Master thesis
An investigation into whitened paint areas on a painting by Adriaen
Cornelisz Beeldemaker, Dogs in a landscape
Sofie Dubbeldam
Student number: 11121300
Thesis supervisor: dr. Emilie Froment Second reader: Tamar Davidowitz
University of Amsterdam
June 21, 2019.
2 Dubbeldam, UvA, 2019 TABLE OF CONTENTS
Summary 4
Samenvatting 5
Preface and acknowledgements 6
1. Introduction 7
1.1 Introduction of the painting and research questions 7
1.2 Current scientific knowledge 9
1.2.1 Short discussion on terminology 9
1.2.2 Whitening induced by physical degradation 9 1.2.3 Migration and pigment-medium interactions 10
1.3 Structure of the thesis and methodology 11
2. Technical examination of Dogs in a landscape 13
2.1 Painting support 13
2.1.1 Original strainer and current stretcher 13
` 2.1.2 Canvas support 14 2.2 Size layer 16 2.3 Ground layer 17 2.4 Paint layer 23 2.4.1 Materials 23 2.4.2 Condition 27
3. Characterisation of the whitening phenomenon 32
3.1 Visual characteristics of the whitening and its
distribution across the paint surface 32
3.2 Microscopic study of the whitening 35
3.3 Comparison of cross-sections from paint affected by the whitening and paint not affected by the whitening 38
3 Dubbeldam, UvA, 2019 4. Underlying causes behind the formation of the whitening 43
4.1 Possible mechanism behind the formation of porosity 43
4.2 Provenance and conservation history 44
4.2.1 Provenance and conservation history 44 4.2.2 Condition prior to treatment by Marchand 45 4.2.3 Conditions of Marchand’s treatment that
may have contributed to the whitening 47 4.3 Influence of the canvas on the distribution of the whitening 49
5. Conclusion and propositions for further research 55
References 57
Overview Appendices 59
Appendix 61
Images made by Sofie Dubbeldam unless stated otherwise.
Photograph cover: Adriaen Cornelisz Beeldemaker. Detail of Dogs in a landscape. 1697. Oil on canvas. 44,7 x 49, 1 cm. Dordrecht, Dordrechts Museum.
4 Dubbeldam, UvA, 2019
SUMMARY
The focus of this master thesis is a painting by Adriaen Cornelisz Beeldemaker (1618 - 1709). Dogs in a landscape (1697) and its pendant, that shares the same name, are part of the collection of the Dordrechts Museum. During the recent restoration of the painting, the removal of old retouches in the sky uncovered a disturbing whitened paint layer. The whitening, which is not encountered on the painting’s pendant is confined only to the sky of the painting and has manifested itself in a pattern of vertical stripes. The aim of the research presented in this thesis is to identify the nature of the whitening and to investigate the underlying causes that are responsible for its formation.
The relation of the painting technique to the whitening is considered and involves mostly the study of paint cross-sections with the light microscope and with SEM-EDX. The starting point for the characterization of the whitening is a macroscopic and microscopic study of the paint surface. And the nature of the alteration is further investigated by comparing a cross-section from paint affected by the whitening to a cross-section from paint not affected by the whitening. Similarly the morphology of the surface from whitened and non-whitened paint is compared with SEM. After the characterization of the whitening, the potential influence of the painting’s conservation history on the formation of the whitening is considered.
The research suggests that the whitening is characterised by a physical alteration in the paint film. The whitened paint has areas with a porous structure, whereas in the unaffected paint the binding medium has remained smooth and closed. The formation of these pores in the paint layer can explain the whitening on Dogs in a landscape as an optical effect, namely the scattering of light. The increased porosity of the altered paint does not only distort the visual appearance of the paint layer but it also leads to a loss of cohesion in the paint. This is revealed by the presence of several paint losses in areas of the paint that have affected by the whitening. The mechanism behind the formation of the porosity is not yet fully understood, but the exposure to moisture and heat seem to play a big role. In Dogs in a landscape, the presence of hygroscopic pigments in the painting may have caused the paint layer to be extra sensitive to changes in moisture. The reactivity of the paint to changes in moisture may have played a role in the formation of the porosity in the paint. The exposure to both high levels of water in combination with a lot of heat during a past treatment may explain the paint layer’s current condition.
5 Dubbeldam, UvA, 2019
SAMENVATTING
Het schilderij dat centraal staat in deze master scriptie is Honden in een landschap (1697), geschilderd door Adriaen Cornelisz Beeldemaker (1618-1709). Dit schilderij en zijn
pendant, dat dezelfde naam deelt, zijn onderdeel van de collectie van het Dordrechts Museum. Tijdens de meest recente restauratie, werd bij het verwijderen van oude retouches in de luchtpartij van het schilderij een verstorende witte verkleuring van de verflaag blootgelegd. De witte verkleuring van de verf komt niet voor op het pendant en is alleen aanwezig in de luchtpartij. De wit geworden verf is verspreid over het oppervlak in parallelle verticale strepen. Het doel van het onderzoek dat gepresenteerd is in deze scriptie is de karakterisering van het fenomeen en het uitvinden van de onderliggende oorzaken betrokken bij de vorming van de verkleuring.
