Monte Verde (MV-II)

What happened

1. 1976The site is discovered in southern Chile.
2. 1982First radiocarbon report puts the MV-II occupation at about 13,000 radiocarbon years.
3. 1988A Nature report adds dates near 33,000 radiocarbon years for a possible older layer (MV-I).
4. 1990Lynch's review questions the finds' human origin and Pleistocene association, and asks for replicated cases.
5. 1997Paleoindian specialists examine the collections and site and agree MV-II is archaeological and ~12,500 years old.
6. 1999Fiedel argues the published provenience records cannot demonstrate association.
7. 1999Taylor et al. find no local 14C reservoir offset in modern plants at the site.
8. 2005Outside-lab AMS redating of a discordant gomphothere bone gives 12,460 ± 30 radiocarbon years.
9. 2008Seaweed from MV-II hearths is directly dated to 14,220-13,980 cal BP.
10. 2023Forty-three dates on the sealing peat, from a section ~50 m north of MV-II, place its base near 14,500-14,600 cal BP.
11. 2026Surovell et al. argue MV-II is no older than the mid-Holocene (8,200-4,200 BP), citing an underlying tephra and redeposited wood.
12. 2026Replies in Science eLetters and PaleoAmerica (Waters et al., site team, Meltzer et al.) reject it; Surovell remains confident.

Objections 3 of 7 proved wrong

• ✕
  With the Clovis horizon ever better defined and earlier claims failing, a glacial-age (pre-Clovis) occupation of South America was improbable.

  Contradicts the accepted model · Thomas F. Lynch · 1990

  Proved wrong — A 2019 review: genetics puts people south of the ice sheets between ~17.5 and ~14.6 ka, and sites show them in multiple places at ~15.5-14 ka (S22). Surovell et al. (2026) argue for later arrival.
• ?
  Archaeologists should depend more on unambiguous and replicated cases before accepting glacial-age occupation.

  Other · Thomas F. Lynch · 1990

  Unresolved — Outside checks came from the 1997 panel (S6) and Taylor's lab (S8); the first independent field dating study came only in 2026 (S13, S16) and is itself untested.
• ✕
  Natural processes can mimic informal hearths and simple stone tools; the 'hand of humans' at the site was not established.

  Natural, not human · Thomas F. Lynch · 1990

  Proved wrong — The 1997 panel judged MV-II archaeological (S6); the 2026 challengers also call it an 'archaeological component' (S13). If its dated wood was redeposited, whether those pieces were worked is open again.
• ?
  The artifacts' association with the dated Pleistocene layer was unproven; mixture or secondary deposition could explain it.

  Layers or association · Thomas F. Lynch · 1990

  Unresolved — Accepted by the 1997 panel (S6), and direct hearth-seaweed dates fit (S9). The same secondary-deposition mechanism was re-argued in 2026 (S13) and has no outside test yet.
• ?
  Published provenience records were inconsistent and key artifacts poorly documented, so their association with the dated surface was not demonstrated.

  Layers or association · Stuart J. Fiedel · 1999

  Unresolved — Errors in the 1997 volume led to a 2002 errata booklet (S11), and a 1989 date was misattributed to MV-II (corrected 2023, S21). The team says all finds are documented; Fiedel renewed the charge in 2022.
• ✕
  Two dates on one gomphothere bone disagreed by over 5,000 years, and a postulated local 14C reservoir effect could have made the site's dates too old.

  Dating or samples · R. E. Taylor et al. (tested the concerns) · 1999

  Proved wrong — No offset in modern site plants (S20), though Pleistocene offsets cannot be measured directly; AMS redates of the bone gave 12,460 ± 30 BP (S8); S12 traces the outlier to one lab.
• ?
  Ash dated ~11,000 years lies stratigraphically below the archaeological component, and the Pleistocene-dated wood was redeposited, so MV-II is no older than the mid-Holocene (8,200-4,200 BP).

  Layers or association · Todd A. Surovell et al. · 2026

  Unresolved — Disputed by geoarchaeologists outside the site team (tephra not Lepué; correlation unproven; S15, S17), by the site team and by Meltzer et al. Surovell remains confident (S18). No outside test yet.

Evidence at the time within ~5 years of the claim

• Unambiguous evidencePartly — Wooden structures, knotted reeds and a few bifaces, but most stone tools were simple edge-trimmed pebbles (S10); Lynch warned that natural processes can mimic hearths and tools.
• Documented contextYes — A 1986 report called MV-II a small, well-preserved residential settlement (S2); later reports describe a thin layer sealed by peat (S12). Recording errors surfaced later (S11, S21).
• Independent dating methodsNo — Radiocarbon only (wood, bone, charcoal). Later luminescence dates came from other parts of the site (S10) and nearby creek exposures (S13), not MV-II.
• More than one labUnknown — Lab names and assay years for the 1982-88 dates were not confirmed; S12 says only that dates came from different labs over four decades.
• Professional recoveryYes — University team led by Dillehay with a large interdisciplinary group of specialists.
• Peer-reviewed reportPartly — Short reports in 1982, 1986 and 1988 (S1, S2, S3); the full site volumes came in 1989 and 1997 (S12).
• Open to criticsUnknown — No source found on critics' access in the 1980s; outside specialists examined the collections and site in 1997.
• Independent teamsNo — Many specialists, all within Dillehay's project. Outside checks came later (1997 panel, S6; Taylor's lab, S20); S13 calls itself the first independent investigation.
• Outside stakeNo — None documented in the sources reviewed (an absence of evidence, not proof); results went to scholarly journals (S2, S3).

