REVIEW 4 major objections 6 minor 33 references
Dark Matter Induced Neutron Production Search Limits
T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash
Pith's one-line read Neutron-multiplicity data from a lead cube at 1166 m.w.e. show no dark-matter signal, yielding 90% upper limits near $\sim 10^{-45}\,\mathrm{cm}^2$ for spin-independent interactions from 300 MeV to 100 GeV.
desk verdict A genuinely new reanalysis of archived neutron-multiplicity data, but the printed limit equation inverts the target-area and A-scaling factors and the quoted cross sections do not follow. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The carrying object is the NMDS-II detector: a 305 kg, 30 cm lead cube surrounded by 60 $^{3}$He proportional counters embedded in polyethylene, which records the number of neutrons produced per interaction. Lead acts as a high-gain neutron amplifier, and the analysis models the cosmic-ray background as a power-law multiplicity distribution $k\,n^{-p}$, justified through a sample-space-reducing cascade mechanism in which each neutron emission shrinks the set of available nuclear states. A Monte Carlo simulation propagates muons from sea level through rock to the cavern and generates neutron multiplicity spectra from both muon showers and the two dark-matter models; the search then fits the 1166 m.w.e. data with the mixture $P(\beta;n)=(1-\beta)k\,n^{-p}+\beta S_{\mathrm{DM}}(n)$, and the 90% integration of the resulting likelihood yields the cross-section limits.
What would settle it
Record the neutron multiplicity spectrum at 1166 m.w.e. for about ten times the current 1440-hour exposure: the no-signal claim is falsified if high-multiplicity events ($n\ge 20$) appear at a rate incompatible with the simulated cosmic-ray-muon background. The complete-conversion assumption can be tested separately by measuring neutron yields from protons and pions of known energy in a 30 cm lead cube; if the true deposited-energy fraction is substantially below one, the reported limits must be scaled upward.
Extended reading notes
Core claim
The central claim is that the NMDS-II data contain no resolvable dark-matter signal: the most likely dark-matter fraction in the likelihood fit peaks at a significance around $1.5\sigma$, corresponding to about four excess events for deposited energies between 2 and 20 GeV, which the paper does not claim as a detection. Assuming complete hadronic energy conversion, the 90% upper limits are at the level of roughly $10^{-45}\,\mathrm{cm}^2$ for spin-independent interactions and $2\times10^{-42}\,\mathrm{cm}^2$ for spin-dependent interactions across the dark-matter mass range 300 MeV to 100 GeV. The two deliberately extreme final-state models, a single-proton spallation and a pion fireball with a limiting temperature, produce nearly the same limits, so the bound is not sensitive to the assumed hadronization pattern.
Load-bearing premise
The limits hold only if every dark-matter interaction deposits all of its rest-mass energy as hadrons inside the lead target; if the real interaction converts a fraction of that energy to neutrons, or emits mostly leptons or photons, the quoted cross-section bounds do not apply, and the six events at 1166 m.w.e. carry the entire statistical weight.
Editorial extensions
If this is right
- Spin-independent dark-matter-nucleon cross sections above about $10^{-45}\,\mathrm{cm}^2$ are excluded at 90% confidence for masses from 300 MeV to 100 GeV, under the complete-hadronic-conversion assumption.
- The near-agreement of the spallation and fireball limits means the result is stable across very different hadronic final states, so it does not hinge on the details of how dark matter converts to pions or protons.
- The $A^2$ coherent enhancement for spin-independent scattering is what lets a lead target set competitive limits at low dark-matter mass where direct-detection experiments lose sensitivity.
- The 583 m.w.e. data set, with about 36 times more muon-induced events, serves as a high-statistics control that validates the cosmic-ray background shape used in the 1166 m.w.e. analysis.
- If dark matter instead converts its energy mainly into leptons or photons, or deposits only a fraction of its rest mass in the target, these limits do not apply.
Reading between the lines
- Not tested in the paper: lowering the multiplicity threshold below five neutrons and using the 583 m.w.e. spectrum as a direct background template would extend the search to lower deposited energies, and the reach would then be limited by how well the power-law index is known rather than by the six-event sample.
- Because the 90% upper limit scales roughly as the inverse square root of the number of background events, a year or more of running at 1166 m.w.e. with the same detector would push the spin-independent bound below $10^{-45}\,\mathrm{cm}^2$ without any hardware change.
- A natural continuation would apply the same lead-cube multiplicity method to hidden-sector dark-matter models whose decay products include hadrons; the fireball simulation would need to be redone with the specific hidden-sector decay kinematics before quoting limits.
- The power-law background extrapolation could be checked by comparing the 583 m.w.e. data and simulation at multiplicities above 30, where a break in the power law would signal that the background shape used at 1166 m.w.e. needs revision.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper analyzes neutron multiplicity data from the NMDS-II detector, a 305 kg lead target instrumented with 3He counters, operated at two underground depths (583 m.w.e. for 6504 hours and 1166 m.w.e. for 1440 hours). The authors model cosmic-ray-muon-induced neutron backgrounds with Geant4, find that both data and simulation multiplicity spectra are well described by a power law, and search for an excess at high multiplicities (n >= 5) at the deeper site, where only six events were observed. Two dark-matter interaction models, a spallation model and a pion fireball model, are used to predict the neutron multiplicity distribution from dark-matter energy deposition in the lead target. A likelihood fit with a signal fraction beta is performed, and upper limits on the dark-matter-nucleon cross section are derived. The paper claims spin-independent limits near 1e-45 cm^2 and spin-dependent limits near 2e-42 cm^2 for dark-matter masses from 300 MeV to 100 GeV, assuming full hadronic conversion of the dark-matter rest mass.
Significance. If the quoted limits are correct, the paper would provide an independent indirect search for dark-matter interactions in the sub-GeV to 100 GeV mass range using neutron multiplicity, a channel that is complementary to conventional direct detection. The identical detector operated at two depths is a useful feature for validating the muon-induced background model, and the simulation effort, including muon propagation from sea level and full shower modeling, is substantial. The paper also validates its spallation simulation against external experimental data. However, the central cross-section conversion equation is incorrect as printed, and the statistical definition of the upper limit is non-standard; until these issues are resolved, the numerical limits cannot be accepted. The paper does not compare its results with existing dark-matter limits, which would be important for context.
major comments (4)
- [Section X, Eq. (14)] Equation (14) does not follow from Equations (12) and (13). From Eq. (13), the expected number of conversion events is Nevents = NDM * NPb * sigma_conversion / Starget, and substituting NDM = rho*v*t*Starget/mDM from Eq. (12) cancels Starget, giving Nevents = rho*v*t*NPb*sigma_conversion/mDM. Setting the 90% signal count to beta90*N1166 and solving for the per-nucleon spin-independent cross section yields sigma_chiN,90 = beta90*N1166*Starget / (NDM*NPb*A^2*epsilon). Equation (14) as displayed instead appears to write N1166 / [beta90 * (Starget/A^2) * NDM * NPb * epsilon], which inverts the beta factor (using N1166/beta90 rather than beta90*N1166) and places Starget in the denominator and A^2 in the numerator. Under this reading the expression has units of cm^-2, not cm^2, so it cannot be the correct conversion. Because every quoted limit in the abstract and Fig. 17 passes through this formula, the central numerical claim is unsupported as written. The authors must correct Eq. (14) and verify that the limits in Fig. 17 were computed with the corrected expression.
- [Section VII, Eq. (9), and Section X] The significance measure in Eq. (9) is non-standard. With Ssig = beta*N1166 and Bback = (1-beta)*N1166, the expression Ssig/sqrt(Ssig+Bback) equals beta*sqrt(N1166), not beta/sqrt(N1166) as the printed equality suggests; this is not a likelihood-ratio test statistic and does not account for the number of multiplicity bins, the look-elsewhere effect, or systematic uncertainties in the background shape. The 90% upper limit beta90 is defined by integrating the normalized likelihood over beta from 0 to 90%, which is a Bayesian credible interval with an implicit uniform prior on beta, not a frequentist confidence interval. The paper should state this statistical construction explicitly, including the prior, and ideally provide frequentist coverage checks, since the entire limit rests on only six events.
