Color of Small Ochre Fragments from the Upper Paleolithic Sites Kapova Cave and Kamennaya Balka II (Russia): Combining Visual Color Identification and Cluster Analysis

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper presents an interesting and original approach with a clearly documented and well-structured methodology. It is a considerable work which deserves to be published and will likely have a significant impact on the journal’s readership.

However, the paper has shortcomings that need to be addressed to enhance the clarity and relevance of the findings. These revisions will help the reader better understand the results' purpose, validity, and potential applications.

First, the introduction needs to explicitly state the research question and clarify why the proposed approach is an improvement over established methods of colour determination, in the specific context. The authors reference the work by Elias et al. (2006), which clearly explains how the colour of iron-based earth pigments is derived and what it depends on. In contrast, this paper does not make it clear what specific factors influence the colour of ochre. The discussion of colour should be framed in terms of hue, saturation, and brightness for better clarity.

Additionally, the authors should not overlook that diffuse reflectance spectroscopy is widely regarded as the most accurate method for colour determination. This technique provides information about the nature of the chromophore compound. Then, by analyzing colour in the L*a*b* colour space,  we can understand how the shade is modulated and varied by the purity of iron oxides, the composition of non-colour-bearing minerals, and the morphology (grain size and aggregate structure) of the pigment, which offers insights into the processing of raw materials. The methodology presented in the paper should be rooted in this understanding, with a clear explanation of its practical advantages and limitations.

Regarding diffuse reflectance spectroscopy, the authors state “ One of the main drawbacks of this approach is the necessity of preparing flat samples.”  However, this is not entirely accurate. This is not a real drawback of the method. The real limitation is that portable, handheld instruments, which are typically used on-site, require a flat surface of around 3-4mm in diameter. In many cases, particularly with small or irregularly shaped samples, this can be challenging. However, different instrumental configurations such as fibre-optic devices or hyperspectral imaging techniques can provide equivalent reflectance spectra of samples of any size without the need for a flat surface, overcoming the above limitation.

Line 59-60: Further clarification is needed in the following passage: “The content of goethite (yellow ochre) was correlated with the b * values, and the addition of hematite (red ochre) increased the a* value of the colour.”  This is not accurate, the referenced paper states that  “for differentiating the ochres by a colorimetric parameter, a^* co-ordinate appears more discriminatory and thus more relevant than b^*”.

Line 136: “The pigment sample was placed on the microscope sample table”, stage is the appropriate term.

In terms of scope, while the results and discussion convincingly validate the proposed methodology for classifying ochre pigment samples based on visual comparative colour determination (using the Munsell Soil Colour Charts), they do not align with the paper's title. The paper pertinently discusses how the colour-clustered data might inform the correlation of samples belonging to the same colour cluster in terms of their possible interpretative association and contextualization. Equally interesting conclusions have been drawn from the interpretation of the significant colour homogeneity of the samples in the case of the Kamennaya Balka II site, which indicates a uniform cultural affiliation of the population, but also opens up new research perspectives for identifying the origin of these materials, which form a homogeneous corpus.   

In the contrary, the title suggests a broader discussion of samples' colour characterization, but there is no comprehensive characterization of representative samples from each colour cluster.  The authors should either revise the title to better reflect the content of the paper or include additional data on the characterization of the colour of the studied samples.

 

Finally, while the conclusions suggest that further comparison between colour data and the chemical composition of pigments could be explored in future work (lines 363-365), this analysis is notably absent from the current paper. Although it is understood that such analytical data may not yet be available, it would be valuable to include an analysis of the L*a*b* colour coordinates for (at least the representative samples from) each class/cluster. This would help to discuss the nature of the chromophore, the purity, the content of other minerals and the visual homogeneity of the sample under a stereomicroscope, which are the main factors that determine the hue, saturation/chroma and lightness, the perceptual physical properties of colour.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The article presents an interesting investigation on the issues of reproduction and the study of color in rock paintings. The methodology and results are clear, but for publication, the researchers must address some points and complete the information provided.

