The Mabinogion

The Four Branches are the core and the only tales the medieval text itself groups together. Pwyll tells of the prince of Dyfed who exchanges places with a lord of the otherworld, wins Rhiannon, and loses and recovers their son Pryderi. Branwen turns on the marriage of Branwen to the king of Ireland, her humiliation, and a catastrophic war in which the giant Bran is mortally wounded and his severed head continues talking to his companions for eighty years. Manawydan is a strange, quiet tale of enchantment and an emptied land, resolved by a bargain over a pregnant mouse. Math contains the making of a woman out of flowers, her transformation into an owl, and the magician Gwydion and his nephew Lleu.

The native tales are separate compositions. Culhwch and Olwen is the earliest Arthurian prose tale in any language, probably late eleventh century, and pairs an enormous catalogue of Arthur's warriors with a sequence of impossible tasks. The Dream of Rhonabwy is a sardonic dream-vision, the Dream of Macsen Wledig attaches a Welsh legend to a Roman emperor, and Lludd and Llefelys explains three plagues on Britain.

The three romances, Owain, Peredur and Geraint, correspond closely to Chretien de Troyes' Yvain, Perceval and Erec. The relationship is much argued: whether the Welsh versions derive from Chretien, whether Chretien and the Welsh drew independently on shared Breton or Welsh material, or some combination. No settled answer exists, and the tales differ from the French in ways that do not look like straightforward translation.

Neither medieval manuscript calls this collection anything. The word mabinogi appears as a closing formula at the end of each of the Four Branches, and its meaning is itself uncertain, possibly "a tale of youth", possibly matter concerning the god-name Maponos, possibly the material a trainee storyteller learned. At the end of the first branch, in one manuscript, the form appears as mabynnogyon, treating an already-plural word as a singular and pluralising it again. That single occurrence is generally agreed to be a copying mistake.

The mistake became the title. The form was used in print in 1795 by William Owen Pughe, and Lady Charlotte Guest adopted it for her translation of all eleven tales, published in parts from 1838 to 1845 and collected in 1849. Guest was an English ironmaster's wife who learned Welsh, worked with Welsh scholars including Tegid and Thomas Price, and produced a version that carried the tales into English and thence into European Romanticism and, eventually, into a great deal of modern fantasy writing. Her translation is bowdlerised in places and her Welsh has been superseded, but the name she used has proved permanent, and it now denotes a grouping no medieval Welshman made.

The manuscripts themselves are later than the tales. Linguistic evidence puts the Four Branches in the eleventh or twelfth century in something like their present form, and Culhwch and Olwen earlier still. Some fragments survive in Peniarth manuscripts of the mid thirteenth century, but the complete texts depend on the White Book and the Red Book, and Guest worked chiefly from the Red Book, which is held at Jesus College, Oxford.

Several figures behave in ways that look residual. Rhiannon appears on a white horse that cannot be overtaken at a walk and is later made to carry visitors on her back like a beast, which invites comparison with the Gaulish horse goddess Epona and the Irish sovereignty figures. Lleu Llaw Gyffes corresponds by name to the Irish Lugh and the Gaulish Lugus. Bran the Blessed is a giant who wades the Irish Sea and whose head is buried in London to protect Britain. Manawydan corresponds to the Irish sea figure Manannan mac Lir. Nodens, Nudd, Gwyn ap Nudd and Mabon all have counterparts in inscription or in Irish narrative.

The reasonable inference is that some of these characters descend from figures who were once gods in Brittonic religion. That is inference, not record. There is no Welsh text that says so, no Welsh mythological handbook comparable to Snorri's, and no continuous pagan tradition on the Welsh side to compare against. Nineteenth and early twentieth century scholarship built elaborate reconstructions of a lost Welsh pantheon out of these tales; most of that has been abandoned as unsupportable, and the more disciplined modern approach is to read the tales first as narrative art and only then as evidence for anything older.

The caveat matters because the Mabinogion is regularly presented as Celtic scripture or as a pagan survival, and it is neither. The tales were written down by Christians in a thoroughly Christian country, are shaped by contemporary European narrative fashion, contain no ritual, no theology and no cult, and treat magic as a device rather than a belief. Modern Druids and Celtic reconstructionists use them as a primary source, which is a legitimate contemporary religious choice; it is not the same as the texts having been sacred to the people who wrote them.