The Seeds Beneath the Steps

In 2021, radiocarbon dates on seeds from the layers holding human footprints at White Sands, New Mexico, ranged from 22.86 to 21.13 thousand years ago, deep in the last Ice Age. Recalibrate all eleven from the raw laboratory measurements, then add the pollen dated in 2023 to test whether the seeds took up old carbon from the water. The pollen agrees on an Ice Age date. Its own published error bars show that it cannot confirm the narrow span between the seed dates, and no date can say how the dated material came to lie beside the footprints or when any one person walked there.

Every recalibrated date and range on this page, every count of agreement and every gap is computed in your browser, now: from the released laboratory measurements and the IntCal20 calibration curve, or, in section III, from the authors' own printed ranges, labelled as such. Numbers taken from the papers are labelled as printed or sit inside quotation marks. Nothing leaves the page. Go straight to the trench and add the pollen.

I · The claim, at full strength

Eleven seed dates from one trench

In September 2021, Matthew Bennett and 14 coauthors reported human footprints in the lake-margin mud of White Sands National Park, New Mexico, on surfaces stacked one above another in a trench at a site called Locality 2. Between the footprint surfaces lie thin layers of seeds from ditch grass, Ruppia cirrhosa, a plant that grows under water. Radiocarbon dates on those seeds put the footprint-bearing layers inside the Last Glacial Maximum, which the paper dates to 26.5 to 19-20 ka: a ka is a thousand calendar years, counted back from AD 1950. The authors' one-sentence summary:

In situ fossil footprints show that humans were present in North America during the Last Glacial Maximum.

Bennett and colleagues, accepted manuscript of the 2021 Science paper, one-sentence summary, PDF page 1. The manuscript and its terms.

The main text puts it at its strongest:

These data provide definitive evidence of human occupation of North America south of the Laurentide Ice Sheet during the LGM.

Bennett and colleagues (2021), accepted manuscript, PDF page 2. LGM is the Last Glacial Maximum.

And the number the claim stands on, from later in the same manuscript. The seed layers

…yielded calibrated ages that range from 22.86 ± 0.32 to 21.13 ± 0.25 ka (N=11) and maintain stratigraphic order within the uncertainties between TH2 and TH6.

Bennett and colleagues (2021), accepted manuscript, PDF page 3. TH2 to TH6 are track horizons, the walking surfaces, counted upward.

Older end · USGS-1810

Printed22.86 ± 0.32 ka

Laboratory measurement19,010 ± 120 radiocarbon years

Recalibrated hereComputing…

95.4% rangeComputing…

Younger end · USGS-1803

Printed21.13 ± 0.25 ka

Laboratory measurement17,470 ± 90 radiocarbon years

Recalibrated hereComputing…

95.4% rangeComputing…

Loading the frozen measurements and the calibration curve.

Each raw age is calibrated against IntCal20 with a uniform calendar prior on a one-year grid. The printed ± is not a standard deviation: it is half the width of the 95.4% highest-density range, and the printed age is that range's midpoint. The comparison tolerance, 20 calendar years per end, midpoint and half-span, covers the printed rounding, the curve's native spacing and a whole-year interval routine; it is not extra dating uncertainty.

The span between the two printed midpoints is 1.73 ka; recomputed, Computing…. That is the distance between two dated seed samples, not a measured length of human presence. The paper's estimate of placing humans in this part of North America for approximately two millennia during the LGM came from its OxCal age models, a kernel-density model with Boundary and Interval functions, which this page does not rerun.

Which number did they print?

The release's data dictionary names the printed column Mean_calibrated_ages. A posterior mean is the natural reading of that name. Try it against the middle of the range, on all eleven rows at once.

Each row is one seed determination. Ticks: the recomputed low end, high end and half-span minus the printed ones. Dot or diamond: the chosen statistic minus the printed middle. A filled red mark lies outside the tolerance.

Computing…

All eleven seed determinations, row by row

From the 2021 USGS release. Raw ages in radiocarbon years BP, ± one sigma; calendar values in ka. The printed columns are compared with, never used as input.

