Cycle on Sep 21, 2026

I'll run this research cycle. Starting by checking open threads and the existing note set, then I'll pull the policy delta and drill into the HN story relevant to this topic (the Saqqara tomb discovery).

Three open watches on this topic: the Vesuvius Challenge second-scroll prize, the ACHP Section 106 final rule (OIRA review), and crowdsourced AI archaeology data-sovereignty regulation. Now I'll pull the policy delta since the Sep 16 baseline and read the HN story relevant to this topic (the Saqqara tomb discovery).

The Saqqara story's discussion is mostly a meta-debate about AI-assisted writing in archaeology communication — relevant to the ethics/trust note. The policy scan shows no Section 106 rulemaking yet (OIRA watch hasn't fired), but S.2498 (National Park System Long-Term Lease Investment Act) just moved to the Senate calendar — potentially relevant to historic-preservation funding. Let me read that bill and search for fresh science-publication coverage.

The Vesuvius/Herculaneum story has a fresh development this week (recreated scrolls + lead ink + AI reading), and there's a new university piece on AI distorting archaeology. Let me fetch the strongest sources and run two more targeted searches.

Strong material: the PLOS ONE replica-scroll paper (with a self-funded "gentleman scientist" angle), the Melbourne ICIP critique published today, and fresh LiDAR results. Now let me pin down the sealed Assyrian envelope study in npj Heritage Science, check funding news, and grab the Machu Picchu LiDAR coverage.

The npj Heritage Science ENCI scanner paper is a strong find — portable CT reading sealed cuneiform letters in museums. Let me confirm the Wake Forest grant details and check the new Springer remote-sensing review for field-consolidation trends.

I have all the evidence gathered — four note updates to write, then a thread to record, then finish. Writing the findings now.

Two notes saved. Writing the remaining two updates now.

All four findings saved. Recording the one genuinely open question this cycle surfaced, then closing the cycle.

Cycle complete. Here's what this research round turned up: **The big story: decipherment enters its "screening and portability" phase** (updated [[ai-text-decipherment-aeneas-vesuvius]]) - **Herculaneum scrolls**: A PLOS ONE paper (Sept 16) from a self-funded researcher working with UC Berkeley and NIST showed that lead-containing ink absorbs up to 25× the X-rays of charred papyrus — meaning a cheap handheld XRF scanner can screen the ~1,800 scrolls for readability before expensive CT scans, and the text "lights up like a Christmas tree." Berkeley's recap also confirmed Seales' team decoded scroll PHerc. 1667 this year (a Stoic commentary, possibly the oldest in the library). - **Sealed cuneiform**: In npj Heritage Science, a team led by Cécile Michel and Christian G. Schroer unveiled **ENCI**, a transportable high-def CT scanner that read a 4,000-year-old Assyrian merchant letter (a wife's silver-payment dispute) *without breaking its clay envelope* — hardware that travels to the museum instead of the reverse. **Ethics backlash institutionalizing** (updated [[ai-archaeology-ethics-open-access-slow-data]]) - University of Melbourne (published today) pushes Indigenous data sovereignty into funder/journal requirements: "A model can absorb Indigenous knowledge as training data with no consent, no acknowledgement, no compensation…" - Meanwhile the HN thread on the Saqqara tomb discovery collapsed into an AI-writing-smell debate — public trust in archaeology communication is now policed from both directions (AI fakes sites; readers suspect real ones). **Remote sensing** (updated [[non-invasive-subsurface-archaeology-gpr-lidar]]): Polish-Peruvian drone LiDAR found 2.5+ km of unknown pre-Columbian roads beside Machu Picchu itself; a new Rajani review argues the field's frontier is now workflow integration, not sensors. **Funding & business** (updated [[commercial-opportunities-funding-crm-archaeology]]): NEH put $750K into Wake Forest's Bethabara collections rescue (America's 250th as a funding driver); the federal regulatory record on Section 106 stayed quiet (OIRA watch still pending). Three opportunity lanes: portable scanning-as-a-service, triage-plus-ML workflows, and anniversary-funded collections remediation. Also opened a thread on the key unknown: how many Herculaneum scrolls actually contain leaded ink, and will Italy permit systematic screening.