The Celtic Undercurrents of Bath

Native religion in rebellion (Defixiones, Part 9)

I’ve detailed in this series how magic spread from the Ancient Near East (ANE) right across Europe and eventually to Britain, at the farthest northeast edge of the Roman Empire. How this occurred in these islands—likely similar to other regions—is related by Cameron Moffett, curator of collections at English Heritage:¹

The Romans brought with them both literacy and this extensive material culture, which was more substantial than what had existed in Britain before. And it’s usually in all this new stuff, which was spread across most of mainland Britain by the mechanism of a newly introduced market economy, that we see the evidence of magic.

But some of the specific elements of native beliefs, also in evidence generally in the Celtic world and specifically at Aquae Sulis (modern Bath) are worth examining further.

In fact, there were certain similarities in Celtic and Roman practices that likely made the adoption of some systems of the latter so quick to catch on over and above the elements Moffett mentions. This also muddies the situation and makes it difficult to untangle which is which. For example, like the Romans, the Celts had a reverence for springs and other watery spots.

The Gauls, one of the main groups of Continental Celts, established a shrine at the source of the Seine near modern Dijon in the second or first century BCE, prior to Roman conquest, and another at the spring of Chamalières, the source of the Rhône, near modern Clermont-Ferrand. The former seems to have been consecrated to the goddess Sequana, the patron goddess of the Seine, and indeed the river’s name derives from hers. She is known for her mischievous duck familiars. The latter was to Maponos, meaning “great son”, a god of youth—and likely a trickster himself—who was syncretized with Apollo after the arrival of the Romans.

In both locations, there is evidence of pre-Roman construction as well as the deposition of wooden objects, which are apparently votives. Similar to Aquae Sulis, the Romans, as well as the Romano-Gauls worshiped syncretized versions of the native gods with the deposition of a large array of items, including defixiones (lead curse tablets).

Indeed, disentangling the Roman votives from those that predate their influence becomes quite difficult because of the cross-pollination of some of these traditions. While I think I’ve been able to argue for the ANE as a clear source of cursing traditions, votives, particularly their deposition in bodies of water, is a clearly attested Celtic tradition. So while curse tablets don’t appear before the Roman period, and so we can assume the knowledge of them came with the Romans, we can also see them as a continuation of an ancient Celtic practice of deposition at watery sites.

One noteworthy example of Celtic water deposition is the Battersea Shield. This gorgeous La Tène-style bronze repoussé shield dates from the second–first century BCE and was found during excavation for a previous incarnation of London’s Battersea Bridge in the mid-19th century. The shield is believed to have been deliberately put in the Thames as a votive. This mighty British river was a site where many items of arms and armor were offered in sacrifice in the Bronze and Iron Ages, including other notable finds such as the Wandsworth Shield and the Waterloo Helmet.

The Thames also figures as a locus for divination during Boudica’s doomed uprising against the Romans (ca. 60 CE) when the waters themselves were used as a kind of scrying object. Although Tacitus only mentions it in passing, a vision in the river is given as one of the omens seen by the Britons as fortuitous for the rebellion:²

[…] visamque speciem in aestuario Tamesae subversae coloniae […].

[…] and in the estuary of the Thames had been seen the appearance of an overthrown [Roman] town […].

Other sites were still more important; excavations at Fiskerton, on the Witham, have yielded a rich selection of Iron Age artifacts, including several swords, spearheads, an axe, and a dagger, many of them ritually damaged or destroyed before their deposition in the river. There are several similar sites throughout the British Isles and mainland Europe, such as Llyn Cerrig Bach in Wales, the Lisnacrogher Bog in Ireland, Orton Meadows (on the former course of the Nene) in East Anglia, and the eponymous La Tène on Lake Neuchâtel, Switzerland.

Circling back to Bath, the archaeology is tricky, as the Roman construction overlays and supplants the earlier Celtic structures. But it is generally agreed there was a temple to Sulis sited at modern Bath. Some estimate this could have occurred as much as 10,000 years ago, placing it in the Early Neolithic period, which, to be honest, seems exaggerated, as the Windmill Hill culture only dates to around 3000 BCE. In any case, it seems clear there was a Celtic Iron Age temple to their local deity, Sulis, when the Romans arrived.

Even with all the Roman-period construction, eighteen Late Iron Age coins were found in modernity, hidden in the anaerobic mud of the spring’s reservoir. Given their condition, and barring some unlikely event such as a hoard being dug up and then redeposited, it seems clear they must have been there prior to Roman influence.³

This would seem to invalidate the hypothesis I had previously accepted from Marina Piranomonte that the use of coins as votives was because of the decline in literacy and the ability to inscribe defixiones, but so it goes in science. And perhaps both can be true; at Aquae Sulis the deposition of coins may have returned because of the decline of public epigraphy and in the case of the Fons Annae Perrenae (Piranomonte’s subject) the cross-pollination of an originally Celtic practice might be what’s at work.