Seed sampleRaw age ± 1σPrinted, kaRecalibrated, kaPosterior mean, ka95.4% ranges, kaLargest difference, yearsMidpoint minus printed, years

The oldest recomputed midpoint is not the headline's. USGS-1808, sampled at the same elevation as USGS-1810, recomputes to Computing…, and the release prints 22.87 ± 0.30 ka for it. The 2023 control paper describes the seed samples as those which yielded ages that range from 22.87 ± 0.30 to 21.13 ± 0.25 ka. The 2021 sentence's 22.86 ± 0.32 ka is USGS-1810, a different sample (Figure 2A labels it 3a, against 2a for USGS-1808) released at the same elevation, with a wider range. Both are kept here; the headline is the 2021 sentence.

The authors saw the objection coming

Ruppia grows under water, and water can carry old carbon. The 2021 paper answered that with three lines of evidence. Dates from elsewhere in the basin showed, in the authors' words, that terrestrial and aquatic material yielded concordant 14C ages between ~44 and 25 ka, which demonstrates that hard-water effects in Paleolake Otero were less than a few hundred years during this period. The seeds keep their order in the trench: our calibrated 14C ages maintain stratigraphic order even when samples were separated by centimeter(s), which would not be the case if hard-water effects were large and variable. And the shallow lake margin makes a large effect unlikely, they argued. For these reasons, we conclude our radiocarbon-based chronology is robust.

II · The deciding control

Add the pollen.

If the lake water carried carbon dissolved from old rock, carbon whose radiocarbon had long since decayed, every seed would read older than it is. That is the hard-water, or reservoir, effect. Conifer pollen is made on land from the carbon in the air, so it cannot inherit it. Put both materials on one clock.

Left: the sampled trench, drawn to scale from 1197.05 to 1197.41 m above sea level; the seven released track horizons run from 1197.08 to 1197.39 m. The quartz level is placed from the article's description, just below the lowest radiocarbon sample; the release gives no elevation for it. Right: each determination's calibrated probability, scaled to its own peak, with its 95.4% range as a bar and its midpoint as a tick, ordered by elevation and joined to its height by a line. Pale blue: the Last Glacial Maximum as the 2021 paper gives it, with the uncertain 20 to 19 ka edge lighter.
Shown
Computing…
Seed envelope
Computing…
Seeds and the Last Glacial Maximum
Computing…
Seeds and pollen
Computing…

Computing…

The younger scenario moves every seed 7,500 calendar years, the scale of the modern-analogue argument by Rachal, Dello-Russo and Cuba, whose 2024 abstract says the footprints could be at least ~7500 yr younger. The page applies it as a calendar translation, shape and width unchanged, to show the scale. It is not their calculation and it is not a measured correction; section IV subtracts reservoir ages properly, one seed at a time.

Who asked for the test, and what it was

The objection was in print within four months of the claim. A comment in Science put it plainly:

Reservoir effects may have caused radiocarbon ages to appear thousands of years too old. Independent verification of the ages of the footprint horizons is imperative and is possible through other means.

Madsen, Davis, Rhode and Oviatt (2022), abstract.

The claimants replied in the same issue: On the basis of the geologic, hydrologic, stratigraphic, and chronologic evidence, we maintain that the ages are robust and conclude that the footprints date to between ~23,000 and 21,000 years ago. Later that year the critics reported a test on the plant itself: Ruppia specimens collected in 1947 from nearby Malpais Spring returned a radiocarbon age of ca. 7400 cal yr BP. (Oviatt and colleagues, 2022.)

The independent verification came in October 2023, from Jeffrey Pigati and colleagues; 7 of that paper's 11 authors had also written the 2021 paper. They returned to the trench: we collected large (>1 kg) samples of bulk sediment from the exact same stratigraphic levels as those of the original samples of Ruppia seeds. Each pollen date needed about 75,000 conifer grains, sorted from the sediment by flow cytometry. In their words, The dating techniques, the sample types, and the accelerator mass spectrometry facility involved in this study are independent of those used by Bennett et al.

The control: three ancient pollen measurements from the 2023 USGS release, each calibrated separately from its measured fraction of modern carbon. Calendar values in ka; the printed columns are compared, never used as input.

Pollen sampleElevationMeasured F14C ± 1σPrinted, kaRecalibrated here, ka95.4% ranges, kaLargest difference, years

Computing…

Computing…

Computing…

The article prints the result as the resulting calibrated ages range from 23.4 ± 2.5 ka to 22.6 ± 2.3 ka. At the top layer the pollen's midpoint sits older than the seeds beside it, not younger, which is the opposite of what a large hard-water effect in the seeds would predict. That ordering is weak evidence, though: the pollen's range contains the midpoint of every seed beside it, and the ordering rests on the size of the blank correction described next, which is larger than the gap. Its ranges are wide, and section III asks what that width can and cannot decide.

