Differentiation of Species and Provenance of Palm-Leaf Manuscripts Using Fourier Transform Infrared Spectroscopy and Chemometrics

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper successfully differentiates the species and geographical origins of palm-leaf manuscripts using Fourier Transform Infrared Spectroscopy (FTIR) combined with chemometric methods, demonstrating significant academic value. However, there are some issues in the interpretation of infrared spectra and chemometric statistical conclusions that require revision.

1. The number of cited references is limited, particularly regarding recent advances in FTIR applications to cultural heritage materials and chemometric methods for spectral discrimination.

2. The paper mentions significant spectral differences among manuscripts of different species and geographical origins in the PCA analysis, but it does not provide detailed statistical validation. It is suggested to supplement the analysis with t-tests or ANOVA to verify the statistical significance of these differences.

3. While the PCA analysis successfully distinguishes manuscripts based on species and geographical origin, the limitations of PCA are not discussed.

Author Response

We would like to thank the reviewers for thoughtful critiques of our manuscript. We have adopted the suggestions and think that the manuscript has been substantially improved by these revisions. We hope that you will now find it suitable for publication in Chemosensors. Our point-by-point responses to the comments are detailed below.

This paper successfully differentiates the species and geographical origins of palm-leaf manuscripts using Fourier Transform Infrared Spectroscopy (FTIR) combined with chemometric methods, demonstrating significant academic value. However, there are some issues in the interpretation of infrared spectra and chemometric statistical conclusions that require revision.

The number of cited references is limited, particularly regarding recent advances in FTIR applications to cultural heritage materials and chemometric methods for spectral discrimination.

We added the review article citation [5] which is about FTIR applications in the field of cultural heritage objects. Likewise the citations [13-16] for spectral discrimination and chemometric methods were added to acknowledge the (un)supervised methods that are typically used for these research questions and to show recent advances in the field of FTIR and chemometrics.

The paper mentions significant spectral differences among manuscripts of different species and geographical origins in the PCA analysis, but it does not provide detailed statistical validation. It is suggested to supplement the analysis with t-tests or ANOVA to verify the statistical significance of these differences.

T-test results were added in supplementary table S1 and incorporated in the text.

While the PCA analysis successfully distinguishes manuscripts based on species and geographical origin, the limitations of PCA are not discussed.

We revised the conclusion/discussion section and included the limitations of PCA / unsupervised approaches:

“Furthermore, the chemometric analysis of the manuscript spectra proved that inherent spectral differences could be observed for the taxonomic species of the used palm and the geographical origin, although a complete seperation of the analyzed groups is not always possible. TThis clear separation could be achieved by focusing the evaluation on specific differences using supervised chemometric methods, such as PLS-DA, such as PLS-DA, which were not applied here due to the small number of analyzed manuscripts.”

Reviewer 2 Report

Comments and Suggestions for Authors

The paper reports on the analysis of palm-leaf manuscripts using FTIR and statistically evaluating the obtained data with PCA. The paper is very well written, and the authors managed to distinguish the manuscripts produced from the leaves of two different palm species, and partly also to discriminate between the manuscripts produced in different geographical regions. All conclusions in the manuscript are very well confirmed, either through additional statistical test (Student) or by inspecting the differences between different spectra and PCA loadings, with additional arguments given in Supplementary Materials. The authors further discuss all possible reasons for which the manuscript materials may differ.

The paper is very concise, interesting and thoroughly written, and as such it is already in the state to be accepted in the present form. There is only a small typographic error in line 196 (zhe instead of the). For the readers, who are not that familiar with FTIR and related statistical methods (different from those described in ref. 6) please provide more details on the input data vectors and dimensions of the covariance matrix.

Author Response

We would like to thank the reviewers for thoughtful critiques of our manuscript. We have adopted the suggestions and think that the manuscript has been substantially improved by these revisions. We hope that you will now find it suitable for publication in Chemosensors. Our point-by-point responses to the comments are detailed below.

The paper reports on the analysis of palm-leaf manuscripts using FTIR and statistically evaluating the obtained data with PCA. The paper is very well written, and the authors managed to distinguish the manuscripts produced from the leaves of two different palm species, and partly also to discriminate between the manuscripts produced in different geographical regions. All conclusions in the manuscript are very well confirmed, either through additional statistical test (Student) or by inspecting the differences between different spectra and PCA loadings, with additional arguments given in Supplementary Materials. The authors further discuss all possible reasons for which the manuscript materials may differ.

The paper is very concise, interesting and thoroughly written, and as such it is already in the state to be accepted in the present form. There is only a small typographic error in line 196 (zhe instead of the). For the readers, who are not that familiar with FTIR and related statistical methods (different from those described in ref. 6) please provide more details on the input data vectors and dimensions of the covariance matrix.

The error was corrected and we added the exact data points per spectra that are used for PCA analysis:

“Hence the area from 1900 to 2400~cm^-1, which can show peaks from carbon dioxide of ambient air, was removed for chemometric analysis. This reduced the number of analyzed data points per spectrum to 1356.”

Reviewer 3 Report

Comments and Suggestions for Authors

This is a simple study looking to identify whether IR can separate manuscripts. 

I did not find any issue with the work. I would just ask the authors to clarify 2 points:

• t-test. what were the t-tests performed on? scores? That should be clarified
• The relevance of looking at the influence of oil should be further explained. How is a testing done today relevant to finding on century old manuscripts? Could the oil affect the manuscripts over time that the testing performed today cannot anticipate? Finally, authors did a simple spectral subtraction. A PCA test on these would show significant changes and so I am not sure that as written, this section is achieving its goal.

line 196 - zhe should be replaced by the

Author Response

We would like to thank the reviewers for thoughtful critiques of our manuscript. We have adopted the suggestions and think that the manuscript has been substantially improved by these revisions. We hope that you will now find it suitable for publication in Chemosensors. Our point-by-point responses to the comments are detailed below.

This is a simple study looking to identify whether IR can separate manuscripts.

I did not find any issue with the work. I would just ask the authors to clarify 2 points:

t-test. what were the t-tests performed on? scores? That should be clarified

We added this explanation in the material section to clarify that tests were applied to the scores of the PCA:

“For statistical testing, two-tailed, two-sample t-tests (function: t.test() with equal variances were applied to the scores of the respective component of a PCA. Although we assesed all resulting principal components from PCA analysis, those with less than five percent of variance were not reported (which usually concerned the fifth and higher principal components).”

The relevance of looking at the influence of oil should be further explained. How is a testing done today relevant to finding on century old manuscripts? Could the oil affect the manuscripts over time that the testing performed today cannot anticipate?

We added further explanation. The oiling is applied frequently to preserve the manuscripts. We added the following text in section 3.2.3:

“The application of oils is a typical repeated treatment for the preservation of palm leaf manuscripts.”

“From this very simplified imitation of the conservation process, it can be concluded that treatment with oil does not significantly alter the spectrum of palm-leaf manuscripts which greatly simplifies the application of DRIFTS as a fingerprinting approach for these manuscripts in practice. “

Finally, authors did a simple spectral subtraction. A PCA test on these would show significant changes and so I am not sure that as written, this section is achieving its goal.

The subtraction to see a difference spectrum is more a visual help. DRIFT is not especially equipped to quantify the spectral intensities, therefore we refrain from a more precise evaluation. We changed the text to a more cautious version to clarify this:

“From this very simplified imitation of the conservation process, ...”

line 196 - zhe should be replaced by the

it was corrected.

Reviewer 4 Report

Comments and Suggestions for Authors The manuscript describes and discusses logically designed information expected to be of great interest to the scientific community. It is a fascinating study with an exciting approach. The paper, on the whole, is well-designed, and the results sound. Nevertheless, the manuscript needs a minor revision:   Point 1: The paper's main aim should be highlighted in the introduction, and additionally, the novelty of carried research work should be discussed.   Point 2: How do the Authors select the subject? The rationale behind the choice is missing and should be discussed.   Point 3: The quality of the figures must be improved.  

Author Response

We would like to thank the reviewers for thoughtful critiques of our manuscript. We have adopted the suggestions and think that the manuscript has been substantially improved by these revisions. We hope that you will now find it suitable for publication in Chemosensors. Our point-by-point responses to the comments are detailed below.

The manuscript describes and discusses logically designed information expected to be of great interest to the scientific community. It is a fascinating study with an exciting approach. The paper, on the whole, is well-designed, and the results sound. Nevertheless, the manuscript needs a minor revision:

Point 1: The paper's main aim should be highlighted in the introduction, and additionally, the novelty of carried research work should be discussed.

