Translated from french (please notify us of errors)
There exists, in the museums of north-western Europe, an object whose purpose no one knows. Cast in hollow bronze, ranging in size from a fig to an apple depending on the specimen, it presents twelve pentagonal faces each pierced by a circular opening — all of different diameters — and its twenty vertices are adorned with small solid spheres.
Over a hundred have been catalogued, distributed between insular Britain and Pannonia, between Hadrian’s Wall and the south of France. No ancient text mentions them. No fresco, no mosaic depicts one in use. And yet, between the 2nd and 4th centuries of our era, craftsmen took the trouble to cast these objects using the lost-wax method, to polish their exterior faces carefully and leave their interior faces rough, as though the inside was never intended to be seen.
The archaeological community refers to them as the Gallo-Roman dodecahedron or bossed dodecahedron. Nearly three centuries have passed since the first discovery, in 1739 at Aston in Hertfordshire, and the mystery remains complete.
An object without ancestor or descendant
The first paradox of the object is its geographical distribution. Dodecahedra are found exclusively in the north-western provinces of the Empire, across a territory that corresponds more or less to ancient Celtic civilisation: approximately sixty per cent of known specimens come from the Gaulish and Germanic provinces, some twenty per cent from Roman Britain. None has been unearthed in Italy, Spain, Africa, Greece, or the eastern provinces. Until recently, the most easterly was that of Brigetio (today Szőny, Hungary), on the Danubian limes. In 2021, a fragment discovered at Deonica, in central Serbia, pushed this frontier somewhat further south-east: it now constitutes the most southerly and most easterly specimen in the corpus.
This Serbian fragment — studied by Miroslav Vujović and held at the Regional Museum of Jagodina since 2006 — is made of brass (90% copper, 5% zinc). It preserves only three partial faces and a single vertex with its sphere, but the angles between the faces, close to 108 degrees, are sufficient to identify the original form without ambiguity. The measured openings give 15 mm on one face and 25 mm on another; the height as reconstructed by 3D modelling would be approximately 65 mm. Its presence at Deonica is plausibly explained by the proximity of a secondary Roman road linking Aureus Mons to Horreum Margi and by the presumed existence of Roman agricultural estates along that route.
More troubling still are the three fragments from western Slovakia, studied in 2025 by Petra Dragonidesová. Two come from Hurbanovo, one from Chotín — two Germanic sites situated barely 6 km north of the Danube, opposite Brigetio. These fragments, likewise of copper alloy, did not end up there by chance: their state of deliberate fragmentation suggests they were recovered, during a raid or after the destruction of the Roman fort at Iža in 179 of our era, with a view to being remelted as raw material. The Roman object had become barbarian scrap.
What the contexts say… and what they conceal
For the great majority of catalogued specimens, the context of discovery is unknown, the object having passed through the antiquities market. Among the few documented cases, the diversity of situations discourages any hasty conclusion. Dodecahedra have been found in military camps (Feldberg, Zugmantel, Carnuntum), in bathhouses (Arles), near a theatre (Besançon), in richly furnished tombs (Krefeld-Gellep, Bachem), in a 4th-century coin hoard (Saint-Parize-le-Châtel), in the fill of a well, and in riverbeds (Nijmegen, Trier, Zurich). According to the statistical survey drawn up by Michael Guggenberger, more than half of specimens with a known location come from civil contexts — urban or rural — fewer than a fifth from military contexts, and approximately 8.5% from contexts with a plausible sacred connotation.
Among these latter cases, several deserve attention. The Pfofeld specimen (Germany) was uncovered in the immediate vicinity of a bronze statuette of Mercury-Hermes-Thoth. A fragment from Lydney (Britain) comes from the sanctuary of the local god Nodens. A fragment from the southern shore of the Severn estuary formed part of a deposit identified as originating from a temple — perhaps that of Diana at Gloucester. And a specimen found north of Paris was discovered one metre from a goddess statuette — Juno, perhaps.
Among finds in an aquatic context, the Newhall specimen, near Harlow (Essex), merits particular attention: it was unearthed between 2019 and 2022 in a pond fed by a natural spring, at the heart of a corridor-plan Roman villa of about ten rooms. The pond had collected knives, tools, coins, and this dodecahedron of copper alloy, 50 mm high and nearly complete. Ordinary debris, or a deliberate deposit at a watery place? Post-excavation analysis is still under way, as its author James Alexander notes, who simply observes that the deposition of iron objects in aquatic contexts is a well-documented phenomenon in the Iron Age and the Roman period.