The pollen carries one large assumption, and the article states it. Isolating the grains took extensive chemical processing, and the dates needed blank corrections that are approximately an order of magnitude larger than what is typically used in radiocarbon dating. The blank was pollen from sediment beyond the reach of radiocarbon, and the correction rests on a stated premise: Assuming that the same amount of contamination was introduced to all of the samples analyzed in this study, which is equivalent to an increase of ~2200 14C years or 2500 calendar years for the WHSA Locality 2 sequence. That shared correction is why each range spans close to five thousand years, and why three pollen dates are not three independent votes. The authors add that even had they not used their pollen-specific approach for the blank correction, the calibrated pollen ages would still have fallen within the Last Glacial Maximum; the release gives only the corrected fractions, so this page cannot rerun that alternative. The authors also tested the reworking objection directly: The 14C content of the playa conifer pollen (22) is indistinguishable from current atmospheric 14C levels (29), which shows that reworking of old pollen is inconsequential in this part of the Tularosa Basin today.

III · The control on the control

What could the pollen have told apart?

Grade B · published sensitivity

Grade B, published sensitivity. This page recalibrates the released pollen fractions, but it trusts the control authors’ analytical uncertainties, their processing correction and the premise that the pollen dates the deposition of its layer. It cannot rerun the laboratory control, so it cannot measure how often the control would miss a real offset. The ledger’s control_can_confirm cell is therefore false, and the questions the resolution cannot reach are marked INCONCLUSIVE.

So ask a narrower question, using only numbers the two sets of authors printed. Given how wide the published pollen ranges are, which differences could they resolve?

Every bar is a printed range from the releases. The last bar is the full seed envelope moved younger by the scenario's 7,500 calendar years. Pale blue: the Last Glacial Maximum.

An Ice Age date against the younger scenario

Computing…

… between the moved seed envelope and the youngest printed pollen edge.

Computing…

The smallest calendar shift of the whole seed envelope that clears every printed pollen range: …. A resolution threshold, not a limit on any real offset.

The narrow span between the seed dates

Computing…

Printed pollen widths of … against a … span between the headline midpoints (recomputed: …).

Computing…

Agreement on an old age does not establish the span, the duration of occupation, or which footprint belongs to which moment. Nor could the pollen bracket the span at any width: its released elevations run from 1197.27 to 1197.36 m, and none is at 1197.143 m, where the older headline seed was sampled.

The control's own printed ranges, the grade-B inputs. Every pollen range lies … the Last Glacial Maximum on its narrower reading.

Pollen samplePrinted range, kaWidth, kaContains the headline intervalWider than the spanInside 26.5 to 20 ka

What these comparisons stand on

None of these can be tested by overlapping intervals. That is why the coarse result above is stated with its conditions, and the rest is INCONCLUSIVE: neither confirmed nor ruled out.

IV · A further result, computed by this page

How much depends on which seed?

Six seed dates share a released elevation with a pollen date. Pick a pair, subtract an assumed reservoir excess R from the seed's radiocarbon age, and recalibrate. The pollen does not move. Watch where the two ranges part.

R runs from 0 to 8,000 radiocarbon years in steps of 100. Corrected age = measured age − R. The sources agree on no ancient value of R; the slider explores an assumption, it does not measure one.

Brown: the seed after the chosen move. Dashed: the same seed as measured. Green: the pollen, unchanged. Both curves are scaled to their own peaks, so height is not evidence weight.

Computing…

Seed midpoint moved by
Computing…

Computing…

Computing…

Every pair, every scenario

The sweep starts when the page has loaded.

First separation on this 100-year scenario grid is a descriptive comparison, not a confidence limit, a maximum allowable reservoir age or a measured correction. The six comparisons are not six experiments: three seeds share one pollen date, two share another, and every one shares the calibration curve. Envelope overlap ignores any gaps inside a highest-density set; the full ranges are listed.

Pairs are matched by released elevation. The pollen sample numbers 1aa, 1dd and 1ff echo the seed samples 1a, 1d and 1f beside them; USGS-1806, from sample 1e, shares only its elevation with 1aa.