1. Introduction The authors should provide the following information:
• Have the paintings under study been restored? The application of restoration products can also affect the color of pigments. The authors should indicate if these paintings have undergone restoration processes.
• Page 2, Line 54: How many samples were analyzed in the 2007 study reviewed by the authors?
• Page 2, Lines 69-73: The research and techniques described are unclear. The authors should elaborate on this information.
• Page 2, Line 74: Have the authors consulted similar studies, such as those on Roman wall painting or other mural surfaces?
2. The Archaeological Sites
• The authors should include a figure showing the analyzed locations, with a general image and visible paint traces. The authors should indicate the spaces within the cave with a schematic diagram.
3. Materials
• The authors should include two high-resolution images representing the analyzed samples, clearly showing the paint traces analyzed:
• One representative image of Kapova Cave.
• One representative image of Kamennaya Balka II.
• The authors should specify the sampling procedure, container, drying method, labeling, and sample size.
4. Results
• The visual determination of the color of the sample using Munsell Charts and an optical microscope is a procedure based on the subjectivity of the observer. Have the authors considered an alternative method to address this issue? Do the authors consider this a disadvantage for the success of the research?
• Page 7, Line 216: Include the figure number: "The upper part of the figure shows a bar chart corresponding to the 'branches' of the..."
• Page 7, Lines 230-231: Include the figure number: "In the lower part of the figures (on the tree diagram) the average colors for each of the three clusters A, B, and C are also shown as circles."
5. Conclusions
• It would have been very interesting for this article if the authors had conducted chemical analysis of the samples taken. The color of the sample is undoubtedly determined by its chemical composition, so a more in-depth study of the chemical composition of the pigments present should be planned.

Comments for author File: Comments.pdf

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript “Characterization of the Color of Small Ochre Fragments from the Upper Paleolithic Sites Kapova Cave and Kamennaya Balka II” presented a colorimetric approach to classify pigment samples from two Upper Paleolithic sites in Russia. Hundreds of ochre pigments (for a total of 867 fragments), excavated in the two archeological sites, were visually observed under the stereomicroscope, which allowed a reproducible lighting set-up, to determine the color by the Munsell color chart. Munsell values were then converted to CIELab coordinates and finally subjected to cluster analysis. Results of this approach allowed the classification of the ochre fragments into three (arbitrary) distinct groups for each archeological site. On the basis of colorimetric features of the ochre fragments, which in turn have originated distinct clusters, the authors investigated possible relationships among chronologically different settlements, and the relationships between a color palette and its use in pictorial representations.

The study is interesting  and it can be useful in the field because it proposes a noninvasive and affordable method to distinguish ochre samples, but at the same time I have some doubts about the reliability of the approach. This is why I recommend resubmitting the work after adding a validation of the method.

Some major concerns are reported below:

11-      The color of mineral pigments is primarily due to its mineralogical composition and thus I think that authors must include this central information. The colorimetric method proposed by the authors needs to be validated by other reliable techniques: e.g. FITR and Raman spectroscopies, XRD, etc.. Difference in the mineralogical composition of the ochre samples can imply different supply sources of raw material; difference in the mixture of pigments to produce a specific color or hue can imply an improved technical capability in mural painting or an increment of the color palette, and so on. Therefore, I believe it is essential to address this chemical/mineralogical aspect to support the conclusion.

22-    It is often unclear the subject of the study: whether pigment samples (that means powder or chunk of raw pigment), ochre samples (powder or raw earth pigment), or fragments from mural paintings. I think this point should be clarified for the reasons explained above in point 1.

33-      The authors stated in lines 219-220: “There are no strict rules for choosing the "cutoff" level, that is, the difference in color that we consider sufficient to divide samples into clusters. It is recommended to strike a balance between the final number of clusters and the variability of the sample characteristics within a cluster.” I was wondering whether this method can be useful to compare different sets of data. Maybe it would be better to arrange all data in one dataset and to process it with cluster analysis? Although still “subjective”, this way should provide a comparison between the two caves, and it should be less affected by arbitrary cutoff levels.

44-      Loading and scores should be commented on PCA scatter plots.

55-      Lines 362-363 “it can be concluded that several sites of the Upper and Middle Paleolithic showed a relationship between the shades of pigment and minerals in the paint composition”, as commented above in point 1, mineralogical composition has not been addressed and thus the method lacks validation.

66-      Lines 366-367 “Such studies may provide additional information about the role of color and related properties of ochre for Paleolithic human.” along all the manuscript this information remained quite unclear. The authors should clarify this relevant relationship and the significance of the study.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

Comments and Suggestions for Authors

Manuscript ID: heritage-3284884

Title: Characterization of the Color of Small Ochre Fragments from the Upper Paleolithic Sites Kapova Cave and Kamennaya Balka II

 

The manuscript reports the results of an interesting methodological approach for the characterization of the color of small pieces of ochre from the Kapova Cave and Kamennaya Balka II Upper Paleolithic sites (Russia).

The methodology includes color determination using a stereomicroscope and Munsell Soil Color Chart, conversion of CIE L*a*b* coordinates, cluster analysis and PCA.

Although the methodological approach is not fully new (e.g., Bouillot et al., 2017; Cuadros et al., 2020), in my opinion the main value of the present study is represented by the application to a significant number of pigment samples to identify artistic practices in parietal cave.

The manuscript is well-written, properly structured and pleasant to read. However, some very minor fails have been noticed (see the attached annotated manuscript).  

 

 

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have satisfactorily addressed all the review comments. The revised title is appropriate!

 

Some additional comments to the revised manuscript.

Line 59: “...which it is difficult to compare their color with the….”

Line 63: “…such as CIE L*a*b*.

Conversion of the sample color from the Munsell coordinates to the CIE L*a*b*

The Conversion of the sample color from the Munsell coordinates to the CIE L*a*b* is an important step in the proposed methodology. As mentioned in lines 91-93 and 208-210.

Line 91-93: The approach also includes the conversion of data from Munsell coordinates to CIE L*a*b* coordinates, making it possible to use cluster analysis to search for patterns in the obtained data.

And

Line 208-210. Conversion of the sample color from the Munsell coordinates to the CIE L*a*b coordinates using the Colour library. Color data were also approximately converted to RGB 209 coordinates for plotting (for illustrative purposes only).

From the paper it is assumed that this step has been accomplished using the software referenced in Ref [22] Mansencal, T.; Mauderer, M.; Parsons, M.; Shaw, N.; Wheatley, K.; Cooper, S.; Vandenberg, J.D.; Canavan, L.; Crowson, K.; Lev, 495 O.; et al. Colour 0.4.2 2022.

Is this correct?  It is not clearly stated. However, an additional specific reference is required to document the conversion method, which is not straightforward or obvious.

Basically, the publication that specifically describes the conversion method used.

Another publication on this topic is found in Vodyanitskii, Y.N., Kirillova, N.P. Conversion of Munsell color coordinates to Cie-L*a*b* system: Tables and calculation examples. Moscow Univ. Soil Sci. Bull. 71, 139–146 (2016). https://doi.org/10.3103/S0147687416040074

Also, please consider that there is updates for ref [22], the newest is Published October 11, 2024 | Version 0.4.6

 

Line 394: “…a relationship between the colors of pigment and minerals in the paint 394 composition…” This sentence is not clear.

 

Line 408: “coloristic” please revise the term…

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

I thank the authors for the revision and rebutal to the comments.

In my opinion, the method still requires adequate validation. However, the authors have clarified that this study represents an initial phase of research, with a more detailed analytical study planned for the near future.  The updated title now better reflects the study's objective.

Major concerns are reported below.

