Process-Based Technical Evidence for a Rotationally Constructed Cubist Painting Associated with Pablo Picasso

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript presents an in-depth technical study of a Cubist painting of controversial historical provenance, signed “Picasso” and dated 1921. The multidisciplinary approach (X-rays, raking light, cross-modal analysis) aligns with current standards in Technical Art History.

Title: The introduction of concepts such as "Rotationally Constructed" requires a more rigorous explanation in the text to avoid subjective interpretative drift.

The name and physical dimensions of the canvas must be cited, and it should be clarified that this article is a continuation of previous work by the authors.

In the “Methods” section, the technical characteristics of the instrumentation used for each methodologies are not specified.

For X-Ray radiography: which are the acquisition parameters, such as kV, MAS or mA and seconds? What are the source and detector used? What are the images format and bit depth?

For UV and IR images: what sources are used? Are they monochromatic, and what are the wavelengths and the detector? Are any optical filters used, and what are their technical characteristics? What are the technical characteristics of the detector? What are the technical characteristics of the images?

Which software is used to process the images in the article?

In lines 95–96, the authors state, “Structural boundaries detected in X-ray (planning seams, armatures) were coarsely registered to IR and UV datasets…” This should be replaced with a quantitative description to ensure the rigour of the images overlap.

Greater consistency in technical language is required for X-ray images. In the text, the authors use "normal" and "inverted polarity", while in the supplementary material they use more appropriate terms such as "grayscale" and "inverted grayscale". However, they need to explain in the "grayscale" description what is meant by white and black, as this depends on the LUT used by the detector.

In line 104, the authors use “Normal-light images…” to describe visible images. More technical rigour is required.

In the supplementary material, "reduced colours" is defined as an RGB image without explaining what operation was performed or for what purpose.

In Figures 1, 2, 3, 7, 8, 9, 10, and 11, the authors include small details of the canvas that requires a metric reference and need to be contextualised on the pictorial surface.

In Fig. 4, images acquired using different methodologies are compared, but the images do not have the same alignment or magnification.

The manuscript cites "Musée national Picasso-Paris 2026" as an online reference accessed in February 2026. Given the current date, this appears to be a chronological error or a reference to a future publication that cannot yet be verified by reviewers.

There are many aspects of this article that require major revision. For example, the methodology lacks the transparency necessary for scientific reproducibility, particularly concerning the digital manipulation of data. Terms such as "reduced colour", "polarity inversion", and "coarse registration" are used without specifying the underlying algorithms, software versions, or mathematical transformations applied to the pixels. I suggest providing a detailed image-processing workflow, including the specific software used and a step-by-step description of how "Reduced Colours" or "Grayscale stress-tests" were generated, to ensure that these adjustments did not introduce any artefacts.

Author Response

Rev #1

The manuscript presents an in-depth technical study of a Cubist painting of controversial historical provenance, signed “Picasso” and dated 1921. The multidisciplinary approach (X-rays, raking light, cross-modal analysis) aligns with current standards in Technical Art History.

We thank the reviewer for the detailed assessment and constructive suggestions. We have revised the manuscript extensively to improve methodological transparency, terminology consistency, and technical rigor. Below we respond point-by-point.

1. Title: The introduction of concepts such as "Rotationally Constructed" requires a more rigorous explanation in the text to avoid subjective interpretative drift.

We have added a precise definition (in bold) in the “Methods’ section 2.2. X‑ray radiography and orientation‑mapping: “Radiographs were assessed at native resolution in both grayscale and inverted grayscale at canonical orientations (0°, 90°, 180°, 270°). Candidate features were advanced to interpretation only when they met four criteria: (1) continuity across adjacent crops; (2) tonal robustness under grayscale inversion; (3) hierarchical logic (primary planning to admit sensible secondary terminations) ; and (4) refindability after rotation and modest tonal changes. Here, “refindability” denotes the ability to relocate and recognize the same structural feature after rotation and tonal adjustments without loss of geometric identity. For each locus, the orientation with maximum geometric coherence was selected, and ±15° fine steps around the optimum were used to confirm stability. A structure is considered ‘rotationally constructed ‘if its geometry becomes maximally coherent only after a specific rotation and remains identifiable under polarity inversion and modest tonal variation. Features are advanced to interpretation only if they satisfy the above criteria: continuity, polarity robustness, hierarchical-seam behavior, and refindability after rotation. This protocol operationalizes orientation‑dependent coherence and mitigates pareidolia by requiring persistence across polarity/orientation stress‑tests (Favero et al., 2017; Delaney et al., 2010; Pouyet et al., 2020). (the workflow is summarized in Appendix A (Figure A1).”