- [Section VI C and Section VII] The paper acknowledges in Section VI C that the Geant4 simulation and the data disagree in the amplitude k of the multiplicity distribution, and it sets this disagreement aside. However, the dark-matter likelihood in Eq. (7) uses the simulated background shape with the index p fixed to the Geant4 value (2.31 at 1166 m.w.e.), while the data are consistent with p = 2.13 +/- 0.37. The uncertainty in p and the normalization is not propagated into beta90 or the cross-section limits. Given the small event sample, the authors should either fit p and beta simultaneously in the likelihood or explicitly propagate the uncertainty in the background shape into the final limits, rather than treating the simulated p as exact.
- [Section VIII and Section X] The limits assume complete conversion of the dark-matter rest mass into hadronic energy in the lead target (f_average = 1). This assumption is stated, but the paper does not quantify how the limits scale with f_average or discuss the model dependence of the two bracketing hadronic models for dark-matter masses near threshold. The abstract states the assumption, but the conclusions should clearly flag that any model in which dark matter deposits only a fraction of its rest mass as hadronic energy, or produces different neutron multiplicities, would weaken the quoted limits. A simple scaling relation sigma ∝ 1/f_average, along with a short discussion of the validity range of the fireball model for masses below about 1 GeV, would make the caveat precise.
minor comments (6)
- [Table I] The header contains a typo: 'Miiyake' should be 'Miyake'.
- [Section VII, Eq. (9)] The displayed equality in Eq. (9) appears to have an algebraic slip: beta*N1166/sqrt(N1166) equals beta*sqrt(N1166), not beta/sqrt(N1166). Please correct the typo.
- [Throughout] There are numerous typographical errors, for example 'indepentently' in Section VI B, 'EXPERIMA T AL' and 'SIMUALTION' in the Section VI header, 'compered' in Section VIII, and 'curvture' in Appendix A. A careful proofread is needed.
- [Section X / Conclusion] The paper does not compare its limits with existing dark-matter constraints, such as direct detection limits at higher masses or other neutron-based searches. A brief comparison, even in a figure inset, would help the reader judge the significance of the result.
- [Section VIII B] The pion fireball model uses a Hagedorn temperature kT = 0.165 GeV taken from proton-nucleus interactions at 70 GeV; for dark-matter masses near the pion-production threshold (300 MeV to about 1 GeV), the applicability of this temperature and the Planck distribution should be discussed.
- [Section V A] The simulation uses Geant4-11.01 while the data were collected in 2001-2003; this is acceptable, but the paper should explicitly state that a modern Geant4 version was used to model the historical detector and that this version choice was validated against the 252Cf calibration.
Circularity Check
No significant circularity: the DM limits follow from a six-event likelihood compared with Geant4 signal and background shapes, not from the fitted inputs; the one self-citation is not load-bearing.
-
self citation load bearing
[Section VIII, Dark Matter Models, validation paragraph before Sec. VIII.A]
"For validation the spallation model simulation was compered to the NESSI(Neutron Scintillator and Silicon Detector) experimental results[27][28] in the range 1.2 GeV to 2.5 GeV. The experimental neutron production and the simulation are in agreement at the 10% level[29]."
Reference [29] is H. Cao's Purdue thesis, so the quoted 10% agreement is supported through a self-citation. However, this is not load-bearing circularity: the external NESSI data are given by [27,28], the Geant4 detector response is separately anchored to the 252Cf calibration, and the DM cross-section limits are obtained from the likelihood of Eqs. 7-9 and the acceptance simulation, not from the thesis result.
full rationale
The paper's central limit-setting chain is self-contained rather than circular. The background spectrum in the likelihood is the Geant4 cosmic-ray-muon neutron-multiplicity shape, checked against the NMDS-II power-law index and against independent muon-flux measurements (Miyake fits and CUPP data), while the total event count N1166=6 comes from data. The DM signal shapes are generated by propagating proton or pion final states in Geant4, with an efficiency correction tied to the 252Cf calibration. The 90% upper limit beta_90 is obtained by integrating the likelihood; converting that limit into a cross section uses Eqs. 12-14 with stated assumptions for rho_DM, v, t, N_Pb, Starget, and acceptance. None of these steps defines the answer as its own input: the no-excess result is a genuine comparison of data to a background-plus-signal model, and the quoted limits are not simply the fitted parameters renamed. The only notable self-citation is [29] for the spallation-model validation against NESSI; since [27,28] provide the external experimental data and this validation is not what forces the final limits, it is minor and non-load-bearing, so it does not raise the circularity score above 2. A separate algebra/consistency concern exists in Eq. 14: as written it does not correctly invert Eq. 13 and the Starget and A^2 factors appear misplaced. That is a derivation-error issue rather than a circular-reduction issue, so it is noted here but not counted as circularity.