The best-documented case remains that of Jublains (Mayenne), published in 2008 by Guillier, Delage, and Besombes: the object comes from the destruction layer of a small building on a sub-floor, dated to the first half of the 3rd century. It measures 59 mm in height, 74 mm in maximum diameter, weighs 81 grams, and presents ten faces with circular openings (diameters from 10.5 to 22 mm) and two opposing faces with oval apertures (21×26 mm) lacking concentric circles — which makes it, in Guggenberger’s terminology, a type 2a specimen, the most common. It is one of the very few dodecahedra for which a reliable stratigraphic context has been published.
The indescribable object
One begins to understand, faced with this accumulation of disparate cases, why interpretations have proliferated. Robert Nouwen catalogued an impressive list in his 1994 survey: surveying instruments, aspersoria, lamps, jewellers’ gauges, coin-blank callipers, toys or cup-and-balls, sceptre pommels, gaming dice, multi-gauge candlesticks, masterpieces of craft skill… In 2025, Guggenberger and Leach bring to approximately fifty the number of distinct theories that have been formulated since 1739.
The hypothesis of the Roman rangefinder or theodolite, revived by the engineer Amelia Sparavigna in 2012, deserves mention… only to be promptly set aside. The idea is geometrically appealing: looking through two opposing openings of different diameters would allow, by visual coincidence, the measurement of a distance to a known target. The difficulty is that this hypothesis had already been proposed by Friedrich Kurzweil in 1957, developed by K. Mauel in 1961, and refuted in what its authors consider definitive fashion by F. H. Thompson as early as 1970 in The Antiquaries Journal — a refutation that Guggenberger and Leach confirm in 2025. The dimensions of the objects are too variable and too poorly standardised for a measuring instrument; no graduation, no numerical indication, no associated tripod has ever been found.
The question therefore remains entirely open. But it is perhaps not without an answer.
Plato in the Gauls
This is the principal contribution of the article that Guggenberger and Leach published in 2025 in The Antiquaries Journal: an attempt to connect archaeology and textual sources more precisely than their predecessors had managed. Their starting point is well known: Plato’s Timaeus (c. 360 BCE), in which the dodecahedron is associated with the structure of the universe as a whole. Having assigned the first four regular solids to the four elements — tetrahedron for fire, octahedron for air, icosahedron for water, cube for earth — Plato adds:
“There was still a fifth construction, which the god used for the whole, embroidering it with figures.”[1]
The four other regular solids correspond to the four elements; the dodecahedron encompasses all.
What the authors add is the articulation between this cosmology and the Platonic notion of hypodoché — the “receptacle of all becoming”, the third ontological genus of the Timaeus, distinct from intelligible forms and sensible things. This receptacle is described as a neutral, malleable matrix, capable of receiving all forms without retaining any of them durably, comparable to “a mass of wax” (Tim. 50b–c). It is space (chôra) and seat (hedra) of all that comes to be. And — this is the crux of the argument — the second term of the Greek word dodekáhedron is precisely hedra, “seat” or “face of a solid”. The dodecahedron is etymologically a seat with twelve faces.
But the central argument of Guggenberger and Leach is a hitherto neglected passage from Iamblichus’s De Vita Pythagorica (c. 245–325 CE). The Neoplatonist philosopher describes therein the manner in which Pythagoras honoured the gods: not in statues of human form, but in divine hidrýmata (ἱδρύματα) — receptacles that:
“embrace all things, provide for all things, and have a nature and a form similar to the whole universe.”[2]
A hollow object, cast in bronze, of quasi-spherical form, symbolising the universe in its totality — the description fits a Gallo-Roman dodecahedron remarkably well, even if Iamblichus does not name it. The authors note that Iamblichus was writing at precisely the time when these objects were being made and used. And in the same passage, a detail that Guggenberger and Leach do not fail to remark, Iamblichus cites the Celts among the peoples from whom Pythagoras is said to have drawn his philosophy — alongside Egyptian priests, Chaldaeans, and the mysteries of Eleusis.
For this argument to hold, the populations who used these dodecahedra — principally in the Tres Galliae and in Britain — must have had access to Pythagorean and Platonic ideas. Now, several ancient sources, from Diodorus Siculus to Julius Caesar by way of Strabo, Ammianus Marcellinus, and Valerius Maximus, explicitly describe the druids as Pythagoreans. Caesar notes that they used the Greek script and taught secretly for twenty years. Valerius Maximus, not without irony, remarks that “what those barbarians in trousers believe is precisely the faith of the Greek Pythagoras himself”.