What the sweep can and cannot show. Each step subtracts an old-carbon offset from a copy of one real seed measurement and reruns the unchanged calibrator against the real pollen, so it is an offset of the kind the critics propose, planted and then recovered through the page's own code: every pair has separated from its pollen by R = 2,200 radiocarbon years. Each separation happens where the seed's older edge passes the pollen's younger edge, and that edge sits where the shared blank correction and its error budget put it. A planted seed offset cannot test a bias the pollen shares, or one inside the pollen's own processing, which is why section III stays at grade B.

Same layer, different answer

Why should the pairs differ so much when they share a level? Partly because the seeds disagree with each other. The 2023 paper's Figure 2A labels every seed date with its sample number, and four sample numbers are shared by two aliquots each. The page takes aliquots that share a number as one layer, which their identical released elevations support. Two aliquots of one layer, dating one moment, should agree within their laboratory errors.

Radiocarbon ages ± one sigma, as released. The test statistic is the standard one for radiocarbon ages, the squared difference over the summed variances; p is two-sided.

Sample numberAliquotsRadiocarbon agesDifference, yearsStandard errors apartpConsistent at 5%

Computing…

Sample 1a holds USGS-1803, the younger end of the headline range. The test cannot say why its two aliquots disagree. The layer may record more than one moment; the seeds may carry different amounts of old carbon; some may be reworked from older mud; or the laboratory errors may be understated. Each reading means something different for the reservoir argument. The critics' 2024 paper makes the general point in one line, Extant age estimates are not as internally consistent as suggested. The claimants' 2026 fact sheet answers the reworking reading directly: Seeds chosen for radiocarbon dating were often still attached to their stems and within well-defined layers, which ensures they have not moved since the plants were alive.

How much of the headline range rests on that one aliquot?

Computing…

The test above cannot say which aliquot of sample 1a is off, and this is the spread of dated seed midpoints, not the paper's modelled duration.

we searched the open web, GitHub-indexed results and the repository's existing strata on 2026-09-22 and did not find an interactive six-pair sweep of White Sands seed reservoir scenarios against separately recalibrated pollen measurements.

The reservoir hypothesis and the modern-analogue correction are scientific prior work by Oviatt and colleagues and Rachal and colleagues. Where the Curve Goes Flat is this site's computational precedent.

V · An approximation you can catch

The fraction before the age

A pollen laboratory measures a fraction of modern carbon, F14C, and the radiocarbon age is −8033 × ln F. A symmetric error on the fraction becomes a lopsided error in years, longer on the old side. Replace the measured fraction with its rounded age and a symmetric ±, and the calibrated range slides younger.

Green: calibrated from the measured fraction. Orange, dashed: calibrated from the rounded age with a symmetric error. The chosen method is drawn strongly with its 95.4% range; the other is ghosted, with its range as the grey dashed bar beneath.

Computing…

Midpoints and ranges in calendar ka; the shift and the source difference in calendar years.

PollenChosen method, midpoint ± half-span95.4% rangesFraction minus age midpointLargest difference from the release

The approximation never replaces the fraction calculation in the control. Both paths run through the same calibration and interval functions; only the measurement domain changes.

The check

What went into every number

Calibrator
IntCal20, linearly interpolated onto a one-year grid from 0 to 55,000 calendar BP, with a uniform calendar prior and the curve's own uncertainty added to each measurement's. The 95.4% highest-density set keeps separate ranges separate.
Seed likelihood
v(t) = laboratory σ² + curve σ(t)²
p(t) ∝ exp[−(age − curve age(t))² / 2v(t)] / √v(t)
Pollen likelihood
curve F(t) = exp[−curve age(t) / 8033]
curve σF(t) = curve F(t) × curve σ(t) / 8033
the same Gaussian form, in fraction space
Tolerances
20 calendar years for seeds, 25 for pollen, per range end, midpoint and half-span. The pollen fractions are released to four decimals; moving a fraction and its error by half a last digit moves a range end by at most 17 years here. Neither tolerance is dating uncertainty.
What is approximate
First-order propagation of the curve error into fraction space; rounded released measurements; whole-year interval ends. This is a close recalibration, not a replay of the laboratory blank subtraction or of OxCal's internals.
What is not rerun
The 2021 kernel-density, boundary and duration model; the 2023 sequence model, which combines pollen with the very seeds under test and so cannot serve as an independent control; the luminescence minimum-age model.
Free choices, all yours
Which evidence the trench shows; the younger scenario; which statistic is compared with the printed column; the seed and pollen pair; R; how the seed is moved; the pollen measurement domain. Reset restores every default. None of them touches the anchor or the dated status.
Fixed choices
The curve, the prior, the grid and the 95.4% mass are fixed above and were not tuned. Pairs are matched by released elevation, never by nearest-looking ages. Sample numbers come from the control paper's Figure 2A and are joined to the release by their printed values.
The null test
Not applicable. These are chronometric measurements on different materials, not a search statistic whose false discoveries could be reproduced by running the claimants’ procedure on noise. The modern pollen and the procedural blank are real laboratory checks, but they do not give this page a complete, rerunnable contamination model.
Same-layer test
Two aliquots that share a Figure 2A sample number are compared with T = (a₁ − a₂)² / (σ₁² + σ₂²), chi-squared with one degree of freedom, two-sided, at the 5% level, using the released one-sigma laboratory errors. With four sample numbers tested, the smallest p is also given after a Šidák adjustment. The test says whether two ages could be one; it cannot say why they are not.
Planted faults
The verifier adds 0.02 to a copy of one pollen fraction and 500 radiocarbon years to a copy of one seed age, runs the unchanged engine, and requires the reproduction to fail and the result to move. These are fault checks, not power tests.
The rows that are not ancient dates