Assumptions & free parameters
free parameters (5)
- Power-law background index p at 1166 m.w.e. =
2.31 +/- 0.01 (Geant4); data MLE 2.13 +/- 0.37
- Fireball Hagedorn temperature kT =
0.165 GeV
- Effective active 3He diameter =
1.33 cm
- Miyake muon flux normalization A =
(2.97 +/- 0.114) x 10^6 (m.w.e.) m^-2 s^-1
- Signal fraction beta =
beta_peak up to about 0.6 for 2 to 20 GeV; not significant
assumptions (6)
- domain assumption Geant4 QGSP-BERT-HP accurately describes muon and hadron induced neutron production in lead and standard rock.
- domain assumption Sea-level muon spectrum Eq. 2 propagated through standard rock reproduces the underground muon energy-angle distribution.
- ad hoc to paper All DM rest-mass energy is converted to hadronic energy inside the Pb target.
- ad hoc to paper The spallation and pion fireball models bracket physical DM-matter neutron production.
- ad hoc to paper Natural lead spin-dependent response scales as 1/A per nucleon.
- domain assumption Neutron multiplicity backgrounds follow a power law k n^-p with a common index.
Cite this review
Pith. "Pith review of Dark Matter Induced Neutron Production Search Limits." pith.science (2026). https://pith.science/paper/TV5M57JC
@misc{pith2026250622659,
author = {Pith},
title = {Pith review of: Dark Matter Induced Neutron Production Search Limits},
year = {2026},
howpublished = {\url{https://pith.science/paper/TV5M57JC}},
note = {Machine review of arXiv:2506.22659}
}
abstract
An independent indirect detection search for Dark Matter-Matter (DM-M) interactions is undertaken to set cross section limits based on neutron production data collected by the NMDS-II detector for 1440 hours at 1166 m.w.e. and 6504 hours at 583 m.w.e.. The detector system consists of a 30 cm cube Pb-target instrumented with 60 $^3$He neutron counters. The neutron detector system calibrated with a $^{252}$Cf source yields a single particle detection efficiency of 23.2\%$\pm$1.2\%. During data collection, the highest neutron multiplicity event observed 54 neutrons. The neutron multiplicity, n, distribution, fits well to a power law $k \times n^{-p}$, for both the data and cosmic ray muon induced neutron production in Geant4 simulations. Two DM-M interaction models were used to set limits. The first, a spallation model, assumes a single proton with kinetic energy equal to the DM-M interaction energy. The other, a fire-ball model assumes an annihilation between DM-M producing pions with a limiting Hagedorn temperature. The two extreme models produce similar upper DM-M cross section limits over the DM mass range between 300 MeV to 100 GeV. Limits assume all the DM energy is deposited in the Pb-target. Spin independent limits, proportional to A$^{-2}$, are at the level ~10$^{-45}~ cm^2$. Spin dependent limits, proportional to A$^{-1}$, are at the level, $2\times 10^{-42}~cm^2$.
