Revisiting Albarracín Rock Art Through Multivariate pXRF Analysis of White, Black, and Red Pigments^†

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The submitted manuscript entitled “Revisiting Albarracín Rock Art Through Multivariate pXRF Analysis of White, Black, and Red Pigments” is, in terms of dataset volume, methodological design, advanced analytical procedures, and its attempt to understand pigment technologies in Levantine art, a strong and valuable contribution. The use of a large sample set of 102 measurements, the combined pXRF approach with statistical and multivariate analyses, significantly enhances the quality of the work.
However, the manuscript is generally long, highly data-driven, and heavily statistical, and in several parts it would benefit from clearer synthesis, simplification, and stronger connection to archaeological questions.

A number of comments and suggestions for improvement are listed below:

* The analyses are very thorough, but the manuscript rarely links the results to archaeological interpretation. It should be clarified what these geochemical differences actually imply regarding painting sequences, workshop practices, or raw material choices. Are these differences indicative of different painting episodes, distinct raw material sources, or simply related to variations in pigment layer thickness?
I believe the interpretive/archaeological layer within the Discussion needs to be strengthened. As it stands, the results read more like a geochemical report than a paper on Levantine art.

* In the pXRF data, some elements—such as Si, Mg and Sr—are highly influenced by the rock substrate. The authors mention this, but in practice a large portion of the spatial discrimination between shelters appears to be substrate-driven rather than pigment-driven. Please consider this challenge more explicitly.

* Please emphasise the statistical limitations for shelters with very small sample sizes; in some cases fewer than three samples are available.

* Was any surface cleaning performed before XRF measurement? I did not see this clearly described. Please clarify this aspect.

* Regarding the identification of carbon of possible bone origin: the presence of phosphorus alone is not sufficient. Phosphorus may also derive from substrate materials or degradation processes. Please express this interpretation with more caution and clearly mention these alternative possibilities.

* The manuscript is excessively long; I recommend condensing certain sections, especially within the statistical results.

* In the Discussion section, some data are repeated unnecessarily. This part should focus more strongly on theoretical implications and cultural interpretations.

Author Response

• RESPONSE TO REVIEWER 1

   

  Reviewer 1 – Comment Q1

  "The analyses are very thorough, but the manuscript rarely links the results to archaeological interpretation. It should be clarified what these geochemical differences actually imply regarding painting sequences, workshop practices, or raw material choices. Are these differences indicative of different painting episodes, distinct raw material sources, or simply related to variations in pigment layer thickness? I believe the interpretive/archaeological layer within the Discussion needs to be strengthened. As it stands, the results read more like a geochemical report than a paper on Levantine art."

  Response: We have substantially revised the Discussion (specifically Section 4.6, now titled "Implications for Understanding Prehistoric Painting Practices") to explicitly address these points. We now argue that the compositional heterogeneity observed within individual shelters (e.g., the coexistence of Mn-rich and Mn-poor black pigments at Ceja de Piezarrodilla) indicates distinct painting episodes or the use of multiple "recipes" over time, rather than a single standardized workshop tradition. Conversely, the homogeneity at Casa Forestal suggests a more coordinated raw material procurement strategy. We also clarify that the specific enrichment patterns (Fe in reds, P in blacks) exceed what could be attributed to layer thickness variations alone.

  Manuscript changes:

  • Abstract: Revised the final sentences to emphasize these archaeological implications regarding technological heterogeneity in Levantine art.
  • Discussion, section 4.6: Complete rewrite to focus on technological choices (recipes) and procurement strategies.
  • Conclusions (final paragraph): Revised to mirror Abstract emphasis on archaeological implications rather than purely methodological achievements.

   

  Reviewer 1 – Comment Q2

  "In the pXRF data, some elements—such as Si, Mg and Sr—are highly influenced by the rock substrate. The authors mention this, but in practice a large portion of the spatial discrimination between shelters appears to be substrate-driven rather than pigment-driven. Please consider this challenge more explicitly."

  Response: We agree that the dominance of substrate-related elements in spatial discrimination patterns requires central, explicit treatment throughout the manuscript. We have added a dedicated explanation in the Introduction and Methods clarifying that elements like Ba, Sr, Si, and Mg are treated as "provenance markers" (reflecting the local geology of the shelter), while Fe, Mn, and P are treated as "pigment markers" (reflecting anthropogenic addition). We explicitly state that the high classification accuracy in the Linear Discriminant Analysis (LDA) is largely driven by this geological fingerprinting of the substrate, which confirms the autochthonous nature of the rock art but should not be confused with pigment recipe differences.

   

  Manuscript changes:

  • Introduction: Added a statement of pXRF limitations.
  • Methods, new section 2.4: Added a paragraph classifying elements into "Substrate-dominated" (Ba, Sr, Si, Mg) and "Pigment-informative" (Fe, Mn, P).
  • Discussion, section 4.2: Clarified that the large effect sizes (Cohen’s d) for Ba and Sr reflect geological variation between shelters, not necessarily pigment technology.

   

  Reviewer 1 – Comment Q3

  "Please emphasize the statistical limitations for shelters with very small sample sizes; in some cases fewer than three samples are available."

  Response: We agree that explicit emphasis on sample size constraints and their statistical handling is necessary. The small sample sizes in certain groups reflect our strategy to avoid artificial replication (sampling the same figure multiple times) and the scarcity of certain colors at specific shelters. We have added a section detailing how these small groups were handled: they were excluded from inferential parametric tests (ANOVA/MANOVA) to maintain rigor, but were retained in exploratory multivariate analyses (PCA) where they provide valuable qualitative data.

   

  Manuscript changes:

  • Methods, section 2.3: Added justification for the sampling strategy and the exclusion of n<3 groups from parametric testing.
  • Methods, section 2.5.1: Clarified dual-tier statistical approach.
  • Discussion, section 4.5: Added paragraph on statistical limitations.

   

  
  

   

  Reviewer 1 – Comment Q4

  "Was any surface cleaning performed before XRF measurement? I did not see this clearly described. Please clarify this aspect"

  Response: No surface cleaning was performed. Measurements were taken on the rock surface "as is". This decision was intentional to preserve the integrity of painted surfaces and any potential biofilms or patinas that might provide chronological or environmental information. We have added an explicit statement to this effect.

   

  Manuscript changes: Methods, section 2.3: Added an explicit statement.

   

  Reviewer 1 – Comment Q5

  "Regarding the identification of carbon of possible bone origin: the presence of phosphorus alone is not sufficient. Phosphorus may also derive from substrate materials or degradation processes. Please express this interpretation with more caution and clearly mention these alternative possibilities."

  Response: We agree that P enrichment alone cannot definitively establish bone char origin without supporting evidence from complementary techniques. We have rephrased the text to state that the P-Ca association is "consistent with" or "suggestive of" bone-derived materials (hydroxyapatite), while acknowledging alternative sources such as environmental contamination (guano) or substrate accessory minerals. We have also added a "Future Directions" section proposing Raman spectroscopy to definitively confirm the carbon species.

   

  Manuscript changes:

  • Results & Discussion: Changed phrasing throughout from definitive to qualified.
  • Discussion, Section 4.4: Added a specific discussion on alternative phosphorus sources.

   

  Reviewer 1 – Comment Q6

  "The manuscript is excessively long; I recommend condensing certain sections, especially within the statistical results"

  Response: We appreciate the reviewer’s concern about manuscript length and have implemented strategic condensation while preserving essential information for reproducibility and transparency. However, we respectfully note that the journal does not have page limits for online publication, and the extensive Supporting Information is necessary for data reusability—a core principle of open science. The revision focuses on improving navigability and interpretive emphasis rather than radical shortening. We have condensed the descriptive text in the Results section, reducing the word count from 4,835 to 2,386 words (a ~51% reduction). We retained the statistical output tables in the main text to ensure immediate accessibility of the quantitative data, focusing the text instead on the key patterns necessary for the narrative.

   

  Manuscript changes: Results: Condensed Sections 3.1–3.3.

   

  Reviewer 1 – Comment Q7

  "In the Discussion section, some data are repeated unnecessarily. This part should focus more strongly on theoretical implications and cultural interpretations."

  Response: We have removed the repetition of numerical results (means, standard deviations) in the Discussion, replacing them with interpretive synthesis and references to the relevant tables.

  .

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript presents an interesting and timely contribution to the study of rock art, effectively connecting this work with several recent publications and contributing to a deeper understanding of Paleolithic artistic practices. Although I am not an expert in chemometrics and therefore cannot fully assess the methodological framework, the general approach appears promising, particularly in its capacity to process large datasets.

However, the paper suffers from several weaknesses that require careful consideration before a final decision can be made. First, the manuscript is very long and at times difficult to follow. The supplementary materials are extensive—over 20 pages—and in practice constitute a second paper. As a result, the overall organization risks becoming unclear as the reader attempts to follow the connections among datasets, analyses, and results.

Furthermore, the range of chemical elements considered in the discriminant analysis is not always easy to understand, nor is the distinction between those elements used for discrimination and those attributed to the background signal. Although the authors state that the background contribution was subtracted based on measurements from unpainted stone, the methodological flow is difficult to follow. A major concern relates specifically to the sampling of unpainted surfaces: according to the supplementary table, it appears that only one background measurement per cave was collected. Given that stone surfaces are often subject to alteration and degradation, a more robust statistical basis may be required for reliable background subtraction.

Additionally, the Discussion and Conclusion sections would benefit from a synthesizing figure or schematic that helps guide the reader toward the main findings. Considering that the authors analysed 102 pigments across several shelters, it is challenging to grasp the overall implications without a clear, integrative summary. Some technical details—particularly those repeated systematically for each colour or sample—could be moved to the supplementary materials, while key interpretive information might be more effectively incorporated into the main text.

Author Response

RESPONSE TO REVIEWER 2

 

Reviewer 2 – Comment Q1

"First, the manuscript is very long and at times difficult to follow. The supplementary materials are extensive—over 20 pages—and in practice constitute a second paper. As a result, the overall organization risks becoming unclear as the reader attempts to follow the connections among datasets, analyses, and results."

Response: We have improved the navigability of the manuscript by adding a "Guide to Supporting Information" at the beginning of the SI file and improving cross-referencing in the main text. We maintain the extensive SI to ensure reproducibility (Open Science), but we have streamlined the main text to focus on the narrative.

 

Reviewer 2 – Comment Q2

"Furthermore, the range of chemical elements considered in the discriminant analysis is not always easy to understand, nor is the distinction between those elements used for discrimination and those attributed to the background signal. Although the authors state that the background contribution was subtracted based on measurements from unpainted stone, the methodological flow is difficult to follow. A major concern relates specifically to the sampling of unpainted surfaces: according to the supplementary table, it appears that only one background measurement per cave was collected. Given that stone surfaces are often subject to alteration and degradation, a more robust statistical basis may be required for reliable background subtraction."

Response: This comment addresses a critical methodological point. We wish to clarify two points:

• No mathematical subtraction: We did not perform an arithmetic subtraction of the background spectrum (which can introduce errors). Instead, we used the substrate values as a qualitative baseline to assess enrichment.
• Sampling: The values in Table S2 refer to the mean value derived from triplicate acquisitions on three representative unpainted areas at each shelter (i.e., n = 9). We have clarified this in the text. We have also updated the caption of Table S2 in the Supporting Information to explicitly state that the reported values represent the mean of n=9 measurements per shelter.

 

Manuscript changes:

• Methods, Section 2.3: Clarified that substrate values represent the mean of triplicate measurements on three representative unpainted surfaces and are used for comparative baselines, not mathematical subtraction.
• Discussion, Section 4.5: Added a note on the limitations of substrate sampling.
• Supporting Information, caption of Table S2: Now explicitly states that the reported values represent the mean of n=9 measurements per shelter.

 

Reviewer 2 – Comment Q3

"Additionally, the Discussion and Conclusion sections would benefit from a synthesizing figure or schematic that helps guide the reader toward the main findings. Considering that the authors analysed 102 pigments across several shelters, it is challenging to grasp the overall implications without a clear, integrative summary. Some technical details—particularly those repeated systematically for each colour or sample—could be moved to the supplementary materials, while key interpretive information might be more effectively incorporated into the main text"

 

 

Response: We agree that a visual synthesis is useful. We have designed a new integrative schematic figure (Figure 11) to synthesize the main patterns across pigment colors, shelters, and technological interpretations. This figure has been placed at the transition between Results and Discussion to serve as a visual roadmap for subsequent interpretation.

 

Manuscript changes: New Figure 11: "Integrative summary of pXRF analysis of 102 pigment samples from 12 Albarracín shelters”.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This manuscript reported Albarracín Rock Art Through Multivariate pXRF analysis of white, black, and red pigments, it is crucial to study the types of the involved pigments based on nondestructive analysis. However, it is inaccurate to study the mineral composition using pXRF. Some comments and suggestion are provided as follows.

1. The presence of Si (~3.7%) and Al (~1.1%) suggests a clay mineral component, it is difficult to state that it is kaolinite due to the high silicon-aluminum ratio, whether quartz is detected in the samples?
2. Kaolinite is a clay mineral instead of clay.
3. In order to confirm the mineral composition of the natural pigments, it should employ portable XRD diffractometer to study the mian mineral compositions combining with pXRF analysis.
4. What is the relationship between the color parameters and mineral composition in the involved pigments?

Author Response

RESPONSE TO REVIEWER 3

 

Reviewer 3 – Comment Q1

"The presence of Si (~3.7%) and Al (~1.1%) suggests a clay mineral component, it is difficult to state that it is kaolinite due to the high silicon-aluminum ratio, whether quartz is detected in the samples? Kaolinite is a clay mineral instead of clay."

Response: The reviewer’s critique is chemically valid and has prompted important revisions. The Si/Al ratio (~3.3) significantly exceeds the stoichiometric ratio for pure kaolinite (~1.18 by weight), indicating excess silica from quartz-rich substrate. We agree that more precise terminology is required. We have corrected the terminology to "clay mineral" and softened the identification to "aluminosilicates consistent with clay minerals (possibly of the kaolinite group)," explicitly mentioning the excess silica from the substrate.

