Herculaneum Recovered A Title, Rapa Nui A Felling Year

In the Villa of the Papyri, the AD 79 eruption carbonised the only large-scale library to survive from classical antiquity. Mechanical opening destroyed as much text as it exposed. Hundreds of rolls stayed shut for that reason. PHerc 1667 has been unwrapped without a blade. Its reconstructed writing surface holds 22 columns or column-equivalents across roughly 860 cm² of papyrus, imaged at 2.4 μm isotropic voxel size on the BM18 beamline at the European Synchrotron Radiation Facility.

Counting the untranscribed portion into the total, the editors reported 33 cm² of uninterpreted surface across the surviving parts of the first three columns. The fragmentary state of the object rules out a secure reading there. Square brackets mark a restored letter. The substrate beneath has gone. An underdot flags a doubtful reading. Established papyrological convention supplies these marks. A second papyrologist can dispute an editorial supplement and cite the brackets around it.

PHerc 139 gave up a title sequence instead of running prose. The sequence names Philodemus and gives the treatise as On Gods. An overlined eta marks the book number. Herculaneum scribes usually split those two across separate lines. This scribe set them side by side, a layout already documented in the final title of PHerc 172. A scribe copied the text and wrote the title too. A Herculaneum title of this kind occupies the end of the roll, not the opening column.

Until this reading, On Gods survived as a first book only, preserved in PHerc 26. The eta does raise the count to eight at minimum. Philodemus is the most extensively attested Epicurean philosopher in the collection. His name dates the roll to the first century BC or the first century AD, with 79 AD as the outer limit. The title line yields three pieces of library metadata. A few square centimetres of surface carried all three. A sealed roll with that much identification has joined a corpus while its prose stays partly unreadable.

There are two ink regimes in these scans. A reviewer can verify the second and must trust the first. The signal in most sealed regions stays too faint for direct volumetric segmentation. Neural models trained from fragment-derived supervision enhance it on surface-conditioned renderings. The teams call those models visibility amplifiers for expert inspection. They do not identify characters, and training never touched transcriptions. PHerc Paris 4 broke the pattern, its optimised scan making ink-bearing deposits visible at apparent thicknesses of roughly 10 to 20 μm.

Collection publishers hit this problem from the commissioning side. A publisher in Papeʻete ordered a catalogue and its gallery material in Rapa Nui and Tahitian. Terminology went into a locked glossary against the source document. Reviewers checked the high-consequence fields individually. A reviewer stopped at a passage already published in an attested version and asked for that version instead of a fresh rendering. A second reviewer flagged a heading built on a pun. No substitution reproduces a pun. The heading went back for a rebuild.

Rapa Nui lies some 3,800 km off the Chilean coast. Settlement came late, between 1150 and 1280 AD. The script drew outside notice in 1864 and survives on twenty-seven wooden objects, none of them on the island. Sporadic raiding began in the early 1800s. Rongorongo production ended in the late 1860s, after the Peruvian slave raids. Missionaries handed the Bishop of Tahiti four inscribed pieces in 1869. Tepano Jaussen took delivery in Papeʻete. The four are in Rome today, held by a religious congregation and catalogued as texts A to D.

Largest of the four is Tahua, an oar-blade shape at 91.2 by 11.5 by 2.8 cm. Its timber is Fraxinus excelsior, European ash. Ash never grew on Rapa Nui. Aruku Kurenga and Mamari come from Thespesia populnea, the Pacific rosewood grown on the island. Aruku Kurenga measures 41.5 by 15.2 by 3.1 cm, Mamari 29.0 by 19.4 by 2.3 cm. Échancrée is Podocarpus latifolia, native to south-eastern Africa. Someone cut notches into its long sides later and reused it as a winding spool for cord.

Dating the four objects split them three to one. Tahua and Aruku Kurenga fall in nineteenth-century ranges. Mamari reaches furthest back among the three, into the late seventeenth century. The Échancrée result predates those three by centuries. Its calibrated interval runs from 1493 to 1509 cal AD at 68.3% probability. Europeans first made landfall in the 1720s, more than two centuries later. The team ran the three younger samples through a Bayesian phase model in OxCal 4.4 and left the older date outside it.

The collection record is what removed two calibrated intervals from consideration. Tahua produced a 48.6% range of 1934 to 1977 cal AD. Missionaries had gathered the Rongorongo in 1869. However, no calibrated range can postdate the gathering. Aruku Kurenga produced a 45.5% range of 1903 to 1947 and lost it the identical way. The accepted intervals are the minority readings, 1862 to 1887 for Tahua and 1832 to 1857 for Aruku Kurenga. A probability figure settled neither case.

While surveying the corpus for wood identification, Orliac recorded an island already described as treeless when Europeans arrived in 1720. The islanders reused old timber habitually. European shipwrights cut ash for rudders and barrels. Shipbuilders in several maritime traditions prized Podocarpus for masts. A plank salvaged from a passing hull is a plausible origin for two of the four Rongorongo. The authors date the plank by its timber and leave the engraving open. The felling year is a terminus post quem, the earliest possible date for the carving.

Speakers of old Rapa Nui called timber and wooden vessels miro. Cognates of that word name Thespesia populnea across the East Polynesian languages catalogued in POLLEX. Islanders planted the tree beside the ahu, the stone platforms under the moai. Two of the four tablets come from it. The species choice for an inscribed object signals more than availability. Thespesia populnea lives about 80 years. The editors count at least 40 annual rings in Aruku Kurenga and in Mamari. That count narrows the margin between felling and carving.

Under the Bayesian phase model the three younger Rongorongo occupy a window of 1722 to 1843 cal AD at 68.3% probability, a span of 121 years, with boundaries constrained to 1694 and 1869. Calibration ran against SHCal20, the Southern Hemisphere curve. Cellulose came out under the BABAB pretreatment protocol. The editors added fifty years plus or minus five for the gap between sample and outermost ring. Their aim was to bound the felling year. A later scholar can rerun the published parameters against a revised curve.

Svenja Bonmann, a historical-comparative linguist at the University of Cologne, studies the Epi-Olmec system of Mexico's southern coast. Its corpus is too small and its context too uncertain for decipherment. Hundreds of Indus seals bear sequences of a few signs. Scholars still argue about whether that system encodes a full language. Proto-Elamite tablets are catalogued down to the sign but survive as administrative fragments outside the known language families. Linear A and Cretan Hieroglyphic resist decipherment while Linear B, an early Greek, gave way in the 1950s. The Phaistos Disc, stamped around 1700 BC, has no sibling text for comparison.

It remains unlikely, in Bonmann's assessment, that programs capable of handling so little data will appear soon. Pattern recognition wants volume. An undeciphered corpus withholds it. She describes these systems as recombining their stored examples, capable of returning an elegant interpretation with no scientific footing. Training bias compounds the problem. A model built mostly on European material will readily propose European relatives. A buyer commissioning Pacific output can test that tendency in Huri's graded Rapa Nui comparisons against a commercial engine.

For a museum or archive in Papeʻete, the two campaigns answered different questions. Herculaneum yielded an identification. Rapa Nui yielded a boundary. The Herculaneum title line names an author and fixes a book number. The felling year dates only the timber. A catalogue entry can name the Herculaneum text. The Rapa Nui entry states a date for the wood and stops there. Huri's editors keep the brackets and the underdots on the page.