Boqueirão da Pedra Furada

What happened

1. 1978Guidon's Franco-Brazilian mission begins excavating the painted shelter.
2. 1984Guidon reports hearth charcoal dated ≥25,000 to 31,500 BP, with flaked quartz and quartzite pebbles.
3. 1986Nature paper: 14C dates to 32,160 BP; fallen painted fragments in a 17,000 BP level.
4. 1990Lynch's review questions the newly reported Brazilian and Chilean sites.
5. 1994After a 1993 visit, Meltzer, Adovasio and Dillehay question the evidence in Antiquity.
6. 1996Guidon's team replies, describing dated layers from 50,000 years b.p.
7. 2003Improved AMS dating of basal hearth charcoal gives ages beyond 50,000 14C years for seven of nine samples.
8. 2014Boëda's team reports new work at the shelter and a nearby open-air site, dated by 14C and OSL to >20,000 years.
9. 2016Capuchin monkeys in the park are shown to flake stone, making tool-like flakes and cores.
10. 2017Fiedel suggests water, gravity or ancestral capuchins could explain the 'tools'.
11. 2021Boëda et al. describe a shaped stone artefact from the nearby open-air site dated ~24,000 cal BP.
12. 2022Gómez Coutouly finds the human origin of the region's pre-LGM artifacts undemonstrated.

Objections 0 of 5 proved wrong

• ?
  Flaked pebbles may be geofacts: cobbles falling from the cliff or moved by waterfall runoff break naturally and can mimic simple tools.

  Natural, not human · Meltzer, Adovasio & Dillehay; Dennell & Hurcombe · 1994

  Unresolved — Proponents cite flaking features and site taphonomy (S15); a 2022 review finds human origin still undemonstrated (S19).
• ?
  The 'hearths' may be natural: wildfires make charcoal, and stone clusters can form without people.

  Natural, not human · Lynch; Meltzer, Adovasio & Dillehay · 1990

  Unresolved — Proponents say hearth pebbles were chosen by people (S15); reviews still judge the Pleistocene evidence ambiguous (S12).
• ?
  Wild capuchins in the same park break quartzite cobbles, leaving flakes and cores like simple human tools; ancestral monkeys could explain the finds.

  Natural, not human · Fiedel · 2017

  Unresolved — Dated capuchin tool sites in the park reach only ~3,000 years (S16); how far back monkeys used stone here is unknown.
• ?
  The claim conflicts with the well-dated Paleoindian horizon; only unambiguous, replicated cases should overturn it.

  Contradicts the accepted model · Lynch · 1990

  Unresolved — Clovis-first has since collapsed (S18), but claims older than the LGM remain contested, so this site is not settled either way.
• ?
  A simple pebble industry that barely changes over tens of thousands of years is unlike known human behaviour.

  Contradicts the accepted model · Borrero · 1995

  Unresolved — Proponents describe a long-lived tool tradition; Borrero says no mechanism for such stasis has been offered (S12).

Evidence at the time within ~5 years of the claim

• Unambiguous evidenceNo — Evidence was flaked pebbles and charcoal in hearths, which natural processes can mimic (S3); the one painted hearth stone lay in a 17,000 BP level (S1).
• Documented contextPartly — Excavated from 1978 in recorded, dated layers, but the fill came from wall decay and valley runoff, leaving the link between stones, charcoal and people open (S1).
• Independent dating methodsNo — Only radiocarbon on charcoal was reported in 1984–86 (S1, S2).
• More than one labNo — All Pedra Furada dates in 1984 carry Gif lab codes (S1); dates from three labs are reported only by 1994 (S5).
• Professional recoveryYes — Excavated by Guidon's Franco-Brazilian mission (EHESS, CNRS) from 1978 (S1).
• Peer-reviewed reportPartly — Short reports in 1984 and in Nature (1986); the site was published in detail only in 2001 and 2003 (S11).
• Open to criticsUnknown — No record found for 1984–89. In 1993 the team invited Meltzer, Adovasio and Dillehay to see the site and evidence (S4).
• Independent teamsNo — Data in 1984–86 came from Guidon's mission and its Gif dating laboratory (S1, S2).
• Outside stakeNo — Academic research funded by CNPq and CNRS (S1); no commercial, religious or nationalist motive found in the sources read.