 

Manuscript changes:

• Global text: Replaced "clay" with "clay mineral" where mineralogically appropriate, “kaolinite” with “aluminosilicate clay minerals (likely of the kaolinite group based on Al-Si association)”, and “presence of kaolinite” with “chemical signature consistent with kaolinitic clays”. Left the historical quote (Beltrán 1968) as "kaolin" to preserve the original citation accuracy, even though it is mineralogically imprecise.
• Discussion, section 4.1: Discussed the Si/Al ratio in the Cabras Blancas interpretation and acknowledged the substrate quartz contribution.

 

Reviewer 3 – Comment Q2

"In order to confirm the mineral composition of the natural pigments, it should employ portable XRD diffractometer to study the main mineral compositions combining with pXRF analysis."

Response: We fully agree that pXRD is the gold standard for phase identification. However, this equipment was not available for this study. We have framed the pXRF analysis as a high-throughput screening method that generates hypotheses to be tested in future campaigns using pXRD and Raman spectroscopy.

 

Manuscript changes:

• Discussion, Section 4.5: Added a "Limitations and Future Directions" subsection explicitly proposing pXRD and Raman to validate the chemical clusters identified here.
• Conclusion: Added a sentence on future directions.

 

Reviewer 3 – Comment Q3

"What is the relationship between the color parameters and mineral composition in the involved pigments?"

Response: The authors acknowledge that this is an important question in pigment studies. However, no quantitative colorimetry (L*a*b*) was performed; color classification was based on visual assessment during fieldwork. Therefore, we cannot provide a statistical correlation between color coordinates and element concentrations. We have noted this as a limitation and a future research line.

 

 

 

Manuscript changes:

• Methods, section 2.3: Explicitly stated that color categorization was visual and no colorimetric instrumentation was used.
• Discussion, section 4.1: Added a paragraph connecting inferred compositions to visual colors qualitatively
• Conclusions: Added colorimetry to future research recommendations.

 

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

The reviewed manuscript is evaluated, certainly, as novel, quite careful, informative and perfectly presented  research work corresponding to the scope and aims the Minerals journal. The research subject of Prehistoric rock painting is one of most intriguing and attractive themes of modern archaeological/ anthropological science. Therefore, the methodology of rock painting studies and analyses is of special scientific importance.  Reviewed manuscript contributes greatly to those methodological aspects of rock painting studies which are connected with problems of the recognizing and identification of natural pigments, raw material sources, technological and temporal patterns of pigments usage, etc. In general sence, this is the insight into the "laboratory" of primirive painters. 

The methodological block of this manuscript seems to have especial value and interest not only for the specialists in Prehistoric rock painting but for all scholars who use in their practice the portable XRF and are close familiar with the problems of measurements data interpretation. Mathematical approach applied to the results of pXRF examination of rock painting samples increase significantly the informative value of the research.

The manuscript's structure is quite logical. Research objectives are stated clearly, and research results form strong background for the suggested conclusions.

The writing style is clear, scientific, and easy for the understanding. The research design is fine, and Supplementary Materials contain the information needed to give exhausitive presentation of the research.

No doubts, this is excellent research. Its publication (recommended strongly) in Minerals would be certain scientific event. 

Author Response

RESPONSE TO REVIEWERS’S COMMENTS

 

 

Global revision strategy

The authors sincerely thank the three reviewers for their thorough and constructive evaluation of the manuscript. The reviews converged on several key areas that have guided our coordinated revision strategy:

• Archaeological interpretation: We have moved beyond the geochemical description to explicitly discuss technological heterogeneity, painting sequences, and workshop practices (rewriting Discussion section 4.6).
• Substrate vs. pigment signals: We have implemented a clearer distinction between "substrate-dominated elements" (Ba, Sr, Si, Mg) used for provenance/discrimination and "pigment-enriched elements" (Fe, Mn, P) used for recipe identification.
• Methodological clarifications: We have clarified that substrate measurements serve as qualitative baselines rather than for mathematical background subtraction, and we have explicitly acknowledged the limitations of pXRF for definitive mineral phase identification (softening claims about kaolinite and bone char).
• Structure and presentation: We have substantially streamlined the statistical results sections to address concerns regarding length. The textual word count of the Results section has been reduced by approximately 51% (from 4,835 to 2,386 words). We have substantially condensed the descriptive text in the Results section to streamline the presentation while retaining all statistical output tables in the main text for immediate reference. We have added a new integrative schematic (Figure 11) to visualize the findings.

Below is the point-by-point response to all comments.

 

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I thank the esteemed authors for their valuable efforts. The concerns raised during the review have been satisfactorily addressed.

Reviewer 3 Report

Comments and Suggestions for Authors

This manuscript can be accepted for publication at the present state.