The control release also lists the seven track horizons, two modern pollen samples and one procedural blank. Offer any of them to the calibrator.

RowWhat it isElevationF14C ± 1σCalibrate

Ask for something the record cannot give

The claim ledger and checksums · seed extract · pollen extract · OSL extract · layer labels · IntCal20 as frozen · sources and reuse terms

VI · The dated record

Five years of argument, one trench

  1. September 2021the claim

    Seed dates from Locality 2 place the footprint layers between about 23 and 21 thousand years ago.

    Bennett and colleagues
  2. January 2022objection

    A published comment argues that a reservoir effect could make the seeds read thousands of years too old, and asks for independent verification.

    Madsen and colleagues
  3. January 2022reply

    The claimants reply that the geologic, hydrologic, stratigraphic and chronologic evidence keeps the ages robust.

    Pigati and colleagues
  4. September 2022objection

    Ruppia collected in 1947 at a nearby spring gives a radiocarbon date of about 7,400 calibrated years before present.

    Oviatt and colleagues
  5. October 2023the control

    Terrestrial pollen from the same levels, and quartz by luminescence, give ages consistent with the seeds.

    Pigati and colleagues
  6. January 2024objection

    A modern-analogue correction from living Ruppia argues the footprints could be at least about 7,500 years younger.

    Rachal and colleagues
  7. August 2024objection

    Critics argue the seeds, the pollen and the luminescence ages may all be too old, and that the dates are less internally consistent than reported.

    Rhode and colleagues
  8. August 2024reply

    The claimants review each objection and reject it.

    Pigati and colleagues
  9. June 2025support

    A paleolake sequence traced into the track-bearing alluvium adds 5 seed dates and 21 bulk-sediment dates from two further laboratories.

    Holliday and colleagues
  10. July 2025objection

    A further critique examines the ecology, depositional context and radiocarbon dating of the Ruppia seeds.

    Rachal and Dello-Russo
  11. May 2026support

    A USGS fact sheet presents the 23,000 to 21,000-year interpretation as the account of the site.

    Springer and colleagues

The claimants' 2024 reply answers the critics directly: We review each of these issues and show they are without merit, often irrelevant, at odds with first principles. The 2025 study, whose first author was also a 2021 coauthor, states: This paper presents 5 AMS radiocarbon dates on seeds and 21 acceptable dates on bulk sediment; and The dates on the seeds generally align with the dates on the mud with no systematic offset. The 2026 fact sheet: Overall, TH 2–6 date between 23,000 and 21,000 years ago, recording nearly 2,000 years of human activity at White Sands National Park during the Last Glacial Maximum. Its series number, 2025-3046, is not its publication year.

Status of the precise chronology, as of 2026-09-22

OPEN

OPEN describes the precise chronology and the link between the dated materials and the footprint surfaces. It does not give every proposed age equal weight: three kinds of evidence, dated in different laboratories, now support the old age. The critics’ published objections go beyond association: they argue that the seed dates carry an old-carbon offset, that the pollen may be reworked from older deposits, that the luminescence ages may be too old, and that the dates are less consistent than reported; the claimants reject each of these. The category is this page’s mapping of a documented dispute into the wave’s vocabulary; no source has voted on the word.

No decision date is asserted. The claim was published in September 2021; as of 2026-09-22 the precise chronology is still argued in the peer-reviewed record.