We added the following sentence to the introduction:

“In this study, we analyze for the first time whether DRIFTS in combination with chemometrics is suitable for obtaining characteristic fingerprints of palm leaf manuscripts, that can be used to classify them regarding provenance and taxonomic identity”

Point 2: How do the Authors select the subject? The rationale behind the choice is missing and should be discussed.

We added the following sentence to the introduction:

“This approach was chosen for its ability to provide mobile, non-destructive analysis, making it ideal for fingerprinting cultural heritage objects.”

Point 3: The quality of the figures must be improved.

We ensured that the quality of the figures is adequate to the journal standards.

Reviewer 5 Report

Comments and Suggestions for Authors

Review of manuscript chemosensors-3610516 “Differentiation of Species and Provenance of Palm-Leaf Manuscripts using Fourier-Transform Infrared Spectroscopy and Chemometrics”

Summary

This study demonstrates the use of FTIR spectroscopy combined with chemometric analysis to differentiate palm-leaf manuscripts by species and geographical origin. The method is non-invasive, reproducible, and minimally affected by writing or conservation oils. Results show clear spectral differences between Borassus flabellifer and Corypha umbraculifera. Partial geographic separation is also possible. The results highlight the potential of FTIR fingerprinting as a valuable tool for manuscript authentication and codicological research.

Commentaries

In my opinion, this manuscript is well-structured and addresses a topic that will likely be of interest to the readers of Chemosensors. The research presents a valuable approach to the non-invasive analysis of historical manuscripts. However, the clarity and depth of the Methods and Results sections could be significantly improved. I believe the manuscript has strong potential for publication in Chemosensors, provided that the authors address several issues to enhance the reliability of the findings and the overall readability of the text.

Abstract

- Overall, the abstract is adequate, but it would benefit from more specific and precise writing that highlights the technical depth and significance of the study.

- Key methodological elements, such as the application of Principal Component Analysis (PCA) ,are not mentioned, despite being central to the research. In addition, including quantitative details, such as the number of manuscripts analyzed, would provide a clearer context for the relevance of the obtained results.

Introduction

- I would have appreciated a more detailed discussion of previous analyses of this type of manuscript, including what kinds of spectroscopic studies—either standalone or in combination with chemometric methods—have been conducted so far.

- The rationale for selecting DRIFTS over other IR-based techniques should be more clearly described. Also, the terminology should be used consistently throughout the manuscript.

- Bringing these points together, the authors should clarify why existing methods are insufficient for the analysis of palm-leaf manuscripts and explicitly state what the present study contributes beyond the current state of the art.

Methods

- I understand the difficulty in accessing this type of material, but do the authors not consider that a sample size of only 11 manuscripts is rather limited for proposing a new methodological approach? Additionally, the criteria for selecting these specific manuscripts should be clearly stated.

- Regarding the “direct inspection” used to assign certain manuscript parameters, more detail is needed. Some form of validation for these assignments would strengthen the reliability of the study. It would also be helpful to provide more information about the conservation and storage conditions of the manuscripts, as these factors could potentially influence the spectral data.

- The application of lemongrass oil is not entirely clear. Why was this oil chosen, and has it been commonly used in real conservation treatments? Has the potential interference of such substances with the spectral results been assessed?

- Finally, the chemometric analysis section should provide more detailed information regarding the preprocessing parameters and the statistical analysis performed.

Results

- The PCA models should include key performance indicators, such as the number of components used and the percentage of variance explained.

- What do the first principal components in Figure 6 actually represent? The use of PC3 and PC4 for interpretation may give the impression of selectively reporting favorable results—this could be seen as a form of cherry-picking.

- Some explanation is needed for the partial overlaps observed in Figure 6A. Additionally, the clear separation of one group of Indonesian manuscripts with more positive PC3 values deserves attention. Similarly, the more negative PC1 values for Sri Lankan manuscripts in Figure 6B raise questions—could these patterns be related to the way the samples were assigned to origin groups?

- It is curious that in Figure 5, multiple clusters appear within the Borassus group, whereas this internal grouping disappears when the dataset is split by species. A possible explanation for this inconsistency should be discussed.

- The variations observed in Figure 7 could potentially reflect differences in conservation treatments or manuscript degradation—this possibility should be considered in the interpretation.

Conclusions

- The conclusions should be completely rewritten, as they currently serve more as a brief summary than a reflective or critical synthesis. The authors should address key questions such as: What does this methodology contribute to the field? What are its main limitations? Are the results meaningful considering the relatively small dataset? Consider moving parts of the Discussion section—particularly those that interpret findings and implications—into the Conclusions, as they would be more appropriate and valuable there than the current text.