Furthermore, the lead pig I mentioned in Part 1 may also have been a votive. One of the original archaeologists surveying the site, Barry Cunliffe, noted it as such.⁴ Its presence is certainly strange, appearing in the temple itself, rather than at some outbuilding where pipes might have been manufactured. Indeed, it is the only such object found on the site, and bears marks appearing to have been made by an axe to ritually damage it prior to deposition.

Another important Celtic tradition is what is known as the cult of the head. Summed up, this cult venerates the head as the source of an individual’s soul, personality, and spiritual potency, and a symbol of the regeneration of life. This is true to such an extent that the physical body is a sometimes disposable element of the complex symbolic structure. Indeed, the cult of the head was a core part of Celtic religious ideology, from the culture’s origins through to its demise, evidenced in its folklore, myth, and art.

While heads on stakes is a well known medieval trope, the message in that context being a warning transgressors will be punished, the same sort of display had an entirely different meaning to the ancient Celts. Classical sources clearly relate—and local vernacular traditions verify—the importance of heads as war trophies, which decorated the exteriors of both dwellings and temples in their villages. Certainly martial prowess is thus shown, but these heads also acted as amulets as well.

One source on the topic is Strabo, who tells us:⁵

[…] βάρβαρον και το ἔκφυλον, ὃ τοῖς προσβόρροις ἔθνεσι παρακολουθεῖ πλεῖστον, το ἀπο τῆς μάχης ἀπιόντας τας κεφαλας τῶν πολεμίων ἐξάπτειν ἐκ τῶν αὐχένων τῶν ἵππων, κομίσαντας δε προσπατταλεύειν τοῖς προπυλαίοις. […] τας δε τῶν ἐνδόξων κεφαλας κεδροῦντες ἐπεδείκνυον τοῖς ξένοις, και οὐδε προς ἰσοστάσιον χρυσον ἀπολυτροῦν ἠξίουν

[T]hey have a barbarous and absurd custom […] of suspending the heads of their enemies from their horses’ necks on their return from battle, and when they have arrived nailing them as a spectacle to their gates. […] The heads of any illustrious persons they embalm with cedar, exhibit them to strangers, and would not sell them for their weight in gold.

Archaeological evidence also appears to back this up, with skulls found in settlements mainly near fortification walls, gates, doorways, etc., just as classical and vernacular traditions suggest. The Celtic homeland areas of central Europe, and in particular the unique temple sanctuaries of southern Provence, such as that at Roquepertuse, have direct and datable archaeological evidence for a head cult making use of votive human skulls. In the case of Roquepertuse, whose temple’s portico featured pillars with cavities for the deposition of skulls, that date is at least third century BCE but possibly even from as early as the sixth century, with the temple’s destruction by the Romans in 124 BCE giving us a clear terminus ante quem.

In Britain, too, finds giving evidence of the head cult are relatively common from the late Iron Age and early Roman period. These include skulls kept as trophies, skulls buried by themselves, and—importantly for our purposes here—skulls found in springs and wells:⁶

[H]uman skulls were frequently offered in ritual contexts at watery places during the Roman period, apparently as a direct continuation of a deeply-rooted native British tradition. One skull found on the site of the Bank of London was found as part of a deliberate filling of an early Roman well, dating from the first to the third century AD, which suggested it was part of a complex foundation ritual. […] The existence of a long-standing tradition of offering skulls to watery places may explain a number of isolated finds in the archaeological record, such as the skull of a young woman […] which was found buried in the lining of a well at a first century settlement in Odell, Bedfordshire. In Brigantia, a well at a Romano-British settlement site at Rothwell near Leeds dating from the fourth or fifth centuries AD yielded a single human skull. […] [?] Merrifield has noted a number of similar instances from Roman London, and another skull from the third century well of a Roman villa at Northwood, Hertfordshire […]. Describing these puzzling finds, he says heads are unlikely to be dropped into wells by accident or as discarded rubbish, and sees significance in the fact that heads are often found as “closing” deposits into wells which previously supplied water for domestic or industrial purposes.

In addition to actual heads, watery contexts for votives symbolic of heads are common. For example, in both the Fontes Sequanae and Chamalières some of the votives I previously mentioned were human heads carved from wood. These seem to date from the pre-Roman period because they show no signs of Mediterranean influence in their style, bearing instead the oval eyes characteristic of Celtic art. The carved jack-o’-lantern of modern Halloween clearly relates to this tradition via the co-opted insular festival of Samhain, even down to the locations in which they are displayed.

We see such symbolism repeatedly in stone heads, including tricephalous and janiform heads, face pots, wooden carvings, masks, and antefixes. One such head is discussed by Professor Anne Ross, thus:⁷

[In the territory of the Belgic Remi tribe] the deity is symbolised by an enormous bearded tricephalos, having a leaf-crown, and usually equated with the classical Mercury. These particular representations would seem to testify to the concept of some autochthonous deity as a head alone, the head sufficing for the total being, the vital part, embued with the power of the whole.

Although Strabo wrote with contempt of the Celtic fascination with severed heads, there is one that appears regularly in the Graeco-Roman tradition as well, even including the apotropaic function: that of Medusa. Also known as a Gorgoneion, the image of this grotesque severed head is a well-known device on armor and shields as well as coins, temple pediments, antefixes, garments, dishes, and weapons. Thus it shared similar ubiquity and longevity to the Celtic head cult, even exceeding it, as it survived well into Christian times and was revived in Renaissance and neoclassical contexts, right down to the present where it appears in the logo of the Versace fashion brand.

The prevalence of the image of the Gorgon’s disembodied head, while of course referring to the Perseus myth, also closely matches the spirit of the Celtic head cult:⁸

It is […] apparent that in her essence, Medusa is a head and nothing more; her potency […] resides in the head […].

If one superimposes the Gorgoneion and the image of the enormous, bearded, disembodied head Ross has given us (minus the triple aspect), it’s hard not to think of one of the more famous images from Aquae Sulis, which she also discusses:⁹

The Gorgon’s head on the shield of Sulis-Minerva in the pediment of the temple is the finest example of the blending of native and classical imagery. The head is male, bearded and moustached, and its ancestry can be traced directly to the human heads which are so prolific on La Tène metalwork. The furrowed brow and two-dimensional features are typical of many examples of Romano-British heads in stone, as is the expression of the face. The convention of the writhing serpents which here spring from the hair and are entwined in the beard and moustache is classical, but the connection of serpents with human heads is found deeply rooted in the native tradition.

Another head emblematic of the site at Bath is that of Sulis-Minerva. This beautiful gilt bronze head evinces Graeco-Roman influence and is believed to have once worn a Corinthian helmet as well. This is generally interpreted as a fragment of a full-body cultic statue, but given the significance of the head in Celtic religious practice I’ve just discussed, I’m not so sure. Obviously there are many factors, but much older finds such as the shields I’ve mentioned are in excellent condition, so the idea that the rest of the statue dissolved in its entirety seems odd. The head isn’t perfect to be sure. There is some pitting on the lower right of the face. But it also shows six layers of gilding, which would have provided additional protection against corrosion and there’s no reason to believe the rest of the statue would not have been similarly gilt. Why then would it not make sense this too was either a disembodied head representing cultic beliefs or even a votive head deposited in the spring?

Certainly Roman religion had some traits in common with that of the Celts, and the interpretatio romana combines the names of their deities, but the Britons didn’t necessarily think of their own gods in this way. Besides Graeco-Roman gods and syncretized ones, the names of distinctly Celtic ones appear in inscriptions from Bath: Nemetona, the Suleviae, Sulis, “the mother goddess”. And even syncretization can be a form of rebellion, as African slaves could secretly worship a native deity such as Ogun, who they recognized in the image of the Christian Saint Peter.

While Romanization was quite thorough in some parts of the Empire, it was less so in Britain. Resistance to the invasion was quite stubborn and prolonged, even though native military tactics were not up to the task. The adoption of Roman customs, too, seems to have been met with little enthusiasm in many parts of the Isles. Rather than building temples in the classical style, Romano-Celtic ones were the norm, and indeed there are many natural sites votive finds attest were sacred, such as groves and springs. These, it is clear, predated Roman influence, and some of them, like that of Sulis at Bath, had structures added to them under Roman rule.

And indeed, there seems to have been a revival of Celtic practices as Roman power waned. For example, already by the late Roman period decapitated burials reemerge, clearly relating to the cult of the head. Many such beliefs continued past the Anglo-Saxon settlement of Britain, even down to its Christianization.

Read Previous Articles in This Series

Part 1: The Curses of Aquae Sulis

Part 2: Malefic Traditional

Part 3: Sympathy for Sauron

Part 4: Bargaining with the Gods

Part 5: Secundina’s Beef

Part 6: More Than Money Can Buy

Part 7: The Punic Curse Trail

Part 8: Hellenism Schmellenism


  1. Episode 93, “Superstition, magic and the Evil Eye in the Roman world”, The English Heritage Podcast, 2020.
  2. Cornelius Tacitus, Annales, 14.32, c. 115–c. 120. I’ve used the Alfred John Church and William Jackson Brodribb translation, 1888.
  3. Barry Cunliffe and Peter Davenport, “The Temple at Bath (Aquae Sulis) in the context of classical temples in the west European provinces”, The Temple Of Sulis Minerva At Bath Vol. I: The Site, 1985.
  4. Barry Cunliffe, Excavations in Bath 1950–1975, 1979.
  5. Strabo (Στράβων), Γεωγραφικά (Geographica), 4.4.5, c. 15 BCE. I’ve used the William Falconer translation, 1903–06.
  6. David Clarke, “The Head Cult: tradition and folklore surrounding the symbol of the severed human head in the British Isles”, 1998.
  7. Anne Ross, “The Human Head In Insular Pagan Celtic Religion”, Proceedings of the Society of Antiquaries of Scotland, 1958.
  8. Jane Ellen Harrison, “The Ker as Gorgon”, Prolegomena to the study of Greek religion, 1903.
  9. Ross, 1958.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: