Two late copies, one older book neither keeps

Claim (verbatim)

The bulk of medieval Welsh prose reaches us through two manuscripts: the White Book of Rhydderch (Aberystwyth, NLW, Peniarth MS 4-5), written about 1350, and the Red Book of Hergest (Oxford, Jesus College MS 111), written about 1400. They share a great deal - the Four Branches, the native tales, the romances - and comparison of the shared texts shows that neither descends from the other; both draw, independently or through intermediaries, on older exemplars now lost. The proof is in the gaps. The White Book has lost leaves and suffered damage, and at those points the Red Book preserves text the White Book no longer has; conversely there are readings and passages the White Book keeps that the Red Book's line dropped or altered. Each late codex, in other words, is a partial witness to a fuller book behind both, and the places where only one of them preserves a passage are the measurable footprints of that lost archetype. The loss mechanism is ordinary manuscript attrition made legible by redundancy: with two independent descendants we can see that the common ancestor held more than either child now shows. Restated: across the tales shared by the White and Red Books there are many passages preserved in only one of the two witnesses - single-witness readings on both sides - establishing a lost fuller exemplar that neither surviving manuscript reproduces whole. Two fourteenth-century copies are the shadow of a thirteenth-century book that did not survive.

Prediction clause (verbatim)

Prediction: collating the tales common to the White Book of Rhydderch (Peniarth 4-5) and the Red Book of Hergest (Jesus 111), the count of substantive passages preserved in only one of the two witnesses - present in one and absent or materially divergent in the other - will be substantial and will fall on both sides of the comparison, not on one alone (primary clause: at least twenty substantive single-witness passages, distributed across both manuscripts rather than confined to one; the verdict follows it). Losses attributable to the White Book's known physical lacunae are tabulated separately so the count is not an artifact of damage alone. The test voids if fewer than three tales are jointly transmitted in a collatable state.

Kill-dataset (verbatim)

Kill: the diplomatic and critical editions of the shared tales - J. Gwenogvryn Evans's White Book Mabinogion and the Red Book of Hergest text, with the modern critical editions (Ifor Williams's Pedeir Keinc y Mabinogi and the editions of the romances) - collated to count substantive single-witness passages on each side.

Novelty / leakage triage

already answered in the literature

The stemmatic result is standard and printed: Ifor Williams's Pedeir Keinc y Mabinogi establishes that White and Red descend independently from a lost common exemplar, the critical editions of the shared tales print the apparatus of divergent and single-witness readings on both sides, and Huws's account of the two codices states the fuller-ancestor conclusion as the field's settled position, with the White Book's physical losses and the Red Book's unique preservations itemised. The >=20 substantive-passage threshold is guaranteed by apparatus already in print - Evans's diplomatic White Book text set against the Red Book readings shows it many times over in the Four Branches alone. The one soft edge, a few late shared texts where Red-Book dependence on the White Book has been argued, does not touch the core corpus the prediction names.

• D. Huws, Medieval Welsh Manuscripts (Cardiff, 2000), chapters on the White Book of Rhydderch and the Red Book of Hergest
• Ifor Williams (ed.), Pedeir Keinc y Mabinogi (Cardiff, 1930), introduction
• J. Gwenogvryn Evans, The White Book Mabinogion (Pwllheli, 1907)

Predictions

No prediction registered yet.