De relatie tussen de schildertechniek en verkleuring is onderzocht en betreft voornamelijk de analyse van verfdwarsdoorsneden, zowel met de lichtmicroscoop als met SEM-EDX. Het vertrekpunt van de karakterisering van de witte verkleuring is de bestudering van het verfoppervlak, zowel macroscopisch als microscopisch. De eigenschappen van de
verkleuring zijn verder onderzocht doormiddel van een vergelijking van
verfdwarsdoorsneden van de niet aangetaste verf en de wit geworden verf. Ook is voor de aangetaste en niet aangetaste verf de morfologie van het oppervlakte vergeleken met SEM. Het onderzoek suggereert dat het wit worden van de verf gekenmerkt is door een fysieke verandering in de verflaag. Het oppervlak van de wit geworden verf laat gebieden met een poreuze structuur zien, het bindmiddel in de onaangetaste verf is gesloten. De vorming van deze poriën kan de witte verkleuring op Honden in een landschap uitleggen als een optisch effect, namelijk als de verstrooiing van het licht. De porositeit van de aangetaste verf heeft niet alleen een visueel verstorend effect, maar het leidt ook tot een structurele verzwakking van de verflaag. Deze verzwakking laat zich zien in het ontstaan van
verfverlies in de wit geworden verf. Het mechanisme achter de vorming van de porositeit is nog niet volledig verhelderd. Het lijkt erop dat de aanwezigheid van vocht en hitte betrokken zijn bij de vorming. De aanwezigheid van hygroscopische pigmenten in Dogs in a landscape kan er voor gezorgd hebben dat de verf extra gevoelig is voor veranderingen in het vochtgehalte. De reactiviteit van de verflaag op de veranderingen in het
vochtgehalte kunnen een rol gespeeld hebben in de vorming van de porositeit in de verf. De blootstelling aan veel water en hitte tijdens een eerdere restauratie verklaart mogelijk de huidige conditie van de verflaag.
6 Dubbeldam, UvA, 2019
PREFACE AND ACKNOWLEDGEMENTS
This Master’s thesis has been written as part of the master Conservation and Restoration of Cultural Heritage, paintings specialisation at the University of Amsterdam (UvA). The painting researched in this thesis is part of the collection of the Dordrecths Museum.
I would like to thank a number of people, without their help this thesis could not have come about. I would like to thank first Dr. Emilie Froment (UvA), my thesis supervisor, for her guidance during the research. I further would like to thank prof. dr. Maarten van Bommel (UvA),prof. dr. Ella Hendriks (UvA) and dr. Rene Peschar (UvA) for providing feedback and advice regarding my proposed research. Additionally I would like to thank prof. dr. Klaas-Jan van den Berg (UvA & Cultural Heritage Agency of the Netherlands (RCE)) for discussing my research with me and for offering advice on the use of the right analytical techniques.
I am grateful towards dr. Maartje-Stols Witlox (UvA) for carrying out SEM-EDX analyses. Likewise I would like to thank dr. Ineke Joosten (RCE) for the study of the paint surface with SEM and for carrying out some additional SEM-EDX analyses. I would also like to thank her for reading and commenting on the texts I had written regarding the analyses. I am very thankful towards Dr. Don H. Johnson (William Marsch Rice University) for carrying out the automated thread counting analysis.
The research described in this thesis has built further upon the technical examination of the painting of interest and its pendant that was carried out earlier by my fellow students Ivana Jerdonekova and Jolijn Schilder. I want to thank them for sharing their findings with me and for letting me use their photographs.
Lastly, I would like to thank my fellow students, Steinnun Hardardottir, Christa Molenaar, Ivana Jerdonekova and Jolijn Schilder for their company and support during this project.
7 Dubbeldam, UvA, 2019
1. INTRODUCTION
1.1 Introduction of the painting and research questions
The research described in this thesis is focused on a painting by the hand of Adriaen Cornelisz Beeldemaker (1618-1709). The painting, named Honden in een landschap (1697), or translated Dogs in a landscape, portrays a scene with three dogs surrounded by a forest landscape (App. I.a). The painting which is part of the collection of the Dordrechts Museum is paired with another painting by Beeldemaker that shares the same name, the same date and has a strikingly similar composition (app. II.a). Hunting scenes, such as those depicted in the paintings in the Dordrechts Museum were Beeldemaker’s specialty and it appears that he sold a great many of these works, in part because of their affordability.1 It is not certain whether the two paintings in the Dordrechts Museum were originally intended as pendants or whether they were painted as part of a series of similar paintings that were intended to be sold separately. Within the Dordrechts Museum however, the paintings now function as a pair. For this reason, and for convenience, the paintings will be referred to as pendants in this thesis.
From the fall 2017 till the summer 2019, the pendants have been examined and restored by students from the Conservation and Restoration of Cultural Heritage programme at the University of Amsterdam.2 One of the paintings, the painting that is central to this thesis, had been extensively retouched in the past in the sky and background of the painting (mapping: App. III.e). The reason for this intervention was elucidated when the removal of the old yellowed varnish and old retouches revealed the presence of a whitened paint surface. No signs of whitening of the paint could be observed in areas other than the sky and the background, nor was any whitening detected on the pendant painting. Interestingly, the whitening has distributed across the surface in a peculiar pattern of parallel vertical stripes (fig. 1.1 – 1.2).
1 J. van Gool, De nieuwe Schouburg der Nederlantsche Kunstschilders en Schilderessen (’s-Gravenhage: author, 1751), 63-64.
2 The examination and restoration on the painting researched in this thesis (inv. no: DM/970/465) was carried out by Ivana Jerdonekova. The pendant (inv. no: DM/970/466) was examined and treated by Jolijn Schilder.
8 Dubbeldam, UvA, 2019 Because the whitening has occurred only in the sky and background of the painting, the phenomenon may be related to the pigments used in this area. It is known that both paintings have been wax-resin lined in 1971 by Léo Marchand. Prior investigations by Schilder and Jerdonekova during treatment of the paintings already pointed out that this treatment involved the exposure of the reverse of the painting to lots of water in order to remove the painting’s original glue size layer to facilitate the impregnation of the wax-resin lining mixture.
The aim of the research conducted in this thesis has been to provide a diagnosis of the whitening present on the paint surface of Dogs in a landscape. The following main research question was formulated:
What is the cause of the whitening on the paint surface in the sky and background of Dogs in a landscape (1697) by Adriaen Cornelisz Beeldemaker (1618-1709)? It is hoped that the identification of the cause of whitening on Dogs in a landscape will contribute to a greater understanding of whitening phenomena. The whitening of paint and varnish layers is commonly encountered but its causes can be versatile and are not always properly understood. Understanding the nature of the whitening, but more importantly understanding the conditions that promote the formation of the whitening is extremely important. In that way, paintings that are susceptible to this phenomenon can be recognized and measures can be taken to prevent the formation or aggravation of the whitening.
Fig. 1.1: Adriaen Cornelisz Beeldemaker.
Dogs in a landscape. 1697. Oil on canvas.
44,7 x 49, 1 cm. Dordrecht, Dordrechts Museum. Photograph: Ivana
Jerdonekova.
Fig. 1.2: Detail whitened sky. Adriaen Cornelisz Beeldemaker. Dogs in a landscape. 1697. Oil on canvas. 44,7 x 49, 1 cm. Dordrecht, Dordrechts Museum.
9 Dubbeldam, UvA, 2019
1.2 Current scientific knowledge.
1.2.1 Short discussion on terminology
Prior to conducting the technical examination and analyses that have been presented in this thesis, a study of the literature regarding the potential causes of whitening of paint surfaces has been carried out. In doing so an understanding was gained of the potential mechanisms of whitening that may be encountered on Dogs in a landscape.
The study of the literature describing whitened paint surfaces can be a confusing enterprise. A variety of terms have been used to describe the whitening of paint surfaces and not all of them are clearly defined. The fact that the terminology is not always used consistently throughout the literature adds to the confusion and requires that the terms are carefully interpreted in the context of each publication. For instance, two terms that are often encountered are ‘blanching’ and ‘blooming’. Both terms describe the whitening of the paint surface but refer, if used properly, to different mechanisms behind the phenomenon. Blanching has for instance been used mostly as a generic term to describe any whitening caused by a physical defect in the paint layer or paint surface.3 However, some authors seem to use the term as a synonym to ‘whitening’ but do not necessarily intend to imply a physical degradation of the paint film.4 The other example ‘blooming’ implies that the whitening on the surface has been caused by migration and precipitation on the paint surface of extractable components from the paint. 5
In lack of a good understanding of the phenomenon and to prevent false implications about the mechanism behind the phenomenon encountered on Dogs in a landscape, it was decided to use the neutral term ‘whitening’ in this thesis.
The whitening of paint or varnish layers is a phenomenon commonly encountered by conservators. Several causes have been proposed and identified. These causes could be roughly divided in two categories and will be discussed below.
1.2.2 Whitening induced by physical degradation
In some cases the whitening, or blanching, is essentially an optical phenomenon that is induced by a physical degradation of the paint film. The presence of voids or
3 K. Groen, “Scanning electron-microscopy as an aid in the study of blanching,” Hamilton Kerr
Institute Bulletin 1 (1988): 48; A. Van Loon, “Colour changes and chemical reactivity in 17th century oil paintings,” (PhD diss., University of Amsterdam, 2008), 122.
4 J. Koller and A. Burmester, “Blanching of Unvarnished Modern Paintings: A Case Study on a Painting by Serge Poliokoff,” in Preprints of the Contributions to the Brussels Congress, 3-7
September 1990: Cleaning, Retouching and Coating, ed. J. Mills and P. Smith (London:
International Institute for Conservation of Historic and Artistic Works, 1990), 138-143; M. Wyld, J. Mills and J. Plesters, “Some Observations on Blanching (with Special Reference to the Paintings of Claude),” National Gallery Technical Bulletin 4 (1980): 48-63.
10 Dubbeldam, UvA, 2019 cracks in the paint layer could lead to a scattering of the light and as a consequence lead to an opacification of the paint surface.6 The formation of voids in the paint layer may be caused by the swelling and shrinking of the paint layer during exposure to solvents or water and lead to a tearing of the binding medium from the pigments. 7
Relatively recently, such formation of micro-voids or pores was indeed demonstrated in whitened paint and varnishes.8 On a set of micro-samples from ‘blanched’ oil paintings it was shown that paints affected by the alteration had a highly porous structure compared to the paints not affected by the whitening. The same phenomenon could also be reproduced on mock-up samples that were prepared according to historical recipes, by exposing the samples to heat and high humidity. It was also shown however that the pigment composition must be involved in the formation of the whitening because samples prepared with green earth and raw umber changed, those with azurite and ivory black did not. The presence of chalk as an extender and a higher concentration of drier (litharge) also facilitated the formation of the whitening. The demonstrated pore formation was thought to be a mostly physical mechanism as no noticeable molecular modifications were detected. 1.2.3 Migration and pigment-medium interactions.
The second category of causes of whitening of paint surface regards migration phenomena and chemical changes in the binding medium due to pigment-medium interactions.
Whitened paint surfaces have often been related to the migration of fatty acids. Free fatty acids can move through the oil network and migrate the towards the paint surface. The unsaturated fatty acids react with oxygen and form oxidation products that evaporate, the saturated fatty acids remain at the surface and crystallize, giving the paint surface a white appearance.9 The source of free fatty acids could come from the paint medium itself, they can be released from triglycerides in the oil medium through hydrolysis.10 But they could also derive from materials applied in conservation treatments such as waxes for consolidation or relining.11
In other cases the whitening may be related to the migration of metal soaps. Metal soaps, or metal carboxylates are a product of the reaction between free fatty acids and metal
6 M. Wyld, “Blanching of the paint film involving possible changes in the medium,” National
Gallery Technical Bulletin 4 (1980): 60.
7 Wyld, “Blanching of paint involving changes in medium,” 60; B. Epley, “Jan Both’s Italian landscape: materials, technique and treatment,” Hamilton Kerr Institute Bulletin 3 (2000): 131. 8 A. Genty-Vincent et al., “Blanching of Paint and Varnish Layers in Easel Paintings: Contribution to the Understanding of the Alteration," Applied Physics A 121, no. 3 (July 2015): 779-788, 9 Koller and Burmester, “Blanching of Unvarnished Modern Paintings,” 138-143.
10 J. Mills and R. White, The Organic Chemistry of Museum Objects, 2nd edition (London: Butterworth-Heinemann, 1994), 34.
11 Koller and Burmester, “Blanching of Unvarnished Modern Paintings,” 141; E. Ordonez and J. Twilley, “Clarifying the Haze: Efflorescence on Works of Art,” Analytical Chemistry 69, no. 13 (July 1997): 418.
11 Dubbeldam, UvA, 2019 ions present in pigments or driers.12 Lead soaps are perhaps most commonly encountered because of the abundant use of lead white in both ground and paint layers in traditional oil paintings. Metal soaps can manifest themselves as two distinct degradation phenomena. In some cases, they have the tendency to aggregate, these aggregations grow and can eventually protrude through the paint surface. 13 In other cases, lead soaps may also migrate through paint layers to the surface without aggregation.14 At the surface, the lead soaps can then precipitate, crystallize and mineralize further with gasses in the atmosphere to form salts such as lead sulphates, chlorides and possibly carbonates. In such a case an insoluble crust may develop on the paint surface.15 The formation of white crusts on the paint surface through this mechanism has been identified and proposed as a cause for the whitening encountered on several paintings. 16 Through a similar mechanism the whitening of paint films containing smalt has been reported. Potassium derived from the smalt pigment can react with the oil medium to form potassium soaps and lead to the formation of potassium-soap containing crusts. 17
1.3 Structure of the thesis and methodology.
In order to present an answer to the main research question of this thesis the investigation has been sub-divided in three more sub-questions.
The whitening on the paint surface may only be properly understood when there is a good understanding of the materials and techniques used in the painting, therefore the first sub-question is: ‘What materials and techniques have been used in Dogs in a landscape?’ The second sub-question deals with the study of the nature of the whitening and has been formulated as ‘How can the whitening be characterized?’. An elaborate description of the phenomenon will be given and the question is intended to reveal whether the whitening can for instance be identified as an optical effect, a chemical degradation or as caused by
12 A. Sawicka et al., “Metal soap efflorescence in contemporary oil paintings,” in Issues in
Contemporary Oil Paint, ed. K.J. van den Berg (Basel: Springer International Publisher AG, 2014),
313.
13 Sawicka et al., “Metal soap efflorescence,” 313; Van Loon, “Colour changes in oil paintings,“ 11. 14 Sawicka et al., “Metal soap efflorescence,” 313. Van Loon, “Colour changes in oil paintings,” 188-192.
15 Van Loon, “Colour changes in oil paintings,” 191.
16 A. van Loon and J. Boon, “The whitening of oil paint films containing bone black,” in ICOM
Comittee for Conservation 14th triennial meeting, The Hague, 12-16 September 2005: preprints, ed. I. Sourbès-Verger (London: James and James, 2005), 511-518; J. Verhave, A. van Loon and P. Noble, “Changes in appearance: unintentional lightening in stable interiors by Philips Wouerman (1619-1668),” ArtMatter: Netherlands technical studies in art 4 (2007): 103-109.
17 M. Spring, C. Higgit and D. Saunders, “Investigation of Pigment-Medium Interaction Processes in Oil Paint containing Degraded Smalt,” National Gallery Technical Bulletin 26, (2005): 66; A. van Loon, P. Noble and J. Boon, “White Hazes and surface crusts in Rembrandt’s Homer and related paintings,” in ICOM-CC: 16th triennial conference, Lisbon, 19-23 September 2011: preprints, ed. J. Bridgland and C. Antomarchi (Almada: Critério, 2011): 1-10.
12 Dubbeldam, UvA, 2019 residues of a material left on the surface during a past restoration. After having characterized the whitening, the last sub-question can be considered: ‘What external influences have been responsible for the formation of the whitening?’. This question investigates what type of conditions must have been necessary for the formation of the whitening and considers if and how Leo Marchand’s treatment may have been involved. The first sub-question is answered in chapter 2. It discusses the build-up of the painting from the support to the paint layer. The technical examination was conducted by means of visual examination, X-radiography and automated thread counting. Paint cross-sections were prepared to study the paint and ground layer composition. The cross-sections were studied in normal light, ultraviolet (UV) light and with scanning electron microscopy and energy dispersive x-ray spectroscopy (SEM-EDX). Although the chapter does not intend to give a comprehensive description of the painting’s condition, some aspects of the condition of the paint layer other than the whitening is discussed because they may prove important for understanding of the whitening. The technical study has focused mostly on the painting affected by the whitening, because within this painting affected and unaffected paint exist next to each other, it was not considered absolutely necessary to prepare additional cross-sections of the pendant painting other than the ones that had already been prepared during the 2018-2019 restoration.
Chapter 3 focuses on the whitening phenomenon. An initial exploration of the whitening phenomenon was carried out both macroscopically and microscopically in normal light and ultraviolet light. The microscopic study was conducted both with a stereomicroscope and with a digital microscope (Dino-Lite and Hirox). Cross-sections prepared from a sample of paint affected by the whitening and a sample of paint not affected by the whitening were compared to see if residues of an external material or if a crust as a result of a chemical degradation of the paint could be detected on the surface. To assess the potential physical degradation as a cause for the whitening of the paint layer, the morphology of paint samples taken from both the affected paint and unaffected paint were studied with scanning electron microscopy (SEM).
Chapter 4 discusses the underlying causes for the formation of the whitening. This chapter encompasses the third sub-question ‘What external influences have been responsible for the formation of the whitening?’ but it is combined with a discussion on the influence of the paint composition and the influence of the canvas on the distribution of the whitening. This was done to ensure a more integrated discussion of each of these factors, which can be difficult to discuss separately. The potential influence of Léo Marchand’s treatment on the whitening was considered by the study of his treatment report. Additional information about his working methods was obtained from the testimonies about his working methods.
13 Dubbeldam, UvA, 2019
2. TECHNICAL EXAMINATION OF DOGS IN A LANDSCAPE 2.1 Painting support
2.1.1 Original strainer and current stretcher
Both paintings were painted on a canvas support. During the wax-resin lining treatment by Léo Marchand the painting was placed on a new stretcher because the original strainer was in poor condition.18 An old photograph of the reverse of the pendant painting, which was taken prior to Marchand’s treatment reveals the structure of the original strainer (fig. 2.1.).
From the photograph it can be deducted that the original strainer consisted of four battens connected either by a half lap joint or a butt joint, the structure was reinforced by four diagonal slats in each corner of the strainer.
The lining canvas was attached on its new stretcher with two rows of nails. The painting’s original tacking edges were preserved by Marchand and the canvas was nailed to the sides of the stretcher through the original tacking edges. The remaining material of the lining canvas was folded over and nailed to the reverse of the stretcher (fig. 2.2). The sides of the painting have been covered by a thick layer of wax, but original tacking margins and the holes caused by the original nails are still visible (fig. 2.3).
18 L. Marchand, Treatment report. see App. IV.d – IV.e. Fig.2.1: Reverse of pendant painting (DM/970/466) prior to restoration by Leo Marchand, 1971. It shows the painting on its original strainer (see also App. IV.a).
14 Dubbeldam, UvA, 2019
2.1.2 Canvas support
The study of the original canvas on the reverse of the painting is obstructed by the presence of the lining canvas. However, the X-radiographs of the paintings still allow the study of the characteristics of the original canvas (App. I.g, App. II.c). A plain canvas weave can be detected. The canvas is not of a particularly high quality, there is a great variation in the thread thickness and the canvas contains a lot of knots in its structure, indicating that the threads used for the canvas were not properly spun (fig. 2.4). Beeldemaker may have compromised on the quality of his canvasses in return for affordability. The type of fibres used for the making of the canvasses has not been analysed, most canvasses were made of linen although hemp fibres were also used in the Dutch sailcloth industry.19
19 J. Kirby, “The Painter’s Trade in the Seventeenth Century: Theory and Practice.” National
Gallery Technical Bulletin 20 (1999): 22.
Fig. 2.3: Detail of the side Dogs in a
landscape (DM/970/465). The original
tacking margins have been preserved. Photograph: Ivana Jerdonekova.
Fig. 2.2: Reverse of Dogs in a landscape (DM/970/465). Photograph: Ivana Jerdonekova.
Fig. 2.4: Detail X-radiograph. DM/970/465. Dogs in a landscape . The canvas weave is irregular. The thread thickness varies significantly, several knots are present (arrow). X-radiograph: Ivana Jerdonekova.
15 Dubbeldam, UvA, 2019 Both paintings were analysed by dr. Don H. Johnson with software presented by the Thread Count Automation Project (App. VIII). The software is able to calculate thread densities across the surface and a map is created in which the areas with a high thread density are coloured red and areas with a low thread density are coloured blue. In this manner a distinctive weave map can be visualized for the canvas. The pattern of these thread density maps can be compared to find matches of canvasses that have been cut from the same piece of cloth. In the case of the Dogs in a landscape paintings, no exact match could be detected between the two paintings, however the canvasses do show very similar thread counts in both the horizontal and vertical direction (table. 2.1).
DM/970/465 DM/970/466 (pendant) Average Standard deviation Average Standard deviation Horizontal thread density (th/cm) 13,0 0,9 13,8 1,2 Vertical thread density (th/cm) 13,9 0,8 13,9 0,8
Although it cannot be proven, it is still possible that the canvasses were derived from the same canvas roll. If the pieces of canvas were not cut in alignment but instead in a diagonal orientation relative to each other, then they would not share any threads in either the weave or weft direction. Even though in such a case the canvasses do derive from the same piece of cloth, the software is not able to then detect these canvasses as a weave match. 20
The average thread counts of the canvasses (table 2.1) do fit in a trend of using coarser canvases towards the end of the seventeenth century. Whereas at the beginning of the seventeenth century, canvasses generally had a thread count of 15-20 threads per cm, from the 1680’s thread counts of 10-16 per cm are more common.21
When the painting is placed in raking light, it is apparent that the surface of the painting is not even. The surface shows wave like deformations, parts of the canvas are slightly raised along the vertical direction (fig. 2.5).
20 D.H. Johnson, personal communication, May 27, 2019.
21 C. Young, “History of Fabric Supports,” in Conservation of Easel Paintings, ed. J.H. Stoner & R. Rushfield (London: Routledge, 2012), 133.
Table 2.1: Average thread counts in the horizontal (weft) and vertical (warp) direction. Analysis: dr. Don H. Johnson (see App. VIII for report).
16 Dubbeldam, UvA, 2019 The effect is subtle, but this uneven surface can also be detected when the painting is studied and viewed in normal light. The relining canvas on the reverse of the painting however, does not show any such deformations, indicating that these irregularities must be inherent to the original canvas. The pendant painting shows a similar surface in raking light (App. II.b). For both paintings, the unevenness cannot be detected in the horizontal direction.
2.2. Size layer
During the study of Dogs in a landscape, no convincing evidence could be found of the presence of a size layer. None of the cross-sections that have been prepared from the painting have revealed the presence of a size layer. In some paint losses, the threads of the canvas are exposed, however in the past the canvas has been impregnated with wax-resin and recently a new varnish has been applied to the paint surface. The microscopic study of these exposed canvas threads is not a reliable method for determining the presence of a size layer, because visually distinguishing a potential size layer from the varnish and wax resin is very challenging if not impossible. The application of a size layer was common practice, it functioned as a means to lay down small fibres and to reduce the absorbency of the canvas. It was believed that in doing so, the amount of ground layer material necessary for the preparation of the support could be lessened.22 Although no physical evidence of a size layer could be identified on Dogs in a landscape it is very likely that it was indeed prepared in this manner. The size could have been one based on animal glues, these are most frequently mentioned in historical recipes from the 16th and 17th century, but flour pastes were also used as a preparation.23
22 M. Stols-Witlox, A Perfect Ground. Preparatory Layers for Oil Paintings 1550-1900 (London: Archetype Publications Ltd, 2017), 72-73.
