One hymn deep

Claim (verbatim)

At Ugarit in the thirteenth century BCE, scribes wrote songs down in a hybrid format: Hurrian text above, Akkadian-language musical instructions below, using the interval vocabulary of the Mesopotamian string treatises. Laroche's edition in Ugaritica V (1968) gathers the whole genre as the h-series - roughly three dozen tablets and fragments, complete with rubrics assigning the tuning (nīd qabli), colophons crediting named composers (an Urḫiya and an Ammiya among them), and scribal signatures: an entire notated song culture, self-describing. But the corpus is one hymn deep. A single tablet, h.6, the hymn to Nikkal - physically a join of three fragments (RS 15.30 + 15.49 + 17.387) - is nearly complete and underwrites every modern performance of 'the oldest song in the world'; everything else is scrap. The mechanism is brutal sampling: clay survives fire but not falling architecture, and the h-series is one shelf's worth of a palace-quarter archive, shattered at the city's destruction around 1185 BCE; survival of a genre through a single near-complete exemplar plus a tail of slivers is precisely what a shelf collapse looks like statistically. The entire audible Bronze Age therefore hangs from one physical join, and the completeness distribution of the h-series should show a cliff, not a slope: one item near-complete, no close runner-up, and a long majority of items too broken to use at all - a distribution measurable line by line in the published edition.

Prediction clause (verbatim)

Prediction: tabulating the h-series in Laroche's edition with later collations, exactly one item (h.6) will preserve a continuous text-plus-music stretch of at least eight consecutive musical instruction units, and no second item will reach that eight-unit bar (primary clause: the one-and-only-one count at the >=8-unit bar; the verdict follows it), while at least two-thirds of the remaining items preserve fewer than five legible lines of their musical section; an instruction unit means one interval-term-plus-numeral pair as printed in the edition; the test voids as a coverage failure if fewer than 25 h-series items are countable.

Kill-dataset (verbatim)

Kill: Laroche's h-series edition in Ugaritica V (1968) with Kilmer's and later re-treatments - count legible instruction units and music-section lines per tablet and rank the completeness distribution.

Novelty / leakage triage

anticipated in the literature — this exact test has never been run

That h.6 - the join RS 15.30+15.49+17.387 - is the h-series' one substantial piece is stated from Laroche's edition onward, and Kilmer's treatments call it the only substantially complete score, everything else being fragments. But the pinned instrument is a per-item completeness tabulation - consecutive legible instruction-unit runs per tablet at a >=8-unit bar, plus a <5-legible-lines share for the remainder - and no publication was located that has run that distribution over the h-series fragments; the qualitative uniqueness is in print, the eight-unit arithmetic is not.

• E. Laroche, 'Documents en langue hourrite provenant de Ras Shamra', Ugaritica V (Paris, 1968), the h-series edition
• A.D. Kilmer, 'The Cult Song with Music from Ancient Ugarit: Another Interpretation', Revue d'Assyriologie 68 (1974), and A.D. Kilmer, R.L. Crocker & R.R. Brown, Sounds from Silence (Berkeley, 1976)
• M.L. West, 'The Babylonian Musical Notation and the Hurrian Melodic Texts', Music and Letters 75 (1994)

Predictions

No prediction registered yet.