1)      In the introduction the authors should explain the types of information and relevance that can be gained from analyzing the color of ochre samples. In this view, the authors should clarify what are the “cultural layers” and what is the relationship between the color and cultural layer. This would help readers understand why the color is significant in the study, describing how the color of ochre samples is connected to or indicative of the cultural layers, possibly in terms of dating, usage, or cultural practices.

 

2)      I still have concerns about cluster analysis. The criteria to select the cut-off level should be better discussed. At page 7 the author said “There are no strict rules for choosing the "cutoff" level, that is, the difference in color that we consider sufficient to divide samples into clusters. It is recommended to strike a balance between the final number of clusters and the variability of the sample characteristics within a cluster. We have chosen this level in such a way that for both data sets (for the Kapova Cave site and for Kamennaya Balka II site), the final number of clusters was equal to three. The cut–off level for the data for the Kapova Cave was established at 85, and for the Kamennaya Balka II site was established at 200. So, the differences between the three clusters for the Kamennaya Balka II site are significantly greater than for the Kapova Cave. This is due to both the greater heterogeneity of the set of samples from the Kamennaya Balka II site and its greater number compared to the set of samples from the Kapova Cave”. I believe that the a priori choice to obtain 3 clusters must be justified by the dataset. Perhaps three macroscopically distinct color types are observed? For instance, a quite numerous number of samples collected from Kamennaya Balka II site seem much more yellowish than the other site.  Or could more clusters be created instead of three clusters? It is important to specify this point because the results of the research are based on the homogeneity/variability of clusters recognized in each archaeological site.

 

3)      I appreciate the plot provided in response to my earlier suggestion to combine data from the two caves into a single dataset and analyze it using cluster analysis. However, my recommendation was to apply agglomerative clustering methods (as performed on pages 6 and 7 of the revised manuscript) to the entire dataset, which could help determine a less arbitrary cut-off level. This involves using the entire dataset to determine a cluster distance based on criteria justified by the data itself (as commented earlier). In this way, the plot reported in response to the reviewer should report the clusters (cluster A, B, C, etc) instead of the archaeological site (cluster site 1, site 2). Using the full dataset may increase the variability of the set bringing to different number of clusters.

 

4)      While not the primary goal of this study, the approach proposed earlier could permit a direct comparison between the two caves. In fact, as the authors commented “the color of the ochre samples from the two sites is significantly different, but there is also some “common” part”. Could this be a potentially informative result worth discussing?

 

5)      The results of the study need to be better explained. The authors should clarify what is the connection between the variability (rather than homogeneity) of color of ochre samples recorded on a site and the cultural layers. As recommended earlier for the introduction, this result should be clarified possibly in terms of dating, usage, or cultural practices. Also, the authors need to provide some examples/descriptions/hypothesis of “activities”, “practices” and “feature of the structure of the studied sites”, “variety of direction of their use”, etc.. The use of these terms is too generic and unclear, as reader I would like to understand what they are meaning in the context of the research. Explaining well this point is also essential to highlight the relevance of the study.

 

6)      I think that the main discussion of the results is reported in line 374-384. I encourage the authors to further clarify the paragraph and restate the concept in the conclusion section.

 

7)      The conclusion should more effectively highlight (in a clear and accurate way) the relevance of the entire study. For instance: in lines 387-391, what is the significant additional information about the features of the structure of the studied sites supported by the color diversity of pigments? Other examples: in lines 398-418, what is the meaning of such differences or similarities? What do we have concluded about “the important additional information about the diverse household and symbolic practices carried out on monuments, as well as to give impetus and direction for advanced study of the issues of obtaining mineral raw materials, methods of its processing, and use both on individual sites and in regional and time comparison.

 

I believe that both the introduction and conclusion of the manuscript should be significantly revised to offer a clear overview of the study's relevance and the types of information derived from it. Moreover, the results and discussion part could be improved in reliability and discussion. Revision of English is recommended as well.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf