PUBLIC RESEARCH PREVIEW Research in progress · uncertainties and source access are recorded throughout.

Published-record audit · October 28, 2026

Visits, residence
and connections.

The dated record supports repeated activity. How long people stayed, how often they crossed the sea, and where they came from remain separate questions.

What this round establishes

Latnija records repeated human activity, while year-round residence and voyage frequency remain unresolved. The published approximate occupation overview of 8,500–7,500 cal BP is retained as an attributed interpretation. This audit does not recalculate the first arrival, identify one continuous population, or count visits from dates.

Review LAT-OCC-001, October 28, 2026. The ledger connects 33 dated charcoal/bone records and five failed bone attempts to their published contexts and model inputs. It separately records 49 shell dates. Fourteen charcoal rows reused in the shell comparison are references to existing samples, not additional dates. The result is 82 distinct dated samples, not 82 occupation events. [S80 · Scerri et al. (2025); S106 · Scerri et al. (2025); S321 · Carleton et al. Latnija and regional Mesolithic–Neolithic chronology: code, sample map …]

From sediment to activity

Published phaseDated charcoal/bone recordsShell datesWhat the association supports
VI100 in this tableUnderlying cave-floor deposits with sparse charcoal. Old charcoal here does not independently establish older human occupation.
V1422Combustion-rich archaeological deposits. In-situ burning is supported at the N2 feature; other parts include locally reworked ash. Phase membership alone is not a unique hearth assignment.
IV23Archaeological material within deposits associated with cave-wall collapse.
III724A cut-and-fill feature with shell and ash dumps. The date of burning or organism death need not equal the time material entered the pit.

These are formation contexts, not four uniform blocks of continuous residence. The lower model also separates two stratigraphic branches, VIA–VA and VIB–VB; they are not two independently demonstrated colonizations. [S80; S106, sections 2 and 4; S321]

Two input disagreements remain explicit

Sample number as printedTable S2: conventional radiocarbon BPPublished model code: radiocarbon BPTreatment
71688, bone, context 1436,100 ± 308,100 ± 30LAT-DATE-001: retain both; no corrected value selected.
65865, charcoal, context 1777,367 ± 197,376 ± 19LAT-DATE-002: retain both; no corrected value selected.

The table and script values were checked against the rendered supplement and the authors’ repository. The first disagreement is 2,000 radiocarbon years; the second is nine. These are differences between published inputs, not estimates of how far the occupation chronology would move. Laboratory documentation and a controlled model rerun would be needed to establish the correct inputs and their consequences. No new arrival date follows from this audit.1

The January 2026 correction addresses uncertainties and boundary choices in the regional Mesolithic–Neolithic comparison. It does not settle these two site-input disagreements. The authors report that the regional correction leaves their pre-Neolithic interpretation intact. [S81 · Scerri et al. (2026)]

The original table’s laboratory heading is retained literally. The main methods attribute bone analysis to CAIS, so a number under the table’s “MAMS” heading is not independent verification of the originating laboratory. [S80 · Scerri et al. (2025); S106 · Scerri et al. (2025)]

Shells support the sequence, with shared dependencies

The shell analysis offers a different material and a useful check on the deposits. Its reservoir adjustments are estimated using charcoal from associated contexts, so the resulting shell chronology is not wholly independent of that charcoal. Published phase adjustments are retained: III, 64.2 ± 64.8 years; IV, 9.5 ± 111.5; V, −57.3 ± 128.6. These are reservoir-model quantities, not occupation dates. [S80 methods; S106 Tables S4–S5]

The ledger preserves individual shell measurements and the printed 95% calibrated BC ranges. It does not turn those ranges into continuous probability distributions or reconstruct unreported sample-exclusion decisions. Likewise, many shells within one dumping context do not demonstrate many separate visits.

Residence, seasonality and overseas contact

Residence: the late boundary remains difficult to constrain, particularly because the overlying Phase II deposits contain mixed material. The published model’s terminal boundary is broad. The evidence does not decide between seasonal use, separate extended stays, and sustained residence with outside contact. [S106, section 4; S321]

Seasonality: the original supplement explicitly leaves seasonal versus year-round exploitation open. A newly indexed 2024 dissertation abstract reports seasonal marine-resource exploitation, but its full text is restricted and was not inspected. That is a specific lead for further evidence; it does not yet let this repository assign an occupation season. Seasonal harvesting of one species would also not, by itself, prove the site was abandoned during the rest of the year. [S106, section 1.5; S322]

The 2026 population argument: the study combines terrestrial environmental productivity, exposed land area and ethnographic population-density relationships. It explores coastal distance and fishing-diet subsets as indirect marine-resource proxies; it does not directly model marine productivity. Its supported census population estimates must also be distinguished from genetic effective-population thresholds. The authors allow seasonal exploitation or separate extended occupations. These model comparisons support a case for repeated contact under their assumptions; they do not identify the travelers’ homeland or recover a voyage schedule. Published alternatives were examined here; no demographic or climate model was rerun. [S109 · Blinkhorn et al. (2026)]

Follow-up: where might the travelers have come from?

The initial origins and motives comparison is now complete. It follows the research order below, conditional on the contact interpretation. It will distinguish immediate departure coast, wider contact network, and biological ancestry.

Candidate or leadWhy investigate itEvidence needed before assigning an origin
Sicily firstThe original study treats Sicily as the most parsimonious source. Southern coasts are departure candidates.Contemporary communities, dated technological or material links, and feasible routes; proximity alone is insufficient.
The single Phase III chert coreIts geological source is reported as unknown.A defensible source identification could establish movement of material; it would not alone identify the ancestry of its carrier.
Sardinian comparisonsThe authors note similarities in expedient lithic technology.Test shared practice against convergence caused by raw-material properties; resemblance alone does not demonstrate migration.
North Africa and the wider central MediterraneanA possible broader network merits comparison.Securely dated contact or mobility evidence, while keeping indirect connections distinct from direct voyages to Malta.

This is a research order, not a ranked probability estimate or a reconstructed migration route. No source population is assigned. The follow-up comparison retains those limits and prioritizes context-linked seasonality and material sourcing. [S80 · Scerri et al. (2025); S106 · Scerri et al. (2025); S109 · Blinkhorn et al. (2026)]

Sources, versions and limits

Canonical observations, source-file hashes and unresolved fields are in research/data/latnija/occupation-audit.json. No laboratory certificates, author clarification, newly calibrated ages or new site measurements were obtained. This completes the defined published-record audit; numerical sensitivity to corrected inputs remains future work, and the origins investigation proceeds conditionally.

Footnotes

  1. Table S2 and model-code disagreement, checked October 28, 2026. Neither alternative is adopted as correct. The published approximate occupation overview is retained as a source attribution, not independently revalidated by this audit. These flags remain open while other research proceeds. ↩

Charcoal and bone: sample-to-context ledger

Table S2 values and model inputs are displayed separately. “No date” means a failed bone-dating attempt. Phase membership is not a unique hearth assignment. The original table’s MAMS heading is preserved; see the laboratory-provenance limit above.

Table sample numberMaterialSquare / contextPhase / model mapTable ¹⁴C BP ± errorModel input ¹⁴C BP ± errorTable 95% cal BC
61861charcoalN3 / 19III / III6,929 ± 276,929 ± 275886–5731
61857charcoalN3 / 25III / III6,956 ± 276,956 ± 275967–5741
54779charcoalN2 / 25III / III5,510 ± 305,510 ± 304446–4268
61851charcoalN3 / 29III / III6,918 ± 226,918 ± 225877–5731
54774charcoalN2 / 29III / III7,041 ± 327,041 ± 325993–5841
61854charcoalN3 / 31III / III6,984 ± 266,984 ± 265979–5776
54781charcoalN2 / 37III / III6,871 ± 356,871 ± 355839–5669
61846charcoalL3 / 104IV / IV6,908 ± 216,908 ± 215842–5727
54768charcoalL2 / 28IV / IV6,997 ± 326,997 ± 325983–5787
54765charcoalL2 / 34V / VB7,096 ± 347,096 ± 346059–5893
54771charcoalL2 / 38V / VB7,058 ± 337,058 ± 336015–5846
62546charcoalN2 / 39V / VA7,078 ± 277,078 ± 276018–5894
62547charcoalN2 / 40V / VA7,081 ± 267,081 ± 266019–5895
62553charcoalL2 / 41V / VB7,008 ± 267,008 ± 265984–5804
54769charcoalN2 / 43V / VA7,229 ± 337,229 ± 336220–6016
62552charcoalL2 / 44V / VB6,970 ± 266,970 ± 265971–5755
62548charcoalN2 / 46V / VA7,375 ± 277,375 ± 276372–6086
62549charcoalN2 / 47V / VA7,112 ± 267,112 ± 266061–5916
54780charcoalL2 / 48V / VB6,998 ± 356,998 ± 355985–5781
65878charcoalLM3 / 143V / VB6,903 ± 196,903 ± 195837–5728
65862charcoalM3 / 167V / VB7,231 ± 197,231 ± 196218–6022
65877charcoalM3 / 168V / VB7,504 ± 207,504 ± 206433–6260
62551charcoalN2 / 51VI / VIc9,917 ± 319,917 ± 319511–9288
65864charcoalL3 / 172VI / VIB9,018 ± 219,018 ± 218286–8231
65863charcoalM3 / 175VI / VIB7,653 ± 207,653 ± 206570–6437
65865 · input disagreementcharcoalL3 / 177VI / VIB7,367 ± 197,376 ± 196357–6086
65866charcoalL3 / 182VI / VIB5,160 ± 175,160 ± 174039–3951
65867charcoalL3 / 183VI / VIB7,740 ± 207,740 ± 206641–6481
65868charcoalM3 / 184VI / VIB7,715 ± 207,715 ± 206596–6473
65869charcoalM3 / 186VI / VIB9,238 ± 229,238 ± 228554–8340
65879charcoalN3 / 190VI / VIB6,932 ± 196,932 ± 195883–5737
65874charcoalMN3 / 198VI / VIB7,068 ± 207,068 ± 206010–5894
71684boneN3 / 171V / not modeledNo dateNot includedUnreported
71685boneN3 / 166V / not modeledNo dateNot includedUnreported
71686boneM3 / 162V / not modeledNo dateNot includedUnreported
71687boneL3 / 161V / not modeledNo dateNot includedUnreported
71688 · input disagreementboneL3 / 143V / VB6,100 ± 308,100 ± 305205–4910
71689boneM3 / 143V / not modeledNo dateNot includedUnreported

