A historiography of the matter of Britain — where the Arthurian legends came from, what shapes they took, and how the memes mutated across fifteen centuries.
Each tale below is a full sub-site of its own — original translation, character cards, story graph, motif index, a unifying mythograph, and a kid-faithful storybook.
The earliest Arthurian story we have (c. 1100), preserved in the Welsh Mabinogion. The boy born in a pig-run, the giant whose eyebrows are propped on forks, the council of the world's oldest animals, and the hunt for the great boar Twrch Trwyth.
read.mino.mobi/culhwch →The masterpiece of late-14th-century Middle English alliterative verse — the anonymous Pearl Poet, surviving in one manuscript. Full translation across all four Fitts, plus character cards, story graph, motif index, mythograph and a kid-faithful storybook with images.
read.mino.mobi/gawain →The c.1300 Middle English Breton lay — Orpheus and Eurydice rebuilt as a king of Winchester and his queen taken not into Hades but into Faerie, and brought back. Full translation across all six movements, plus characters, story graph, motifs, mythograph and storybook with images.
read.mino.mobi/orfeo →The First Branch of the Mabinogi — a foundational Welsh pre-Arthurian tale of exchanges: Pwyll trades places with Arawn lord of Annwn, meets Rhiannon on the Gorsedd mound and outwits Gwawl with the badger-in-the-bag, and recovers his stolen son Pryderi by way of Teirnon's hand on the foal-stealing claw. The complete 7,600-word Middle Welsh tale from the manuscript tradition, in original English translation alongside the Welsh. Annotation layer to follow.
read.mino.mobi/pwyll →The Lady of the Fountain — one of the Three Welsh Romances and the Brittonic twin of Chrétien's Yvain: a knight of Arthur's court rides to a storm-making fountain, defeats its Black Knight and weds the widowed Lady, breaks the term she set and runs mad in the wilderness, and is redeemed with the help of a grateful lion. The complete Middle Welsh tale from the Red Book of Hergest (Rhys & Evans 1887), in original English translation across all nine movements, with the full annotation layer — characters, story graph, motif index, mythograph and a storybook.
read.mino.mobi/owain →The Second Branch of the Mabinogi — the great Welsh tragedy, and the sister-tale of Pwyll. The giant king Bendigeidfran marries his sister Branwen to Ireland to bind the two islands; his half-brother Efnisien's malice turns the alliance to war; the Cauldron of Rebirth gives back the dead but dumb; and Brân's living severed head feasts the seven survivors for fourscore years. The complete Middle Welsh tale from the Red Book of Hergest (Rhys & Evans 1887), in original English translation across all six movements, with the full annotation layer.
read.mino.mobi/branwen →The Third Branch of the Mabinogi — the sequel to Branwen, and its bright mirror. Two of the seven survivors come home; Manawydan weds Rhiannon and rules Dyfed with Pryderi and Cigfa, until a thunderclap and a fall of mist empty the whole land to a wasteland. Driven to live by their hands, then robbed of Pryderi and Rhiannon by a vanishing caer, Manawydan alone keeps his head — and lifts the entire enchantment with a craftsman's patience, a pregnant mouse on a tiny gallows, and a bargain he will not let go. The only branch where wit, not war, wins. The complete Middle Welsh tale (Welsh Wikisource reading text), in original English translation across all six movements, with the full annotation layer.
read.mino.mobi/manawydan →The Fourth and last Branch of the Mabinogi — the branch of magicians. The trickster Gwydion starts a war over Otherworld pigs to wrong the king's foot-holder, and kills Pryderi; the brothers are punished as paired beasts; a boy is denied a name, arms and a wife by his own mother's curses, and given a wife made of flowers — Blodeuwedd — who betrays him to a hunter's spear. He flies off as an eagle, is sung down out of an oak and restored, and the reckoning falls. The complete Middle Welsh tale (Welsh Wikisource reading text), in original English translation across all six movements, with the full annotation layer.
read.mino.mobi/math →Geoffrey of Monmouth's Latin Life of Merlin (c. 1150), and the first Latin text on the site — the deepest taproot of the Arthurian Otherworld. After the slaughter of Arfderydd, Merlin the king-prophet runs mad with grief and becomes a wild man of the Caledonian Wood; sung sane by a harp, chained at court, riding a stag to his wife's wedding, prophesying the kings of Britain, and hearing the bard Taliesin tell of the Isle of Apples where Morgen tends the wounded Arthur. The earliest home of Avalon, Morgan le Fay, and the once-and-future king. A complete original translation of all 1,529 hexameters across fifteen movements, with the full annotation layer.
read.mino.mobi/vitamerlini →With nine full stacks in the same apparatus — Culhwch, Pwyll, Branwen, Manawydan, Math, Owain, Orfeo, Gawain, and the Vita Merlini — the comparative layer the motif index and the story graph open up can do its work. Below: the same Thompson Motif codes, the same Propp functions, the same character archetypes, laid against all nine poems. The complete Four Branches of the Mabinogi stand together on the grid — Pwyll's exchanges, the tragic Branwen, the restoring Manawydan, and the transformation-magic of Math — read together as a single work, with their own historiography, at the Mabinogi hub; and now Geoffrey of Monmouth's Latin Vita Merlini joins them as the deep-mythic outlier — the one Latin text, the one that bends the wonder-tale spine almost out of recognition, and the earliest home of Avalon and the once-and-future king. The shape of that apparatus — and how to add a tenth tale — is on the Method page.
Fifteen hundred years of the legend, on this site.
Forty entries, Gildas to The Green Knight.
The legendary 5th-c. chronology with error bands.
SVG phylogeny of texts and influences.
People, texts, motifs, places, scholarship.
Avalon, Morgan, the Otherworld.
A reader's fun-not-rigor theory.
Primary sources and scholarship.
What we mean by reading a tale deeply, why we mean it, and how a tenth tale would be added to the cycle.
Every tale lives at its own sub-site of read.mino.mobi as a self-contained reader. Each one is built from the same nine layers, and each layer is data first — JavaScript objects, not narrative HTML — so the rendering, the comparisons, and the eventual port to anything else are mechanical.
A reading text in the tale's original language — Middle Welsh for the Mabinogion branches, Middle English for the Pearl Poet and the Breton lays. Public-domain or CC-BY-SA, with manuscript provenance noted. The source is sacred; we never paraphrase it away.
Written movement by movement (passage by passage, fitt by fitt, branch by branch), in segment-aligned parallel with the source. Tuned for readability, not for the philological journals; a footnote layer carries the choices that matter. The Read tab toggles between parallel, English-only and source-only modes.
Every named figure as a card — role, epithet, blurb, appearances by movement, typed relationships to other characters. The same data, force-laid as a graph, becomes the Character web. Click a node, jump to a card; click an appearance, jump to the passage.
The tale laid against Vladimir Propp's 31 narrative functions (Α, β, γ, δ, ε, ζ, …): which functions fire, which are absent, which the tale rearranges. Each function links back to the movement that realises it. The same scaffolding works for every folk-rooted narrative — that's exactly Propp's claim.
The tale by call-number — Stith Thompson's Motif-Index of Folk Literature and the Aarne-Thompson-Uther tale-type catalogue, the folklorists' Dewey decimal for mythic memes. Each motif carries a confidence (well-attested / interpretive / speculative) and the movements where it shows.
The tale laid against A. J. Greimas's actantial model — the functional grammar of wanting beneath the plot. Six actants on three axes: a Subject who desires an Object (the axis of desire), a Sender who sets the quest in motion for a Receiver (the axis of transmission), and a Helper and an Opponent who aid or block (the axis of power). One window.<TALE>.desire object renders a six-box SVG diagram; its unreachable flag dashes the desire-arrow when the Object is never attained. Where Propp asks what happens in what order, Greimas asks what the whole machine is for.
The recurring oral set-pieces — arming, the feast in hall, the boast, the lament, the council, the voyage, the threshold-crossing, the mound, the supplication — that oral-formulaic theory (Milman Parry, Albert Lord) found to be the building blocks of traditional narrative. A window.<TALE>.themes array tags each occurrence to a passage and a shared id; they surface as ⟜ type-scene markers inline in the Read view and as a panel at the head of the Story graph. Because the ids are shared, the same type-scene can be tracked across all the tales at once.
One synergistic graph over everything above. Movements pin to a horizontal gold spine; characters, motifs, Propp functions, type-scenes and the Object of desire hang off the moments they touch, force-laid so cross-cutting elements pull toward the centre. Hover any node; click any node. It's the index of indexes.
A faithful retelling for readers ~8 and up — the whole tale across paged spreads, with art briefs for each illustration. The storybook keeps the apparatus honest: if a tale doesn't fit a child-readable page-by-page retelling, the deeper layers are masking confusion. Images are generated from the briefs by a shared illustration pipeline; the artwork is data too.
Each tale-site stands alone — you can read Culhwch without ever knowing the Pearl Poet exists. But because every layer is structured the same way, the comparative layer at Pendragon can read across them all without lifting a finger: the same Propp function code points at eight different realisations, the same motif call-number cross-cuts every cycle, the same character archetype appears under eight faces. That's the spine of this site.
The apparatus is intentionally read-only at the cross-tale level. The single source of truth for each tale is its own sub-site's data files; Pendragon picks them up and arranges them.
One registry, nine tales. Adding the illustrations for a new tale used to mean a new script. Now it's a record in scripts/illustrate/tales.mjs — a few sentences of house style, a dictionary of character & setting "pins" with the descriptions to inject when each appears, a list of regex triggers that match those pins to a given spread's art brief. A single runner, scripts/illustrate.mjs, takes --tale <slug> and renders only the spreads whose PNGs are missing — safe to re-run on every push. A matrix workflow fans it out across every registered tale.
The point is that the artwork is consequential of the writing. The storybook's illus field is an art note — the illustrator (human or model) reads it; the pipeline assembles a prompt by selecting whichever pins the brief mentions; the picture comes out matching the prose. Change a brief, delete the PNG, push: the image is rebuilt to match.
A tale read deeply once is data forever. The annotation layer is the durable contribution — the place where attention compounds. Every motif we pin and every Propp function we mark is a tiny piece of literary infrastructure that the next tale, and the comparative layer, get to reuse.
The nine tales here are the proof of concept that the same scaffolding works on a 12th-century Welsh prose tale (Culhwch), a 14th-century alliterative romance (Gawain), a c.1300 Breton lay (Orfeo), the complete Four Branches of the Mabinogi (Pwyll, the tragic Branwen, the restoring Manawydan and the transformation-magic of Math), one of the Three Welsh Romances — the Welsh twin of Chrétien’s Yvain — (Owain), and a 12th-century Latin hexameter poem (Geoffrey of Monmouth's Vita Merlini). The interesting work begins when the corpus is large enough to surface its own patterns: which motifs travel across the British Isles and which stop at the Welsh border; which Propp functions the Arthurian romances elide and which the Mabinogion branches keep; which character archetypes wear new faces in every century. The Mythograph is the inside-one-tale view of that; Pendragon is the across-tales view.
The pattern, in the order that has worked eight times:
read/<slug>/: index.html, styles.css, an empty tale.js with the passages skeleton, and the standard app.js view-switching glue copied from the most recent sister.characters.js — every named figure as a card, with role, blurb, appearances and typed relationships. The Character web falls out for free.analysis.js — the Propp pass (where Propp's 31 functions fire, which are absent, which the tale shifts), plus the desire object (the Greimas actantial six — Subject, Object, Sender, Receiver, Helper, Opponent — with *Ref cast-ids so the Mythograph can wire the actants to their character nodes) and the themes array (the Parry–Lord type-scenes, each tagged to a passage and a shared id).motifs.js — Thompson motifs and ATU types, with confidence levels and cross-tale references.app.js; it builds itself from the data files above — characters, Propp functions, motifs, type-scenes and the Object of desire all fall into the one graph.storybook.js — the faithful retelling, each spread carrying a title, prose text and an illus art brief.scripts/illustrate/tales.mjs — house style, character pins, regex triggers. Add the slug to the workflow matrix.The ninth tale, Geoffrey of Monmouth's Vita Merlini — the Latin Life of Merlin (c. 1150) — is now complete and wired into the crosswalk above: the first Latin text on the site, and the deepest taproot it reaches. It is the great outlier of the corpus, the one poem that keeps the apparatus's shape while bending the wonder-tale spine almost out of recognition — its hero flees society rather than questing into it, his lack is his own lost mind, his cure is given by a spring rather than won, and his crown is refused. And it is the earliest literary home of the things the whole hub exists to trace: Avalon (the Isle of Apples), Morgen (the first Morgan le Fay), and the once-and-future king, borne wounded over the sea to be healed. The natural tenth is a sibling of Owain that completes the Three Welsh Romances and extends the Welsh↔Continental axis: Peredur (the Welsh Perceval, the Grail) or Geraint (the Welsh Erec). The apparatus is ready.
The nine-tale crosswalk as a graph. The nine tales are pinned around the edge; the structural elements they share — motifs, story functions, character archetypes — float between them, pulled toward whichever tales claim them. The middle of the graph is where the medieval Insular tradition is densest; the outskirts are where each poem is most distinctively itself.
Tales pinned at the corners. Hover a node for its name; click for the cross-tale gloss. Toggle multi-tale only to hide single-tale leaves and surface the structural backbone.
