-Wear, Haptics, and Replica Discipline
Use-wear and haptics must be separated. Use-wear belongs to the surviving corpus: polish, abrasion, deformation, residue, corrosion patterning, impact damage, heat alteration, suspension wear, or localized stress observed on actual dodecahedra. Haptics belongs to controlled handling: what faithful reproductions reveal about grip, balance, rotation, indexing, aperture access, visual alignment, and repeated manipulation. The first is historical trace evidence. The second is affordance evidence.
Corpus use-wear disciplines reconstruction. No proposed function should be accepted unless it predicts a trace pattern that appears in the corpus or plausibly explains why its trace would not survive. Candle-holder claims should predict heat alteration, soot, wax, residue, or standardized upright stability. Textile claims should predict aperture abrasion, rim smoothing, or fiber-related polish. Suspension claims should predict localized wear. An explanation that predicts no distinctive trace remains a low-discipline completion.
Replica handling disciplines imagination. Faithful lost-wax bronze reproductions cannot prove ancient use, but they can show which interactions are mechanically natural, awkward, unstable, or salient. The strongest replica observation so far is aperture-threshold haptics. In the author's handling of two faithful lost-wax bronze reproductions of the same nominal type, made by the same craftsman within a short interval, the objects appear visually similar while differing sharply in hand. In one example, a particular aperture accepts the handler's little finger and becomes a natural anchor. In the other, the same bodily relationship fails. The distinction is immediate because the aperture crosses a physical threshold: it either admits the finger or it does not. The hand does not infer the difference; it encounters it.
Figure 1. Replica handling as haptic evidence. A hole that accepts the little finger creates a repeatable anchor; a visually similar object can fail to produce the same anchor. This demonstrates a body-indexed threshold, not an ancient function.
This observation needs a precise epistemic frame. What controlled handling demonstrates is what the carrier does now, in a present encounter — not what it did in an ancient one. The replica result belongs to the Carrier + Receiver → A → B side of the transmission model, where the receiver is modern. It tells us the carrier still carries. It does not tell us that B equals the original practice.
With that frame in place, the observation does real work. It establishes that aperture variation is not merely visual or decorative. It can produce object-specific orientation, recognition, and home positions. A handler may learn a particular dodecahedron the way one learns a familiar tool: not by naming every feature, but by recognizing its tactile signature.
This helps explain why aperture variation is stronger evidence than aperture presence alone. One hole can be adapted to many uses; twelve identical holes would still leave the faces largely interchangeable. Twelve differentiated holes create a field of thresholds. Each face can have a different tactile, visual, and sequential value. If a use required moving from hole to hole in sequence — as in a story, mnemonic, ritualized handling, route, genealogy, oath, lesson, or role memory — orientation would matter. Repeated handling suggests that a dodecahedron can develop such a home position.
The present evidentiary state is therefore best described as affordance-dense but procedurally underdetermined. The dodecahedron preserves the feel of a handled object without preserving the instructions for completing the handling. Replica handling disciplines imagination; corpus wear disciplines reconstruction.
10. Concept Carriage and Embodied Transmission
A clarification of evidentiary level governs what follows. The Constraint and Closure framework's Object Primacy requirement holds that valid claims must originate from intrinsic properties: geometry, material, repeatability, internal variation. The concept-carriage model and the three-phase transmission structure belong to the external narrative class. They are not reconstructions in the sense Constraint and Closure reserves for claims uniquely compelled by the surviving object. They are frameworks for understanding how presence-only decay could have occurred — mechanisms consistent with what the falsification axes leave in R, proposed as strongly constrained conjectures rather than derivations from intrinsic properties alone. The constraint-chain analysis establishes what the object rules out; the transmission model proposes a mechanism consistent with what remains. The two operate sequentially: falsification first, conjecture second, and the latter is always answerable to the former.
Objects can carry knowledge in different ways. Some carry it on the surface: labels, measurements, diagrams, signs. Some carry it in standardized form: a coin, weight, or gauge works because its values are shared across a system. Some carry it in the holder-object relationship: the object becomes meaningful only to a trained body within a living transmission chain. The dodecahedron is best approached through the third architecture, and this manuscript names that architecture concept carriage. A concept carrier does not contain a recoverable message. It may stabilize a relation, procedure, office, restraint, recognition, or way of attending. Message is recoverable content; concept carriage is lost competence made materially durable. The dodecahedron is therefore less like an undecoded text than a surviving carrier whose trained completing practice has failed.