Established by

The continuing objection, Rhode and colleagues (2024) and Rachal and Dello-Russo (2025); the claimants' reply, Pigati and colleagues (2024); the supporting study, Holliday and colleagues (2025); and the official synthesis, Springer and colleagues (2026).

What would change it

Independently reproduced dates on securely associated terrestrial material from the footprint surfaces, with a depositional account that directly tests reworking and the tie between each dated layer and each surface; a direct, independently reproduced measurement of a large bias in the seed or pollen dates, going beyond the carbon-isotope arguments for one that are already published and that the claimants’ reply rejects; or a peer-reviewed reconciliation of the two positions.

This page's result

The pollen supports a Last Glacial Maximum date for the footprint layers, but its published error bars are too broad to confirm the 1.73 ka span between the seed dates, and no calibration can test how the dated material came to lie beside the footprints.

These dates concern people present at one place and time. They do not decide a migration route or anyone's ancestry. The primary papers acknowledge the tribes and pueblos whose homeland this is; this instrument does not test cultural traditions and does not pretend to.

Sources

Every source, field by field

  1. The claim. Matthew R. Bennett, David Bustos, Jeffrey S. Pigati, Kathleen B. Springer, Thomas M. Urban, Vance T. Holliday, Sally C. Reynolds, Marcin Budka, Jeffrey S. Honke, Adam M. Hudson, Brendan Fenerty, Clare Connelly, Patrick J. Martinez, Vincent L. Santucci and Daniel Odess. 2021. Evidence of humans in North America during the Last Glacial Maximum. Science 373(6562), 1528-1531. DOI 10.1126/science.abg7586. https://doi.org/10.1126/science.abg7586 · 24 September 2021 (print). Accepted manuscript, quoted here: https://eprints.bournemouth.ac.uk/36202/
  2. The claimants’ data. Matthew R. Bennett, David Bustos, Jeffrey S. Pigati, Kathleen B. Springer, Thomas M. Urban, Vance T. Holliday, Sally C. Reynolds, Marcin Budka, Jeffrey S. Honke, Adam M. Hudson, Brendan Fenerty, Clare Connelly, Patrick J. Martinez, Vincent L. Santucci and Daniel Odess. 2021. Data release for Evidence of humans in North America during the Last Glacial Maximum. U.S. Geological Survey data release. DOI 10.5066/P9ABZEM9. https://doi.org/10.5066/P9ABZEM9 · 23 September 2021. File 5, radiocarbon sample information for Locality 2.
  3. The first published objection. David B. Madsen, Loren G. Davis, David Rhode and Charles G. Oviatt. 2022. Comment on “Evidence of humans in North America during the Last Glacial Maximum”. Science 375(6577), eabm4678. DOI 10.1126/science.abm4678. https://doi.org/10.1126/science.abm4678 · 14 January 2022. Quoted from the abstract in the publisher’s Crossref record.
  4. The claimants’ first reply. Jeffrey S. Pigati, Kathleen B. Springer, Matthew R. Bennett, David Bustos, Thomas M. Urban, Vance T. Holliday, Sally C. Reynolds and Daniel Odess. 2022. Response to Comment on “Evidence of humans in North America during the Last Glacial Maximum”. Science 375(6577), eabm6987. DOI 10.1126/science.abm6987. https://doi.org/10.1126/science.abm6987 · 14 January 2022. Quoted from the abstract in the publisher’s Crossref record.
  5. A modern-plant test of the reservoir effect. Charles G. Oviatt, David B. Madsen, David Rhode and Loren G. Davis. 2022. A critical assessment of claims that human footprints in the Lake Otero basin, New Mexico date to the Last Glacial Maximum. Quaternary Research 111, 138-147. DOI 10.1017/qua.2022.38. https://doi.org/10.1017/qua.2022.38 · 2 September 2022 (online). Quoted from the abstract in the publisher’s Crossref record.
  6. The deciding control. Jeffrey S. Pigati, Kathleen B. Springer, Jeffrey S. Honke, David Wahl, Marie R. Champagne, Susan R. H. Zimmerman, Harrison J. Gray, Vincent L. Santucci, Daniel Odess, David Bustos and Matthew R. Bennett. 2023. Independent age estimates resolve the controversy of ancient human footprints at White Sands. Science 382(6666), 73-75. DOI 10.1126/science.adh5007. https://doi.org/10.1126/science.adh5007 · 6 October 2023 (print). Quoted from the article, pages 73 to 75.