Fifteen centuries from a leader who may never have lived to a meme that will not die. Filter by the kind of artefact; the larger haloed dots mark the pivot points where the whole tradition changed shape.
When the story is set — the legendary post-Roman chronology, c. 400–550 AD — with error bands for how little of it is really datable.
Drag to pan, scroll or pinch to zoom, ⤢ to refit. The diamond is the conventional date; the bar is the uncertainty. Badon and Camlann carry the widest bands — they are the legend's two load-bearing dates, and the sources still argue over both.
A reader's fun-not-rigor theory: that King Arthur grew out of a real common soldier the British legions made emperor. We pass it through the seeder and see how it holds up.
Flavius Claudius Constantinus was, by the scraps that survive, an ordinary soldier with a very lucky name. In 407 the Roman army in Britain — twitchy, underpaid, watching the Saxons come and the empire look away — mutinied through three would-be emperors in quick succession and settled on him, reportedly because Constantine was an auspicious name to conjure with (the great Constantine had himself been hailed emperor on British soil, at York, a hundred years before). He took the British field army across the Channel, fought through Gaul, and made his capital at Arelate — modern Arles, on the warm rim of the Mediterranean. For about four years (407–411) he ruled Gaul and Spain as Augustus, was grudgingly recognised by the legitimate western emperor Honorius and then un-recognised, besieged at Arles, and finally captured and executed in 411. Stripping the garrison out is part of why Britain was left to fend for itself — the moment (the Rescript of Honorius, 410) when the empire told the British cities to look to their own defence.
Run down the checklist and it is startling:
The kicker: when Geoffrey of Monmouth finally gave Arthur his full biography in 1136, he had him do this very thing — conquer Gaul, kill a Roman official, and march on Rome to be crowned emperor, only for treason at home to call him back. The shape of Geoffrey's Arthur is the shape of a British usurper's Gaulish campaign.
Chronology. Constantine III dies in 411. The Arthur of legend is pinned a couple of generations later — Badon around 500, Camlann around 537 (see the In-world timeline). So he is less “Arthur” than the doorway Arthur walks through: his revolt is literally the 410 end-of-Roman-Britain moment the whole legend is set just after.
“Rome too busy fighting Constantinople.” Gently: the western court at Ravenna wasn't at war with the eastern capital — East and West were nominally one empire with two courts. It was busy being sacked by Alaric's Goths in 410 and juggling other usurpers. But “Rome had bigger problems” is dead right.
The instinct is sound, and there's a well-trodden version of it. The clearest case of a British-army usurper becoming legend isn't Constantine III but Magnus Maximus — proclaimed by the British legions in 383, invaded Gaul, ruled the West from Trier until 388. The Welsh remembered him as Macsen Wledig, gave him a dreamy romance in the Mabinogion (The Dream of Macsen), and planted him as a founding ancestor at the head of medieval Welsh royal genealogies. That's the template this theory is reaching for — and Constantine III is the very next iteration of it, one generation on, with the bonus of the Arles palace.
And for an actual “Arthur was a Briton who invaded Gaul” thesis, meet Riothamus — a 5th-century “king of the Britons” who really did ship an army across to Gaul around 470, was betrayed, and vanished after a defeat near a Burgundian town called… Avallon. Geoffrey Ashe built a whole book (The Discovery of King Arthur, 1985) on identifying him with Arthur. The Avalon echo writes itself.
The best touch is the palace. Constantine's Arelate was a genuine imperial seat on the Mediterranean; the south of France is still, to a Briton, a vision of warmth and ease. You can absolutely imagine veterans limping back to the rain with tales of a British general who took the legions south, made himself emperor, and held court beside a turquoise sea — and you can imagine those tales, retold for three centuries, hardening into a once-and-future king. Whether or not it's true, it's how legends actually form. (See the historical-Arthur question for how thin the real evidence is.)
A phylogeny of texts and the matter they carry — which work descends from which, and where the marvellous Celtic strain runs through it.
Columns are strands of the tradition; vertical position is time (oldest at top). Solid lines are direct textual descent, dashed lines looser influence, dotted lines mere background. Hover a work to light up its ancestry; click the haloed pivots and ⟡-marked nodes to open the wiki.
Drag to pan, scroll or pinch to zoom, ⤢ to refit. The single fattest bundle of solid lines converges on Malory — the confluence of the whole medieval tradition into the canonical English text.
The people, texts, motifs and places — cross-linked. Search, filter by kind, or follow the “see also” trails.
How important is the fairy element? It is the legend's oldest layer — and its most persistent.
Where to read the texts themselves, and the scholarship that argues over them. Links go to stable archives and reference gateways.