Author Response

We would like to thank the reviewers for thoughtful critiques of our manuscript. We have adopted the suggestions and think that the manuscript has been substantially improved by these revisions. We hope that you will now find it suitable for publication in Chemosensors. Our point-by-point responses to the comments are detailed below.

Summary: This study demonstrates the use of FTIR spectroscopy combined with chemometric analysis to differentiate palm-leaf manuscripts by species and geographical origin. The method is non-invasive, reproducible, and minimally affected by writing or conservation oils. Results show clear spectral differences between Borassus flabellifer and Corypha umbraculifera. Partial geographic separation is also possible. The results highlight the potential of FTIR fingerprinting as a valuable tool for manuscript authentication and codicological research.

Commentaries: In my opinion, this manuscript is well-structured and addresses a topic that will likely be of interest to the readers of Chemosensors. The research presents a valuable approach to the non-invasive analysis of historical manuscripts. However, the clarity and depth of the Methods and Results sections could be significantly improved. I believe the manuscript has strong potential for publication in Chemosensors, provided that the authors address several issues to enhance the reliability of the findings and the overall readability of the text.

Abstract: Overall, the abstract is adequate, but it would benefit from more specific and precise writing that highlights the technical depth and significance of the study. Key methodological elements, such as the application of Principal Component Analysis (PCA) ,are not mentioned, despite being central to the research. In addition, including quantitative details, such as the number of manuscripts analyzed, would provide a clearer context for the relevance of the obtained results.

PCA and the number of analysed manuscripts has been added to the abstract.

Introduction

- I would have appreciated a more detailed discussion of previous analyses of this type of manuscript, including what kinds of spectroscopic studies—either standalone or in combination with chemometric methods—have been conducted so far.

We added a discussion regarding FTIR applications to cultural heritage materials and applied chemometric methods. To the best of our knowledge, apart from the examples given, no further research has been conducted in this field with (palm-leaf) manuscripts.

- The rationale for selecting DRIFTS over other IR-based techniques should be more clearly described. Also, the terminology should be used consistently throughout the manuscript.

The selection of DRIFTS was further explained in the introduction:

“Unlike ATR, DRIFTS analyses a comparatively large sample area, making it particularly well-suited to fingerprinting. This is particularly useful when examining materials with an inhomogeneous macrostructure, such as palm leaf manuscripts.”

DRIFTS was explicitly used throughout the manuscript to be more consistent.

- Bringing these points together, the authors should clarify why existing methods are insufficient for the analysis of palm-leaf manuscripts and explicitly state what the present study contributes beyond the current state of the art.

The current state of the art for analysing palm-leaf manuscripts is actually not existing. Therefore, all available methods applied to valuable cultural heritage objects are a reasonable choices. With this study, we have laid the foundation for DRIFTS in combination with chemometric analysis to become a standard method for the analysis of palm leaf manuscripts. We have added this in the conclusion:

“The work presented here thus lays the foundation for a new approach in codicology to classify palm-leaf manuscripts according to specific material properties, which could also be applied in a similar way to manuscripts made of other materials.”

Methods

- I understand the difficulty in accessing this type of material, but do the authors not consider that a sample size of only 11 manuscripts is rather limited for proposing a new methodological approach? Additionally, the criteria for selecting these specific manuscripts should be clearly stated.

This is indeed a fair point. We added more details regarding the sample selection in the methods section, which was that the origin could be determined. Actually we are in progress to apply this approach to a larger collection (for which, this and other initial research is the basis). Permission to analyse such objects is not easy and requires trust and understanding of the methods for example from librarians that are not familiar with the scientific details. This is why this basic research is important to earn this understanding. We also write in the discussion and conclusion that this approach should be applied to more manuscripts:

“Future research is needed to validate this conclusion and confirm the results obtained here, also comparing or combining DRIFTS with other analytical methods for classification.”

“It is likely that supervised chemometric approaches can make this distinction even clearer, which should be tested in a next step with spectra from a larger number of manuscripts.”

- Regarding the “direct inspection” used to assign certain manuscript parameters, more detail is needed. Some form of validation for these assignments would strengthen the reliability of the study. It would also be helpful to provide more information about the conservation and storage conditions of the manuscripts, as these factors could potentially influence the spectral data.