Pliny the Elder, in his Naturalis historia, offers an unexpected echo of this convergence. He describes a druidic ritual object specific to the Gauls, called urinum: a ball formed by intertwined serpents,
“the size of a medium-sized round apple, covered by a cartilaginous shell bristling with cups similar to the suckers on the arms of an octopus.”[3]
The object is spherical, of animal origin according to Pliny, but its use is clearly cultic: the druids employ it “to win lawsuits and gain access to kings”. One will not claim that the urinum is a dodecahedron. But the coincidence is striking: a globular Gaulish object, its surface covered with regular protuberances, reserved for secret druidic practices, which Pliny himself says he saw with his own eyes without claiming to understand its nature — at the same period, in the same provinces, where bronze spheres of twelve faces and twenty spheres were being manufactured.
The Pythagorean secret and the druidic secret would together explain why no ancient text ever describes the use of the object.
The thesis of Guggenberger and Leach is therefore as follows: the Gallo-Roman dodecahedron would be a theurgic receptacle, a ritual object intended to receive a divine presence or to practise a form of divination by numbers, within the framework of an original synthesis between Pythagorean philosophy and the Romanised druidic tradition. Its presence in tombs, sanctuaries, votive aquatic contexts — and its total absence from the Mediterranean heart of the Empire, where this Celto-Pythagorean synthesis had no reason to exist — accords better with this hypothesis than with any purely utilitarian explanation.
The authors remain cautious nonetheless: “That there was a Pythagorean-Druidic aspect to the Gallo-Roman dodecahedron, though not yet proven, seems plausible.” Archaeology, for the time being, neither confirms nor refutes this. But the next time a dodecahedron is uncovered in a stratified context, we shall know which questions to ask: are there traces of another substance inside? Microscopic signs of wear? If it is broken, was it broken deliberately?
In the meantime, the object continues to rest in its museum display cases, its twelve faces equally silent, its twenty spheres pointing in as many directions — an empty receptacle, or perhaps a receptacle for all that one has imagined placing within it.
Modern Studies
- Guggenberger, M. et Leach, S., “The Gallo-Roman Dodecahedron and the Receptacle of All Becoming”, The Antiquaries Journal, 2025. DOI: 10.1017/S000358152510036X
- Dragonidesová, P., “Mysterium Insolutum. Fragments of Roman Dodecahedrons from Hurbanovo and Chotín”, Raitarivs LXX. Proceedings in Honour of Ján Rajtár, Študijné zvesti AÚ SAV – Supplementum 3, 2025, p. 29–36.
- Alexander, J., “A Roman villa at Newhall, Harlow, Essex”, Archaeology International, 26 (1), 2023, p. 208–214.
- Vujović, M. B., “Fragment of Roman Dodecahedron from Deonica (Serbia)”, Instrumentum, 54, 2021, p. 37–40.
- Guillier, G., Delage, R. et Besombes, P.-A., “Une fouille en bordure des thermes de Jublains (Mayenne): enfin un dodécaèdre en contexte archéologique!”, Revue archéologique de l’Ouest, 25, 2008, p. 269–289.
- Nouwen, R., “Les dodécaèdres gallo-romains ajourés et bouletés. Histoire et problèmes”, Bulletin de l’Institut archéologique liégeois, t. CVI, 1994, p. 85–108.
- Duval, P.-M., “Comment décrire les dodécaèdres gallo-romains, en vue d’une étude comparée”, Gallia, t. 39, fasc. 2, 1981, p. 195–200.
[1] Plato, Timaeus 55c: ἔτι δὲ οὔσης συστάσεως μιᾶς πέμπτης, ἐπὶ τὸ πᾶν ὁ θεὸς αὐτῇ κατεχρήσατο ἐκεῖνο διαζωγραφῶν.
[2] Iamblichus, De Vita Pythagorica 28, 151: οὐ ταῖς ἡμετέραις συνεζευγμένους μορφαῖς, ἀλλὰ τοῖς ἱδρύμασι τοῖς θείοις, πάντα περιέχοντας καὶ πάντων προνοοῦντας καὶ τῷ παντὶ τὴν φύσιν καὶ τὴν μορφὴν ὁμοίαν ἔχοντας. […] καὶ περὶ τοὺς Κελτοὺς δὲ καὶ τὴν Ἰβηρίαν.
[3] Pliny the Elder, Nat. hist. XXIX, 53: vidi equidem id ovum mali orbiculati modici magnitudine, crusta cartilagineis velut acetabulis bracchiorum polypi crebris insigne.
Other articles in English from the Nunc est bibendum blog