It is important to be exact about what carriage does and does not assert. In a living practice — Phase One in the model of Section 15 — the carrier is completed: the right holder encounters the object inside the right practice, and the concept is activated in the hand of one prepared to receive it. The carrier-side of the transmission model is Originating practice/concept → Carrier → ?: the carrier does not guarantee that what reaches a later receiver is the original practice. Three outcomes must be kept distinct. A later B may stand in lineage with the original — a genuine, if altered, descent. A later B may be a transformation — descent so altered that calling it the same concept misleads. Or a later B may be an independent regeneration — a concept that arises under encounter with no descent from the original at all. The carrier does not record which of these has occurred. It constrains the field in which B arises; it does not certify B's pedigree.
These three outcomes — lineage, transformation, independent regeneration — are ethically equivalent under the interval claim developed in Understanding as Ethical Intervention. Whether a modern encounter produces a B that descends from the original concept, transforms it, or arises as a new concept under encounter, the interval itself is not changed by that outcome. In every case B arises within a field whose form constraints were imposed by the originating practice — those constraints rule out arbitrary completion and certify that B was not blurted. The ethical case for keeping the object's status open concerns the field of possible encounter, not the pedigree of any particular B that arises within it.
Independent regeneration is in fact evidence that the carrier is still carrying: it demonstrates that the form continues to deliver receivers into a bounded, constrained field, even though the field's content is not transmitted but generated. An object capable of independent regeneration has not failed as a carrier. It has demonstrated the one thing the framework can verify about Phase Three: that encounter still produces a determinate field, and that field still warrants protection from premature closure.
Oral-embodied transmission is one likely means by which such a carrier works. It is not inferior memory. Literate cultures often underestimate how much complexity oral cultures can preserve when knowledge is repeated, socially reinforced, embodied, and ritually or practically necessary. The relevant weakness is not cognitive capacity; it is failure mode. When the performance community breaks, disperses, assimilates, or stops needing the practice, there may be no recoverable written residue. The object remains, but the competence that completed it is gone.
Roman mediation is not an incidental problem in this case; it is part of the theoretical structure. If a native or provincial practice was transmitted orally, socially, ritually, or procedurally rather than through Roman written administration, then surviving access to it would predictably be indirect, hostile, partial, or absent. This does not license free speculation about druidic, Celtic, or native practice. But it does mean that the absence of native written explanation cannot be treated as decisive evidence against a native knowledge system.
The dodecahedron's unsigned character is therefore compatible with several architectures: security, esoteric restriction, ordinary non-literacy for the practice, anti-documentary institutional habit, socially bounded competence, or a concept that simply did not require marks. The framework should preserve all of those possibilities rather than collapse immediately into clandestine resistance.
11. Distributed Knowledge and Institutional Sparsity
A repeated but sparse object class suggests a social distribution neither fully centralized nor broadly common. If the dodecahedron had been a state instrument, one might expect inscriptions, administrative references, standardization, or concentration around official contexts. If it had been an ordinary household object, one might expect far more finds and a wider distribution. The observed profile points toward a smaller recipient class.
The strongest institutional reading is sparse-redundant distribution: enough nodes to reproduce the form across geography, but few enough that each specimen may correspond to a specific position, role, lineage, status, or local function rather than an unrestricted household possession. The safer claim is one object per unknown position within a practice community. The holder may change; the object remains the regalia, working instrument, concept carrier, or recognition device of that position — with mnemonic framing as one possible mechanism among these rather than the governing one.
This reading also explains why multiple examples at one site, if confirmed, need not imply ordinary abundance. They may reflect succession, deposition over generations, hub-status sites, or exceptional local histories.
12. Distribution: Rivers, Native Continuity, and Roman Pressure
The drafts propose that the dodecahedron distribution may follow waterways and native-continuity landscapes more closely than formal Roman roads or purely Roman-imposition settlements. The revised claim must be narrower. Proximity to water is not, by itself, diagnostic. Human settlement, Roman military occupation, agriculture, industry, shrines, villas, roads, ports, and cemeteries all tend to leave archaeological traces near water because people need water in order to live, travel, farm, manufacture, and organize territory.
The relevant claim is comparative and hydrographic: among findspots with reasonably reliable provenance, do the objects show a recurring relationship to route-bearing water contexts — navigable corridors, confluences, estuarine access, river bends, headwaters, portage-like transitions, and basin-to-basin corridors — at a rate or in a configuration that differs from comparable Roman-period sites where no dodecahedra have been found? The detail of that geographic coding is taken up in the Appendix.
A dodecahedron not found near a waterway is weak negative evidence, not decisive disproof. These objects are portable, durable, and archaeologically mobile. After roughly eighteen centuries, a recorded findspot may reflect plough displacement, erosion, river-course change, reuse, hoarding, antiquarian collection, construction disturbance, spoil movement, or imprecise reporting rather than original functional placement.
The most defensible position is a middle claim. The dodecahedron may have moved through provincial river-linked landscapes and local-elite or native-continuity networks generated under Roman pressure. Rome may have supplied infrastructure, stress, metal economies, and frontier conditions, while the object's operational meaning may have remained socially bounded and nondocumentary. The pattern is not "anti-Roman." It is provincial, contact-zone, and potentially sub-imperial. Geography can strengthen or weaken a proposed function only after the mechanical and social affordances of the object have been independently constrained.
13. The Generative Condition: Rome-Dependent, Not Rome-Surviving
A Presence-Only Artifact does not necessarily arise from heritage being preserved intact. It may arise when an existing population faces new pressure and invents a material response. For the dodecahedron, the required ingredients would include skilled metalworking, a need for non-documentary or restricted transmission, a sparse institutional network, and a durable form capable of surviving the failure of that network. Under this reading, Rome is not the maker but the occasion. The object is not simply Celtic survival and not simply Roman administration. It is a provincial artifact produced by the contact zone: native knowledge, imperial pressure, local metalworking, and institutional vulnerability converging into a form that outlived the practice that required it.
The strongest supported claim about that generative condition is structural rather than psychological. Production and use of the dodecahedron appear to co-vary with Roman provincial conditions and to fail to survive their withdrawal. The class appears during Roman provincial consolidation and fades when Roman provincial administration withdraws — the behavior of an instrument bound to the conditions that produced it, not of an heirloom that would outlast occupation. This is the sense in which the dodecahedron is Rome-dependent rather than Rome-surviving. The claim is about co-variation and durability. It does not, by itself, assert a motive.
A narrower reading is available inside R, and must be marked precisely as one subtype rather than as the leading claim. One candidate that fits these constraints is provincial resistance — not resistance as armed revolt, but resistance as the preservation of a competence the occupier was not allowed to see, a way for locals to deal with Rome from outside Rome's categories, kept off Rome's books, and abandoned when the conditions that required it ended. This resistance reading is a strongly constrained conjecture: a candidate inside R against which a number of independent constraints converge, but for which the convergence falls short of unique compulsion. It is one subtype of the Rome-dependent structural account. It is not that account's foundation, and it must not be cited as a reconstruction or read as the manuscript's claim about what the dodecahedron was.
The structural claim — covariation and non-survival — is what the evidence currently favors. Withholding from Rome is a candidate, not a finding.
Three claims must be kept separate. First, the surviving Roman record is silent: no known ancient text, inscription, manual, administrative reference, military account, inventory, or later explanation identifies the dodecahedron's function. Second, Roman knowledge cannot be ruled out at the individual level. Third, intentional secrecy at the class level is plausible but not demonstrated by silence alone. The stronger and safer claim is that the object was not publicly legible within Roman explanatory systems.
That the transmission chain did not survive is what the surviving record supports. That B was lost is a different and stronger claim. Non-transmission is not loss. The companion document B Was Not Lost develops this distinction.
14. The Family Clock: Decay at Household Scale
The family clock should not be mistaken for a romantic theory of inheritance. It argues the opposite: households, family lines, and small workshops preserve objects more easily than they preserve the names, procedures, relationships, and stories attached to them. Inheritance is not the preservation of meaning. Often it is the preservation of matter after meaning has already begun to fail.
In a literate, bureaucratic, photograph-saturated culture, many people cannot name their great-great-grandparents, identify the faces in unlabeled family photographs, explain the use of inherited tools, or reconstruct the work, language, and obligations of relatives only a few generations removed. That failure is not ignorance in a moral sense. It is what memory does when it is not anchored by writing, ritual, active use, or institutional repetition.
The family clock names this decay curve. A bracelet, photograph, tool, charm, badge, clock, or keepsake may carry a precise story for one generation, a softened story for the second, fragments for the third, and only inherited importance for the fourth. The object remains; the index thins.
Bronze can cross generations passively. The competence that completes a concept carrier cannot. Unless that competence is stabilized by writing, institutional teaching, repeated practical use, or ritual performance, each generation becomes a bottleneck. One person forgets a distinction. Another preserves only the memorable part. Another substitutes a plausible explanation. Eventually the object survives as grandfather's thing, an old bronze oddity, a curiosity, a charm, a deposit, or junk.
The dodecahedron is especially vulnerable to this household-level decay because it lacks an ordinary function anchor. A sword remains a sword after its owner is forgotten. A coin remains recognizable as money. A fibula remains recognizable as a fastening device. The dodecahedron carries no such stabilizing category on its surface. Once the trained completing relation faded in any one family line, nothing obvious remained to keep that specimen socially legible inside the household that held it.
The family clock therefore does not require a single dramatic collapse event. Silence can be produced household by household, in unsynchronized small losses spread across generations. A tool becomes a keepsake; a keepsake becomes an heirloom; an heirloom becomes a curiosity; a curiosity becomes archaeological presence.