Shell measurements and associated charcoal

All 49 shell determinations are retained. Reservoir corrections are shared by phase and depend on associated charcoal. These are source measurements and printed calibration envelopes, not 49 independent visits.

Table sample numberContext / phase¹⁴C BP ± error95% cal BCReservoir group
6174519 / III7,811 ± 246219–5877LAT-MRE-III
6175419 / III7,531 ± 245974–5628LAT-MRE-III
6175519 / III7,790 ± 246211–5860LAT-MRE-III
5927519 / III7,563 ± 276001–5646LAT-MRE-III
6172919 / III7,517 ± 255966–5621LAT-MRE-III
6177219 / III7,861 ± 226236–5906LAT-MRE-III
6172425 / III7,380 ± 255851–5559LAT-MRE-III
6177025 / III7,348 ± 225834–5557LAT-MRE-III
6169529 / III7,932 ± 256257–5932LAT-MRE-III
6173529 / III7,625 ± 246062–5703LAT-MRE-III
6173729 / III7,511 ± 235961–5618LAT-MRE-III
6174029 / III7,776 ± 246201–5843LAT-MRE-III
6175229 / III7,479 ± 255929–5591LAT-MRE-III
5927329 / III7,801 ± 286215–5865LAT-MRE-III
6171329 / III7,600 ± 226036–5676LAT-MRE-III
6171929 / III7,450 ± 255901–5581LAT-MRE-III
6176929 / III7,454 ± 215903–5581LAT-MRE-III
6177329 / III7,477 ± 215925–5591LAT-MRE-III
6180729 / III7,539 ± 195977–5633LAT-MRE-III
6180929 / III7,681 ± 206118–5741LAT-MRE-III
6181029 / III7,765 ± 206197–5837LAT-MRE-III
6181129 / III7,865 ± 216240–5909LAT-MRE-III
6181331 / III7,655 ± 206082–5721LAT-MRE-III
6180837 / III7,591 ± 196025–5669LAT-MRE-III
6176428 / IV7,905 ± 266466–5956LAT-MRE-IV
6178128 / IV7,647 ± 256218–5722LAT-MRE-IV
6178428 / IV7,404 ± 215979–5526LAT-MRE-IV
6175134 / V7,592 ± 256236–5671LAT-MRE-V
6176034 / V7,467 ± 236099–5555LAT-MRE-V
6170834 / V7,790 ± 236442–5884LAT-MRE-V
6179034 / V7,245 ± 245951–5462LAT-MRE-V
6179234 / V7,419 ± 236056–5536LAT-MRE-V
6180634 / V8,091 ± 206609–6080LAT-MRE-V
6173439 / V7,750 ± 246410–5843LAT-MRE-V
6171439 / V7,685 ± 226356–5775LAT-MRE-V
6178039 / V7,708 ± 266376–5801LAT-MRE-V
6173640 / V7,406 ± 236043–5528LAT-MRE-V
6169041 / V7,345 ± 235991–5495LAT-MRE-V
6169641 / V7,971 ± 256580–6041LAT-MRE-V
6174741 / V7,602 ± 256248–5680LAT-MRE-V
6174941 / V7,420 ± 276056–5536LAT-MRE-V
6179741 / V7,339 ± 195985–5492LAT-MRE-V
6178546 / V7,403 ± 216038–5526LAT-MRE-V
6179346 / V7,776 ± 196431–5871LAT-MRE-V
6180146 / V7,906 ± 226544–5993LAT-MRE-V
6178347 / V7,564 ± 226216–5657LAT-MRE-V
6180247 / V7,586 ± 196230–5672LAT-MRE-V
6177748 / V7,544 ± 216200–5641LAT-MRE-V
6174348 / V7,449 ± 256081–5546LAT-MRE-V

The 14 reused charcoal references are 61861, 61857, 54774, 61851, 61854, 54781, 54768, 54765, 62546, 62547, 62553, 62548, 62549, 54780. Their Table S5 calibration values remain separate from Table S2 in the canonical data.

Malta research collection → · Model source →