2.`The name and physical dimensions of the canvas must be cited, and it should be clarified that this article is a continuation of previous work by the authors.

We now provide full  information in Introduction, line 44:  Red Guitar, is a 35.7 cm × 25.9 cm oil on cardboard, signed “Picasso,” and dated 9 April 1921. We also changed  the sentence on line 62: “Within this framework, the present study the present study extends the earlier analyses presented in Bakovic et al. (2022) and evaluates the painting through process-based lenses.”

3-5  Missing technical specifications for X‑ray/UV/IR  and image software

Detailed technical characteristics of the instrumentation and image software are described in a  new Methods section 2.1. Imaging acquisition and processing :

“X-ray radiographs were acquired using a portable digital X-ray system (Siemens Mobilett) with low‑kV, low‑mAs exposures (40–70 kV, 0.5–6 mAs), with a CsI/a‑Si flat‑panel detector (100–150 µm pixel pitch). After 200s exposure raw images were processed for display and storage using Digital Image Management System (DIMS) (Bakovic et al. 2022).

The UV imaging was performed using a Darkbeam DB‑2UV lamp emitting at approximately 365 nm. The IR light was produced by a Kodak No. 2 safelight ruby optical glass filter (Cat. 152 1525) transmitting at 600–650 nm, but also near IR (700–900 nm) and mid/far IR (>1000 nm). The only optical filter used was the Kodak No. 2 deep-red filter.

Images were acquired with an Android digital camera equipped with a back‑illuminated CMOS sensor and RGB Bayer filter array, which is only sensitive to visible light. RAW capture was not enabled, and all images were recorded as JPEG files generated by the device’s internal processing pipeline. On a Microsoft Windows PC, images were viewed and arranged in Microsoft PowerPoint using the minimal color display (reduced color images) and no‑color display (grayscale images) modes. These settings modified only on‑screen rendering and did not alter the underlying pixel data, and no destructive image processing was applied. The grayscale and racking light grayscale inversion was conducted using Pine Tools (https://pinetools.com/invert-image-colors). The X-ray orientation mapping, polarity inversion and cross modal registration were performed with Image J (NIH) using native‑resolution radiographs. All non‑X‑ray image transformations were restricted to display‑layer luminance visualization (LUT‑based rendering) and did not modify stored pixel values; interpretive conclusions do not depend on compression‑sensitive color metrics “

7            In lines 95–96, the authors state, “Structural boundaries detected in X-ray (planning seams, armatures) were coarsely registered to IR and UV datasets…” This should be replaced with a quantitative description to ensure the rigour of the images overlap.

We replaced qualitative wording with a fully quantitative description in Section  2.3. Cross-modal registration: “Structural boundaries detected in X‑ray (planning seams, armatures) were registered to IR and UV datasets and to high‑resolution visible images using stable landmarks. Landmark-based alignment used at least six anchors (contour breaks, seam junctions, guitar sound-hole geometry). Pixel-shift tolerance during manual refinement was kept within 1–2 px”. The examples are shown in Appendix A (Figure A2 ).

 

8-10      Greater consistency in technical language is required for X-ray images. In the text, the authors use "normal" and "inverted polarity", while in the supplementary material they use more appropriate terms such as "grayscale" and "inverted grayscale". explain in the "grayscale" description what is meant by white and black, as this depends on the LUT used by the detector. However, they need to In line 104, the authors use “Normal-light images…” to describe visible images. More technical rigor is required.

We updated the manuscript to consistently use: “grayscale” and “inverted grayscale” instead of “normal/inverted polarity”  and  “visible-light imaging” instead of “normal-light”

11          In the supplementary material, "reduced colors" is defined as an RGB image without explaining what operation was performed or for what purpose.

We now explicitly describe the transformation: luminance-only via linear transformation; Purpose: confirm relief-based rather than hue-based legibility; No color-selective filters applied.

12          In Figures 1, 2, 3, 7, 8, 9, 10, and 11, the authors include small details of the canvas that requires a metric reference and need to be contextualized on the pictorial surface.

This is now included.

13          In Fig. 4, images acquired using different methodologies are compared, but the images do not have the same alignment or magnification.

We re-acquired or re-registered the images and matched magnification across modalities. New overlays added.

14          The manuscript cites "Musée national Picasso-Paris 2026" as an online reference accessed in February 2026. Given the current date, this appears to be a chronological error or a reference to a future publication that cannot yet be verified by reviewers

We corrected this citation. The original URL displays digital catalogue material that was updated in 2025/26. The date now reflects the access date only.

15                I suggest providing a detailed image-processing workflow, including the specific software used and a step-by-step description of how "Reduced Colors" or "Grayscale stress-tests" were generated, to ensure that these adjustments did not introduce any artefacts

As we described in 3-5 above,  for visible images we used standard Microsoft PowerPoint  minimal color display  mode to produce ‘reduced color ‘images and no‑color display  mode to produce grayscale images. These settings modified only on‑screen rendering and did not alter the underlying pixel data, and no destructive image processing was applied. The grayscale and racking light grayscale inversion of Jpeg images  was conducted using Pine Tools (https://pinetools.com/invert-image-colors).

The X-ray orientation mapping, polarity inversion and cross modal registration were performed with Image J (NIH) using native‑resolution radiographs.

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Reviewer’ comments

I read the paper with great interest and would like to commend the authors for their innovative approach to the research topic. This paper addresses a subject that is worthy of publication. However, I have a few comments that may help enhance the work.

The article is well-structured, and the topic is particularly engaging as it sheds light on new research related to Picasso’s working process. A multi-step workflow was employed to evaluate whether painting exhibits a coherent structure both in depth (through X-ray analysis) and on the surface (considering visible and light characteristics). This innovative approach has provided new insights into the artist's working process. Below are my comments regarding minor revisions:

 

Comments

Figures 1, 2, and 3 are unclear, making their meaning difficult to understand. They should be replaced with higher resolution images.

The reviewer suggests that the authors should provide more details on whether their results illuminate any previously unknown aspects of the artist's painting process.

Given that Picasso's works are highly recognized and have been extensively studied, the reviewer also recommends that the authors enhance their bibliography with relevant literature. This would demonstrate that they are aware of existing research and clarify how their own work contributes to the field.

The reviewer believes that the article's conclusions should clearly outline the new insights this study offers regarding Picasso's painting process.

Author Response

Rev 2

Reviewer’ comments

I read the paper with great interest and would like to commend the authors for their innovative approach to the research topic. This paper addresses a subject that is worthy of publication. However, I have a few comments that may help enhance the work.

The article is well-structured, and the topic is particularly engaging as it sheds light on new research related to Picasso’s working process. A multi-step workflow was employed to evaluate whether painting exhibits a coherent structure both in depth (through X-ray analysis) and on the surface (considering visible and light characteristics). This innovative approach has provided new insights into the artist's working process. Below are my comments regarding minor revisions:

 We thank Reviewer #2 for the very positive evaluation and helpful suggestions.

Comments

Figures 1, 2, and 3 are unclear, making their meaning difficult to understand. They should be replaced with higher resolution images.

All three figures have been replaced with higher-resolution versions and improved annotations. They are sections of the  X ray images- the  high resolution of all images  in n=provided in Appendix B

1. The reviewer suggests that the authors should provide more details on whether their results illuminate any previously unknown aspects of the artist's painting process.

The Discussion new section 4.7. Previously unreported process markers and why they matter now explicitly describe our new process signature that extend existing literature:

“This study identifies four interacting process markers that, taken together, clarify aspects of Picasso’s construction workflow not previously formalized within a single, non‑destructive protocol.

(1) Orientation‑dependent armature: X‑ray structures become geometrically coherent only after controlled rotation and remain stable under inverted grayscale viewing; this operationalizes rotation as a diagnostic stress test, distinguishing intentional planning from incidental density variation (Fig. 5; Appendix A). This extends prior technical studies that documented underdrawings and hidden compositions by adding a formal rotation‑mapping criterion to discriminate planning structures, complementing work on La Miséreuse accroupie (Langley et al. 2013; Barten & Delaney 2018) and related Cubist/Blue Period (Casadio 2010; Gedo 1986) analyses (Favero et al, 2017; Pouyet et al.; 2020; Bakovic et al. 2022)

(2) Hierarchical seam architecture: a primary base‑layer seam governs later edits that terminate mechanically against it, and this geometry retains identity across rotations and modalities; the result is an explicit, reproducible map of planning hierarchy rather than an inferred one (Figs. 2–3; Appendix A). This codifies a seam‑hierarchy framework that aligns with Picasso’s iterative revisions on reused supports (Langley et al. 2013; Barten & Delaney 2018) while making the planning logic explicit through cross‑modal overlays, thereby systematizing observations scattered in earlier technical overviews (Arslanoglu et al., 2013).

(3) Stratigraphically embedded writing: incised names, place, and distributed year numerals (1912–1919) are earlier than overlying paint, verified by groove filling/bridging and persistence under grayscale and grayscale inversion, thereby constraining a multi‑year palimpsest culminating in the 1921 signature and reframing text as a constructional act (Figs. 7–9). This reframes textual elements from surface annotation to process‑integrated construction, consolidating earlier observations about Picasso’s use of letters and numerals within composition by grounding them in stratigraphy and relief tests(Delaney et al. 2010; Jiménez, 2021)

(4) Microscale relief motifs: repeated circular/semi‑circular relief forms and pigment reservoirs in tuner‑area quadrants evidence deliberate manipulation of semi‑wet paint, adding a micro‑tier of process control that extends beyond typical accounts of revision or support reuse (Fig. 11). This introduces a reproducible microscale relief tier, rarely isolated as a procedural marker in Picasso technical literature, thereby broadening the toolkit beyond macro‑level pentimenti and material IDs.  

Taken together, these four markers form a non‑destructive, reproducible process signature that integrates rotation awareness, seam hierarchy, stratigraphic text anchoring, and microscale relief into a single analytical framework, extending earlier technical insights on reused supports and concealed compositions with an auditable protocol.”

 

2. Given that Picasso's works are highly recognized and have been extensively studied, the reviewer also recommends that the authors enhance their bibliography with relevant literature. This would demonstrate that they are aware of existing research and clarify how their own work contributes to the field.

We have doubled the number of  references to contextualize the workflow within current Picasso technical art history.

3. The reviewer believes that the article's conclusions should clearly outline the new insights this study offers regarding Picasso's painting process.

The revised Conclusion now summarize the novel contribution of the study:

“Using only non‑destructive imaging, we reconstruct a unified, multi‑campaign workflow for the painting: a rotation‑sensitive base plan (X‑ray) organizes later edits; incised textual elements predate overpaint and anchor a 1912–1919 palimpsest that resolves with the 1921 signature; and recurrent microscale relief features record intentional surface shaping during semi‑wet stages. Cross‑modal co‑location shows that depth and surface obey the same geometric constraints, indicating that planning, inscription, and revision were not independent additions but components of one evolving construction system. These findings refine what is known about Picasso’s working methods by formalizing orientation mapping and seam hierarchy, demonstrating stratigraphically embedded writing as part of construction, and introducing microscale relief as a reproducible marker of process. The resulting technical footprint is mechanically coherent and difficult to fabricate, supporting an originality‑positive interpretation when considered alongside materials context and curatorial review.  These findings do not constitute stylistic attribution; rather, they demonstrate a mechanically coherent, historically plausible constructional logic within the documented technical repertoire of the relevant period”.

.”

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The revised manuscript shows some improvement, but significant concerns remain regarding methodological clarity, reproducibility, and technical support for the main interpretive claims.

The imaging methods are still not documented in sufficient detail, particularly with respect to X-ray acquisition parameters, the characterization of the “IR” modality, and the image-processing workflow. Cross-modal registration also remains insufficiently quantified. As a result, several conclusions appear stronger than the evidence presently documented.

Some formal and editorial problems also remain unresolved, including inconsistencies in references and supplementary material.

In its present form, the manuscript still requires substantial revision before it can be considered for publication.

Comments for author File: Comments.pdf

Author Response

We thank Reviewer 1 for careful reevaluation of the revised manuscript and for the detailed, technically informed comments. We appreciate the emphasis on methodological clarity, reproducibility, and precise distinction between visualization and analysis. Below we address each point raised and indicate the corresponding revisions made in the manuscript.

 

General:-Concerns remain regarding methodological clarity, reproducibility, and the distinction between visualization and analytical image processing.

 

We have revised the manuscript to clearly distinguish analytical imaging procedures from illustrative visualization steps. In particular, reduced‑color images are now explicitly defined as interpretive visualization aids intended to support visual inspection and comparative assessment, rather than as algorithmic, quantitative, or classification‑driven image‑processing outputs. The purpose, epistemic role, and limitations of these visualizations are now stated explicitly in the Methods section and relevant figure captions. Throughout the manuscript, we have clarified that analytical observations are grounded in native acquisition data, while visualization products are used solely to assist interpretation and communication.

 

1. Date inconsistency (Line 43)

Corrected access date in Introduction and References.

2. X‑ray tube parameters are incomplete. Exposure time, feasibility at stated kV/mA, and file format/bit depth are missing (Lines 79–82)

The Methods section has been revised to explicitly report the requested X-ray parameters. We  clarify that while images were processed for display using DIMS, all analytical observations were made at the native detector resolution. No analytical decisions were based on resampled, down‑sampled, or display‑optimized images.

3. The UV method is ambiguous; the term “approximately” should be avoided. The technique (UV‑induced visible fluorescence vs UV reflectography) and filters are unclear. a

This ambiguity has been corrected. The manuscript now explicitly specifies the technique as UV‑induced visible fluorescence, including excitation wavelength range, emission filtering, and imaging conditions. Approximate language has been removed, and the distinction from UV reflectography is now clearly stated in the Methods section.

4. Insert bibliographic citation and version of ImageJ (Line 93)

ImageJ, version 1.54p (NIH) is added in Section 2.1;  Reference: Schneider, C.A.; Rasband, W.S.; Eliceiri, K.W. NIH Image to ImageJ: 25 years of image analysis. Nature Methods 2012, 9, 671–675.

5. Definitions are subjective (“click into place”) and require more rigorous operationalization (Lines 98–109; Appendix A)

Appendix A and the Methods section have been extensively revised to replace informal phrasing with explicit, operational criteria. Subjective language such as “click into place” has been removed. The revised text clearly differentiates between visualization‑assisted interpretation and analytical observation, with criteria framed in terms of observable spatial correspondence, continuity, and material features.

6. Cross‑modal registration details (Lines 112–118) Transformation model, resampling, image resolution, and error metrics are missing.

This section has been fully rewritten. The revised text now specifies the transformation model used for cross‑modal alignment, the resampling approach, native and working image resolutions, and the qualitative basis used to assess registration accuracy. The limitations of visual registration are explicitly acknowledged, and no quantitative error metrics are implied where they were not measured.

7. Non‑technical terminology is used-“normal‑light images” (Lines 120–127)

Terminology has been standardized to “visible‑light imaging” throughout the manuscript.

8. Grayscale is not an independent modality. (Lines 130–133)

We agree. The text has been revised to clarify that grayscale and reduced‑color renderings are visualization controls derived from visible‑light imaging and are not independent acquisition modalities. References to multimodal corroboration are restricted exclusively to physically independent imaging techniques.

9. Definitive statements should be refrained from without quantitative tests. (Lines 145–159)

Interpretive language has been softened across the Results and Discussion sections. Definitive claims have been replaced with conditional phrasing, and interpretations are now explicitly framed as visually and materially informed, rather than statistically validated.

10. Methods for these products are not reproducibly defined Generation of heatmaps, contours, difference maps(Lines 174–191)

The Methods section has been expanded, and Supplementary Method Sections S2 and S3 have been added to Appendix A. These sections provide explicit procedural descriptions for the generation of heatmaps, contours, and difference maps. These products are now clearly identified as interpretive visual syntheses grounded in explicitly defined, physically observable image features, rather than as outcomes of automated algorithmic segmentation or classification.

11. Claims of reproducibility remain qualitative (Lines 254–309)

The manuscript has been revised to clarify that references to reproducibility in relation to tunner-area micro-relief motifs refer to recurrent morphology and consistent mechanical behavior observed across imaging modalities, rather than to statistical replication. Interpretive claims based on these observations have been correspondingly narrowed and carefully qualified.

 

We again thank the reviewer for their thorough and technically sophisticated critique. The revisions have significantly strengthened the manuscript’s clarity, reproducibility, and terminological precision, while preserving its central contribution: a process‑based, nondestructive framework for evaluating orientation‑dependent construction and stratigraphic integration in early 20th‑century painting practice.