23 Stols-Witlox, A Perfect Ground, 50.
Fig. 2.5: Raking light from right. DM/970/465. Adriaen Cornelisz Beeldemaker. Dogs in a
landscape. 1697. Oil on canvas. 44,7 x 49, 1 cm.
Dordrecht, Dordrechts Museum. Photograph: Ivana Jerdonekova.
17 Dubbeldam, UvA, 2019
2.3 Ground layer.
The colour of the ground layer on Dogs in a landscape was revealed by the presence of several paint losses on the surface of the painting. These losses show that Beeldemaker prepared his painting with a light brown ground (fig. 2.6).
By studying the paint cross-sections it can be determined that the preparation of the canvas consists of a single ground only (fig. 2.7). The colour of the ground appears much warmer and lighter than it does on those areas where the ground layer has been exposed on the surface. This difference may be explained as a consequence of the strong illumination to which the paint samples are exposed when they are studied with the light microscope. The cross-section of the pendant painting that had already been prepared during the restoration, does unfortunately not clearly reveal the ground layer (App. V.f). No additional samples were taken from the pendant and the paint surface did not present any large losses that clearly revealed the colour of the ground layer. Along the abraded edges of some of the aging cracks in the paint layer the ground layer has been exposed (fig. 2.8).
Fig. 2.6: Dino-Lite photograph. 50x. DM/970/465. A loss in the paint layer has exposed the brown ground layer.
Fig. 2.7: Cross-section DM/970/645#4B. Bright Field (BF). Microscope: Leica DM2500. The ground layer consists of a light coloured matrix, it contains large yellow translucent particles and few red and black pigment particles.
Fig. 2.8: Dino-Lite photograph. 250 x. DM/970/466 (pendant). Abrasions of the paint along the aging cracks reveals a brown ground with a similar colour to the ground layer on its pendant.
18 Dubbeldam, UvA, 2019 The colour looks similar to that of the ground layer on the pendant. Considering the large similarities between the two pendants it may be assumed that the composition of the ground layer of the two paintings is similar, but it cannot be properly proven.
The cross-sections reveal that the ground contains large translucent yellow particles which are present in a light matrix, mixed with a few red and black pigment particles (fig. 2.7). EDX measurements have revealed that the ground layer is predominantly rich in lead, but also contains silicon and smaller amounts of various other elements; Al, Ca, Fe, K, Na and Cl (App. VI.c see pt. 4).
The high signal that is measured for lead, indicates that lead white is a main component in the ground layer composition. The lead white particles are not easily distinguished when the cross-sections are studied with visible light. However, the distribution of the lead white particles is easily recognized in the backscattered electron image (BSE), because of its high atomic mass, the lead particle register brightly in the BSE image (fig. 2.9). It appears that not much care was taken to grind the lead white to a fine gradation. There is a great variation in the particle size, the smallest particles are approximately 1 µm, but lead white particles or aggregates in the range of tenths of microns were also found.
The elemental distribution maps as measured by SEM-EDX have revealed the nature of the large translucent yellow particles in the ground. The particles are clearly defined by a high concentration of silicon and oxygen, this suggests that they consist of silicon dioxide (SiO2) (fig. 2.10). In pure form, SiO2 naturally occurs as quartz or otherwise referred to as ‘sand’.24
24 N. Eastaugh et al., Pigment Compendium (Oxford: Butterworth-Heinemann, 2008), 321. Fig. 2.9: cross-section: DM/970/465#6A. Left: BF, scale bar: 50 µm. Right: Backscattered electron image (BSE). Dotted line indicates the border between the paint and ground layer. The lead white particles register as white in the BSE, there is a great variation in the lead white particle size. 20kV. SDD detector. NovaNanoSEM450 (Thermo Scientific). Pathfinder Software. BSE image: Ineke Joosten (RCE).
19 Dubbeldam, UvA, 2019 Not all of the particles appear to contain SiO2 in pure form however. The SEM-EDX elemental distribution maps further reveal that several of the SiO2 containing particles are also rich in aluminum and/or sodium (fig. 2.10) (App. VI.b, VI.d). This indicates that there are also other silicate minerals besides quartz present.
Fig. 2.10: Cross-section DM/970/465#6A. Top left: BSE, top middle: BF 500x, top right and bottom: selection of SEM-EDX elemental distribution maps (see App. VI.d). See also App. VI.b for similar results in cross-section DM/970/465#4B.