  7. The control’s data. Jeffrey S. Pigati, Kathleen B. Springer, Jeffrey S. Honke, David Wahl, Marie R. Champagne, Susan R. H. Zimmerman, Harrison J. Gray, Vincent L. Santucci, Daniel Odess, David Bustos and Matthew R. Bennett. 2023. Data release for Independent age estimates resolve the controversy of ancient human footprints at White Sands. U.S. Geological Survey data release. DOI 10.5066/P9E36U4B. https://doi.org/10.5066/P9E36U4B · 5 October 2023. Tables 1 and 2 and the data dictionary.
  8. The younger scenario. David M. Rachal, Robert Dello-Russo and Matthew Cuba. 2024. The Pleistocene footprints are younger than we thought: correcting the radiocarbon dates of Ruppia seeds, Tularosa Basin, New Mexico. Quaternary Research 117, 67-78. DOI 10.1017/qua.2023.74. https://doi.org/10.1017/qua.2023.74 · 10 January 2024 (online). Quoted from the publisher’s abstract.
  9. The continuing objection. David Rhode, Christina M. Neudorf, David Rachal, Loren G. Davis, David B. Madsen and Robert Dello-Russo. 2024. Unresolved: Persistent Problems with the White Sands Locality 2 Geochronology. PaleoAmerica 10(1), 10-27. DOI 10.1080/20555563.2024.2345979. https://doi.org/10.1080/20555563.2024.2345979 · 5 August 2024 (online). Quoted from the abstract as indexed by OpenAlex; the publisher page refused automated retrieval (HTTP 403).
  10. The claimants’ reply. Jeffrey S. Pigati, Kathleen B. Springer, Harrison J. Gray, Matthew R. Bennett and David Bustos. 2024. The Geochronology of White Sands Locality 2 is Resolved. PaleoAmerica 10(1), 28-44. DOI 10.1080/20555563.2024.2376298. https://www.usgs.gov/publications/geochronology-white-sands-locality-2-resolved · 5 August 2024 (online, Crossref); the USGS record is dated 4 August 2024. Quoted from the abstract on the USGS publication record.
  11. Supporting evidence. Vance T. Holliday, Jason D. Windingstad, Jordon Bright, Bruce G. Phillips, Joel B. Butler, Ryan Breslawski and James E. Bowman. 2025. Paleolake geochronology supports Last Glacial Maximum (LGM) age for human tracks at White Sands, New Mexico. Science Advances 11(25), eadv4951. DOI 10.1126/sciadv.adv4951. https://pmc.ncbi.nlm.nih.gov/articles/PMC12175891/ · 18 June 2025 (online). CC BY-NC 4.0. Quoted from the full text.
  12. The continuing objection. Dave Rachal and Robert Dello-Russo. 2025. Seeds of controversy: Ecology, depositional context, and radiocarbon dating of Ruppia cirrhosa at the White Sands trackway. Journal of Archaeological Science 179, 106232. DOI 10.1016/j.jas.2025.106232. https://doi.org/10.1016/j.jas.2025.106232 · July 2025 issue. Cited by title only: the publisher page refused automated retrieval (HTTP 403) and no abstract was available in Crossref, so nothing is quoted from it.
  13. The latest official synthesis found. Kathleen B. Springer, Jeffrey S. Pigati, David Bustos, Thomas M. Urban and Matthew R. Bennett. 2026. Fossil footprints and Ice Age ecosystems of White Sands National Park. U.S. Geological Survey Fact Sheet 2025-3046. DOI 10.3133/fs20253046. https://pubs.usgs.gov/publication/fs20253046/full · 7 May 2026. Its series number is 2025-3046; it was published on 7 May 2026.
  14. The calibration curve. Paula J. Reimer and 41 coauthors. 2020. The IntCal20 Northern Hemisphere Radiocarbon Age Calibration Curve (0-55 cal kBP). Radiocarbon 62(4), 725-757. DOI 10.1017/RDC.2020.41. https://doi.org/10.1017/RDC.2020.41 · 2020. CC BY 4.0. Curve file: https://intcal.org/curves/intcal20.14c

The two USGS releases state that the product is for the most part in the public domain and may contain copyrighted items; this page ships only narrow numeric extracts of the fields it needs. Article text appears here only as the short attributed quotations above. The full terms are in NOTICE.txt.