Figures
Figures from the paper (11 more)
Reference graph
Works this paper leans on
-
[1]
P. D. Group, R. Workman, V. Burkert, V. Crede, E. Klempt, U. Thoma, L. Tiator, K. Agashe, G. Aielli, B. Allanach, et al., Review of particle physics, Progress of theoretical and experimental physics 2022, 083C01 (2022)
work page 2022
-
[2]
J. Aalbers, S. AbdusSalam, K. Abe, V. Aerne, F. Agos- tini, S. A. Maouloud, D. Akerib, D. Y. Akimov, J. Ak- shat, A. Al Musalhi, et al., A next-generation liquid xenon observatory for dark matter and neutrino physics, Journal of Physics G: Nuclear and Particle Physics 50, 013001 (2022)
work page 2022
- [3]
-
[4]
M. F. Albakry, I. Alkhatib, D. Amaral, T. Aralis, T. Ara- maki, I. Arnquist, I. A. Langroudy, E. Azadbakht, S. Banik, C. Bathurst, et al., A strategy for low-mass dark matter searches with cryogenic detectors in the su- percdms snolab facility, arXiv preprint arXiv:2203.08463 (2022)
arXiv 2022
- [5]
-
[6]
X. Cui, A. Abdukerim, W. Chen, X. Chen, Y. Chen, B. Dong, D. Fang, C. Fu, K. Giboni, F. Giuliani, et al., Dark matter results from 54-ton-day exposure of pandax- ii experiment, Physical review letters119, 181302 (2017)
work page 2017
-
[7]
X. Collaboration, E. Aprile, J. Aalbers, F. Agostini, M. Alfonsi, L. Althueser, F. Amaro, M. Anthony, F. Ar- neodo, L. Baudis, et al., Dark matter search results from a one ton-year exposure of xenon1t, Physical review let- ters 121, 111302 (2018)
work page 2018
-
[8]
M. Doro, J. Conrad, D. Emmanoulopoulos, M. A. S` anchez-Conde, J. Barrio, E. Birsin, J. Bolmont, P. Brun, S. Colafrancesco, S. Connell, et al., Dark matter and fun- damental physics with the cherenkov telescope array, As- troparticle Physics 43, 189 (2013)
work page 2013
Show all 33 references
-
[9]
Abbasi, M
R. Abbasi, M. Ackermann, J. Adams, S. Agarwalla, J. Aguilar, M. Ahlers, J. Alameddine, N. Amin, K. An- deen, G. Anton, et al., Search for neutrino lines from dark matter annihilation and decay with icecube, Physi- cal Review D 108, 102004 (2023)
2023
-
[10]
Adri´ an-Mart ´ ınez, A
S. Adri´ an-Mart ´ ınez, A. Albert, M. Andr´ e, G. Anton, M. Ardid, J.-J. Aubert, T. Avgitas, B. Baret, J. Barrios- Mart ´ ı, S. Basa,et al., Limits on dark matter annihilation in the sun using the antares neutrino telescope, Physics Letters B 759, 69 (2016)
2016
-
[11]
V. A. Kudryavtsev, L. Pandola, and V. Tomasello, Neutron- and muon-induced background in underground physics experiments, Eur. Phys. J. A 36, 171 (2008), arXiv:0802.3566 [astro-ph]
2008 arXiv
-
[12]
Lovchikova, A
G. Lovchikova, A. Trufanov, M. Svirin, V. Vinogradov, and A. Polyakov, Spectra and mean energies of prompt neutrons from 238 u fission induced by primary neutrons of energy in the region e n¡ 20 mev, Physics of Atomic Nuclei 67, 1246 (2004)
2004
-
[13]
Lestone, A comparison between two evaluations of neutron multiplicity distributions, arXiv preprint arXiv:1410.2148 (2014)
J. Lestone, A comparison between two evaluations of neutron multiplicity distributions, arXiv preprint arXiv:1410.2148 (2014)
2014 arXiv
-
[14]
G. E. Knoll, Radiation Detection and Measurement (John Wiley and Sons, Inc., 2000) Chap. 14, pp. 518– 519
2000
-
[15]
Shalev, Z
S. Shalev, Z. Fishelson, and J. Cuttler, The wall effect in the counters, Nuclear Instruments and Methods 71, 292 (1969)
1969
-
[16]
displayed in Figure 11
The search for excess neutron multiplicity production is conducted using the data at 1166 m.w.e. displayed in Figure 11. For the search, β is limited to the physical positive signal region [0,1] and θ is the signal significance. There are mass regions in Figure 16 in both mode...
-
[17]
For spin-independent cross sections, the sensitivity per nucleon requires dividing the factor of the atomic number A, twice
Spin dependent and independent limits based on the fireball and spallation models. For spin-independent cross sections, the sensitivity per nucleon requires dividing the factor of the atomic number A, twice. The first factor of A is simply due to the target nucleon number. The...
2000
-
[18]
S. W. J.J. Beatty, J. Matthews, Review of particle physics - cosmic rays, Progress of Theoretical and Ex- perimental Physics 2019, 083C01 (2019)
2019
-
[19]
KF, Cosmic rays at Earth: researcher’s reference man- ual and data book(Elsevier, 2001) Chap
P. KF, Cosmic rays at Earth: researcher’s reference man- ual and data book(Elsevier, 2001) Chap. 4, pp. 485–486
2001
-
[20]
Enqvist, A
T. Enqvist, A. Mattila, V. F¨ ohr, T. J¨ ams´ en, M. Lehtola, J. Narkilahti, J. Joutsenvaara, S. Nurmenniemi, J. Pel- toniemi, H. Remes, et al., Measurements of muon flux in the pyh¨ asalmi underground laboratory, Nuclear In- struments and Methods in Physics Research Section A...
2005
-
[21]
Khachatryan, A
V. Khachatryan, A. M. Sirunyan, A. Tumasyan, W. Adam, T. Bergauer, M. Dragicevic, J. Er¨ o, C. Fab- jan, M. Friedl, R. Fruehwirth, et al., Measurement of the charge ratio of atmospheric muons with the cms detector, Physics Letters B 692, 83 (2010)
2010
-
[22]
Kiryunin, H
A. Kiryunin, H. Oberlack, D. Salihagi´ c, P. Schacht, and P. Strizenec, Geant4 physics evaluation with testbeam data of the atlas hadronic end-cap calorimeter, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated ...
2006
-
[23]
R. W. Johnson, An introduction to the bootstrap, Teach- ing statistics 23, 49 (2001)
2001
-
[24]
Corominas-Murtra, R
B. Corominas-Murtra, R. Hanel, and S. Thurner, Understanding scaling through history-dependent pro- cesses with collapsing sample space, Proceedings of the National Academy of Sciences 112, 5348 (2015), https://www.pnas.org/doi/pdf/10.1073/pnas.1420946112
2015 doi
-
[25]
Fujii and J
K. Fujii and J. C. Berengut, Power-law intensity distribu- tion of γ-decay cascades: Nuclear structure as a scale-free random network, Physical Review Letters 126, 102502 (2021)
2021
-
[26]
Corominas-Murtra, R
B. Corominas-Murtra, R. Hanel, and S. Thurner, Sample space reducing cascading processes produce the full spec- trum of scaling exponents, Scientific Reports 7, 11223 (2017)
2017
-
[27]
Chupp, D
E. Chupp, D. Forrest, G. Kanbach, E. Flueckiger, and F. Golliez, Solar neutron emissivity during the large flare on 1982 june 3, Astrophys. J. 318 (1987)
1987
-
[28]
Bauche-Arnoult and J
C. Bauche-Arnoult and J. Bauche, Comparison of atomic data modeling with experimental intensities, Journal of Quantitative Spectroscopy and Radiative Transfer 58, 441 (1997), radiative Properties of Hot Dense Matter III
1997
-
[29]
Filges, F
D. Filges, F. Goldenbaum, M. Enke, J. Galin, C.-M. Herbach, D. Hilscher, U. Jahnke, A. Letourneau, B. Lott, R.-D. Neef, et al., Spallation neutron production and the current intra-nuclear cascade and transport codes, The European Physical Journal A-Hadrons and Nuclei 11, 467 (2001)
2001
-
[30]
Letourneau, J
A. Letourneau, J. Galin, F. Goldenbaum, B. Lott, A. P´ eghaire, M. Enke, D. Hilscher, U. Jahnke, K. N¨ unighoff, D. Filges, et al., Neutron production in bombardments of thin and thick w, hg, pb targets by 0.4, 0.8, 1.2, 1.8 and 2.5 gev protons, Nuclear Instru- ments and Metho...
2000
-
[31]
Cao, Indirect Dark Matter Search for Dark Matter, Ph.D
H. Cao, Indirect Dark Matter Search for Dark Matter, Ph.D. thesis, Purdue University (2023)
2023
-
[32]
A. Saha, M. Samanta, D. Pal, R. Nandy, and B. Ban- dopadhyay, Energy distribution of pions in the nucleon– nucleus interaction at 70 gev, Canadian Journal of Physics 65, 527 (1987)
1987
-
[33]
Lipari and T
P. Lipari and T. Stanev, Propagation of multi-tev muons, Physical Review D 44, 3543 (1991)
1991
Reviewed August 6, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.