We transparently presented all available data for this collection. The direct inspection by experts in biology and palm-leaf manuscripts is the best one can get. There will be future research which focus on DNA based species determination, which promise a scientific validation. We must fear that a collection of manuscripts with detailed and validated meta data is just not existent. Mock up samples could provide insides to the influences of deterioration, storage conditions and validated origin. Though, this was not in the scope of this study.

- The application of lemongrass oil is not entirely clear. Why was this oil chosen, and has it been commonly used in real conservation treatments? Has the potential interference of such substances with the spectral results been assessed?

We added more context to the application of oil:

“The application of oils is a typical repeated treatment for the preservation of palm leaf manuscripts.”

Lemongrass oil was (and is) reportedly used for conservation and the main result from this analysis was clearly written in the last sentence of the section:

“From this very simplified imitation of the conservation process, it can be concluded that treatment with oil does not significantly alter the spectrum of palm-leaf manuscripts which greatly simplifies the application of DRIFTS as a fingerprinting approach for these manuscripts in practice.”

- Finally, the chemometric analysis section should provide more detailed information regarding the preprocessing parameters and the statistical analysis performed.

We have listed all the parameter to use the functions from the respective R packages in the material section in 2.4.

Results

- The PCA models should include key performance indicators, such as the number of components used and the percentage of variance explained.

We looked at all resulting principal components from the PCA and reported the results for the first four components. We added the following sentence in the materials section:

“Although we assesed all resulting components from PCA analysis, components with less than five percent of variance were not reported (normally the fifth and higher).”

We also added the percentage of variance of the components to the supplement table S1.

- What do the first principal components in Figure 6 actually represent? The use of PC3 and PC4 for interpretation may give the impression of selectively reporting favorable results—this could be seen as a form of cherry-picking.

We added the following sentence: “Notably, the first two components, which account for the greatest variance, are also influenced by unknown characteristics of the manuscripts unrelated to their geographical origin.”

- Some explanation is needed for the partial overlaps observed in Figure 6A. Additionally, the clear separation of one group of Indonesian manuscripts with more positive PC3 values deserves attention. Similarly, the more negative PC1 values for Sri Lankan manuscripts in Figure 6B raise questions—could these patterns be related to the way the samples were assigned to origin groups?

We added the following sentences:

“However, there is some overlap between these groups, which could be attributed to similarity of individual manuscripts regarding other, unknown characteristics such as storage conditions, handling, and age of the manuscripts.”

“However, the spectra of Indian and Sri Lankan manuscripts overlap, suggesting that they may share similar characteristics in terms of the aforementioned unknown properties.”

- It is curious that in Figure 5, multiple clusters appear within the Borassus group, whereas this internal grouping disappears when the dataset is split by species. A possible explanation for this inconsistency should be discussed.

This is a good remark and we looked at this detail again. The grouping for the individual manuscripts is visible in the PCA scores. This is valid for the complete dataset as well as for the individual splitted datasets. The visual representation does not support a good comparison between those as we want to highlight the differentiation of the origin. Also different PCA components are used in this analysis, nevertheless they can not be compared directly as the dataset has changed and so does the PCA.

- The variations observed in Figure 7 could potentially reflect differences in conservation treatments or manuscript degradation—this possibility should be considered in the interpretation.

We added the degradation state as an additional source of variation: “The reason for these differences could be due to different ageing, but also different treatment or degradation states of manuscripts.”

Conclusions

- The conclusions should be completely rewritten, as they currently serve more as a brief summary than a reflective or critical synthesis. The authors should address key questions such as: What does this methodology contribute to the field? What are its main limitations? Are the results meaningful considering the relatively small dataset? Consider moving parts of the Discussion section—particularly those that interpret findings and implications—into the Conclusions, as they would be more appropriate and valuable there than the current text.

We revised the discussion and conclusion

Round 2

Reviewer 5 Report

Comments and Suggestions for Authors

Review of manuscript chemosensors-3610516 entitled “Differentiation of Species and Provenance of Palm-Leaf Manuscripts using Fourier-Transform Infrared Spectroscopy and Chemometrics”

I appreciate the authors' efforts to revise the manuscript and address my previous comments and suggestions.

In its current form, I consider the manuscript suitable for publication. However, I recommend that the authors carefully proofread the text to correct any remaining typographical or punctuation errors.

Author Response

We carefully revised the manuscript and corrected the remaining errors.