What would settle it for a field partner

1. Blind test: independent lithic analysts sort archived Pleistocene pieces mixed with cobbles broken by rockfall and runoff below the same cliff and by capuchins in the park.Medium cost
   Would show: Whether the Pleistocene pieces can be told from natural and monkey breakage better than chance.
2. Re-map archived piece-plots: test whether Pleistocene flaked pieces cluster around hearths or track distance to the waterfall talus and drip line.Low cost
   Would show: Clustering at hearths favours people; a gradient tied to the talus favours natural breakage.
3. Analyse archived Pleistocene hearth sediments and stones for heating temperature, fire structure (micromorphology) and faecal biomarkers, with local wildfire layers as controls.Medium cost
   Would show: Repeated, contained fires or human faecal stanols would support hearths; wildfire-like signals would not.
4. A permitted team samples a nearby cliff-foot deposit of the same age without archaeology to measure the natural rate of flaked cobbles and burnt stone clusters.High cost
   Would show: If such 'tools' and 'hearths' are as common off-site, the claim fails; if rare, human agency gains weight.

What would change this record

Refuted if blind tests show the Pleistocene pieces match rockfall or capuchin breakage and matched off-site deposits hold similar 'hearths'. Vindicated by human remains, refitted knapping sequences or human biomarkers in sealed Pleistocene layers.

Notes

• Status. The deposits' great age is broadly accepted; whether people made the Pleistocene 'tools' and 'hearths' is not. Proponents cite nearby Piauí sites to ~41 ka; critics say human origin is undemonstrated.
• Accepted view then. Clovis-first: people reached the Americas near the end of the last glaciation, marked by Clovis at ~11,200 14C BP (≈13,100 cal BP, IntCal20).
• Dating basis. 14C on hearth charcoal (31,500±950 BP in S1; 50,000 in S7). Calendar values are this record's SHCal20 medians; the older one nears the curve's 55 ka limit.
• Correction. The claim began before Nature 1986: Guidon (1984) already published Pedra Furada dates of ≥25,000 to 31,500±950 BP with flaked pebbles, so year_claimed is 1984.
• Correction. The dispute is mainly about human agency, not dates: a review notes the Piauí sites have no major chronological problems (Borrero 2016).
• Correction. Doubts predate the 1994 critique: Lynch (1990) questioned the newly reported Brazilian sites.
• Correction. A mundane explanation missing from the starter notes: capuchin monkeys in the same park flake stone (Proffitt et al. 2016; Fiedel 2017).
• Correction. The critics visited in December 1993 at the team's invitation; their Antiquity critique appeared in 1994 and the team's replies in 1996.
• Not read: the body of Meltzer et al. 1994 (paywalled); its arguments are described from replies and later abstracts (S6, S7, S14). Valladas et al. 2003 reportedly found the pebbles burnt 30 to >100 ka ago with no evidence the heating was human; this was seen only in a search summary and is unverified. The Nature abstract prints 32,160±100 BP; other accounts give ±1,000. Calendar ages were computed for this record with SHCal20 (Hogg et al. 2020) for Brazil and IntCal20 (Reimer et al. 2020) for Clovis.

People and evidence

• Proponents. Niède Guidon, Georgette Delibrias, Fabio Parenti, Anne-Marie Pessis, Eric Boëda
• Critics. David J. Meltzer, James M. Adovasio, Tom D. Dillehay, Thomas F. Lynch, Luis A. Borrero, Yan Axel Gómez Coutouly
• Evidence types. radiocarbon, tl, stratigraphy, lithics, features, art · 4 independent lines
• Record confidence. medium — Key dates and positions checked in primary texts. The 1994 critique's body and the 2003 TL and 2023 papers were not read, so some objection detail rests on abstracts and replies.