What would settle it for a field partner

1. At a lab outside both teams, AMS-date blind splits of identified short-lived plants (reed cordage, food plants) and seaweed from archived MV-II collections; apply a local marine reservoir correction to seaweed.Medium cost
   Would show: Pleistocene ages on short-lived remains rule out redeposited old wood; mid-Holocene ages support Surovell et al. (Charcoal can only show a young age.) Needs sampling permits.
2. Blind glass-chemistry fingerprinting, at a lab outside both teams, of the ash Surovell et al. call Lepué Tephra, against reference Lepué samples and any ash in the site team's section ~50 m north of MV-II.Medium cost
   Would show: Whether the ash is the ~11,000-year Lepué Tephra and whether it lies below MV-II-equivalent peat, the core of the mid-Holocene argument.
3. Wood and bone taphonomy (abrasion, rounding, decay) on archived MV-II material, plus micromorphology of any archived MV-II sediment, compared with wood the creek deposits today.Medium cost
   Would show: Whether MV-II wood and bone were carried in by water or used in place.
4. Permitted joint fieldwork, logged by both teams and an outside geologist, tracing the peat and ash beds continuously from Surovell et al.'s creek exposures to the site team's section ~50 m north of MV-II.High cost
   Would show: Whether the dated creek exposures and the site-team section are the same beds; MV-II itself was destroyed decades ago, so the correlation must be traced nearby.

What settled it Site visit, Expert review

What would change this record

An outside lab confirming the ~11,000-year Lepué Tephra beneath MV-II-equivalent peat (e.g. in the section ~50 m north) would move this record to 'open'; mid-Holocene AMS ages on short-lived plant remains from archived MV-II collections would point to 'refuted'.

Notes

• Status. Re-contested since 2026 (S13). Kept vindicated: the challengers' own 14C dates (13,400-16,500 yr) match earlier ones (S14), and their key evidence, a tephra link, is disputed on its geochemistry (S15, S17) and untested by an outside lab.
• Accepted view then. Clovis-first: no human entry into the Americas before ~13,000 cal BP; earlier claims were rejected.
• Now understood. ~14,500 cal BP (majority view; disputed since 2026)
• Dating basis. First reports (1982-88): ~13,000 14C yr BP (S3); the 1997 panel accepted ~12,500 (S6). bp_min/bp_max are the team's current calibrated MV-II range (S12), not a calibration of 13,000.
• Correction. Brief: 'Pre-Clovis occupation at ~14,500 years'. First reports (1982-88) gave ~13,000 14C yr BP (S3); the 1997 panel accepted ~12,500 14C yr BP (S6), which calibrates to ~14,500 cal BP (S12).
• Correction. Brief: 'dismissed for about twenty years before acceptance'. About 15 years passed from the first dated report (1982, S3) to the 1997 panel (S6); about 20 from the 1976 discovery (S14).
• Correction. Brief: 'a 1997 site visit by skeptics is often credited with the shift'. The panel also studied the collections, not only the site (S6).
• Correction. Brief: 'Expected status: vindicated'. That is the majority view, not unanimous: Fiedel (1999, 2022) disputed the provenience records (S7), and Surovell et al. (2026) argue MV-II is no older than 8,200-4,200 BP (S13).
• Correction. Brief: 'The older Monte Verde I claim (~33,000 years) is separate and still disputed'. The team now calls those levels inconclusive and reports possible, unconfirmed horizons of ~18,500-14,500 cal BP (S10).
• Correction. Published accounts: a date printed in 1989 as an MV-II assay came from La Moderna, Argentina; the site team corrected it in 2023 (S21).
• Covers MV-II only; MV-I (~33,000 14C yr, 1988) was left unresolved by the 1997 panel (S6). MV-II itself was destroyed over 30 years ago by logging and flooding (S14), so both new dating programs used nearby deposits: the site team a section ~50 m north (S12), Surovell's team alluvial exposures (S13, S14). The 2026 eLetters are screened, not peer-reviewed; 1997 panelist Meltzer leads one (S15). Tune and Miller (S16) take no side on the age. Methods count (6): 14C, peat-sealed stratigraphy, artifacts, structures and hearths, plant remains, fauna; the one reported footprint (S14) is not counted separately.

People and evidence

• Proponents. Tom D. Dillehay, Mario Pino, Michael B. Collins
• Critics. Thomas F. Lynch, Stuart J. Fiedel, Todd A. Surovell
• Evidence types. radiocarbon, stratigraphy, lithics, artifacts, fauna, plant remains, features, architecture, footprints · 6 independent lines
• Record confidence. medium — Paywalled sources read as abstracts; full texts of S13 and S17 not read, the eLetters (S15) were. The 2026 re-contest is live, and an outside test could change the status.