The large translucent yellow particles, appear dark in the BSE image. They are characterised by a high concentration of silicon and oxygen, some also contain aluminium and sodium. SDD detector. JEOL JSM-5910LV. Noran System 7 Software. Analysis: Maartje Stols-Witlox (UvA).
20 Dubbeldam, UvA, 2019 When EDX measurements for sample #6A are visualized as phases, the presence of two different types of SiO2 based materials becomes even more clearly apparent (fig. 2.11).25
Phase two in figure 2.11 contains the particles that consist of silicon and oxygen only and can thus be identified as quartz (SiO2). The particles in phase 3 are in addition to silicon and oxygen, rich in sodium and aluminum. In nature, quartz is often found together with other minerals such as those from the feldspar group.26 Feldspar is a group name for various aluminosilicate minerals with the general formula (K, Na, Ca) [Al,Si)4O8], these minerals are the most commonly occurring minerals worldwide.27 The SiO2 phase rich in aluminum and sodium in sample #6A could potentially be a plagioclase feldspar. The plagioclase feldspars are a series of feldspars in which the Na and Ca content varies between albite
25 A ‘phase’ represents those areas that have a similar elemental composition. 26 Eastaugh et al., Pigment Compendium, 321.
27 Ibid., 161.
Fig. 2.11: cross-section DM/970/465#6A. top: BSE. Bottom: silicon containing phases in the ground layer with corresponding EDX spectra (see also App. VI.g). Phase 2 consists of quartz. The particles in phase 3 are possibly plagioclase feldspars. SDD detector. NovaNanoSEM450 (Thermo Scientific). Pathfinder Software. Analysis: Ineke Joosten (RCE)
21 Dubbeldam, UvA, 2019 (NaAlSi3O8) and anorthite (CaAl2Si2O8).28 The very low calcium content indicates that these particular particles in phase 3 of sample #6A are very close in composition to albite. It is interesting to note that if this one cross-section, which represents only a tiny fraction of the ground layer, already contains two types of silicate minerals it is not unreasonable to assume that the ground layer on Dogs in a landscape contains an even greater variety of silicate minerals.
Considering the presence of quartz and likely feldspar in the ground layer, minerals which are derived from the earth’s crust it is interesting to consider in what form Beeldemaker applied these materials in his paintings. Some possibilities present themselves, the minerals could have ended up in the painting’s preparation layer through the use of clays, or as a contaminant in earth pigments.
Clays are often found together with crystals of quartz, feldspar and mica.29 Clays consist of hydrated aluminum sheet silicate minerals and they often contain additional elements such as Mg, Fe, Ca, Na or K.30 Although the signal from lead white in the ground layer is most prominent, silicon, aluminum and most of these additional elements have been detected in the ground layer of Dogs in a landscape (fig. 2.12).
28 Eastaugh et al., Pigment Compendium, 161. 29 Ibid., 113.
30 Ibid.
Fig. 2.12: Cross-section DM/970/465#5A. Top: BF. Bottom: BSE with EDX-spectra for pt.1 and pt.4 (see also App. VI.c). Pt.4: The ground layer contains a combination of elements similar as found in clays. Pt.1: the presence of Mn and Fe may indicate the use of umber or sienna. SDD detector. JEOL JSM-5910LV. Noran System 7 Software. Analysis: Maartje Stols-Witlox (UvA).
22 Dubbeldam, UvA, 2019 The earth pigments are natural iron oxide and iron oxide hydroxides. As natural deposit, they are often found in association with clay minerals and also quartz.31 The yellow, red and brown earth pigments exist in three main classes; the ochres, siennas and umbers. According to the classification given by Eastaugh et al., the sienna’s contain minor (<5%) amounts of manganese oxides in addition to the iron oxides and umbers are distinguished by their higher content of manganese oxides (5-20%).32 Sample #5A shows a section of the ground where there is an apparent concentration of not properly mixed brown pigments (fig. 2.12). The EDX measurements for this area reveals that this area of the ground is rich in iron and contains a small amount of manganese (fig. 2.12). This likely points to the use of either sienna or umber.
Seventeenth and eighteenth century art technological sources contain multiple recipes that prescribe the use of an oil bound lead white ground tinted with earth pigments.33 Additionally, recipes occur that advise the use of clays for single canvas grounds. Both Lebrun, in 1635, and Dutch author Simon Eikelenberg in his notes on the art of painting (1679-1704) described the use of oil grounds based on clays used in the manufactory of pottery.34
So-called ‘quartz grounds’ have been identified on more than a quarter of 153 examined canvas paintings by Rembrandt made between 1642 and 1669.35 The term ‘quartz ground’ in the context of Rembrandt essentially refers to the use of a whitish sandy earth as a basis for the ground, various earth pigments could be added to alter the colour.36 There are indications that the type of clay used for the preparation of Rembrandt’s canvases may have been similar to the clay used in the brick and tile making industry and to the raw clay used by the potters in the making of earthenware.37 Although, the ground in Beeldemaker’s Dogs in a landscape does contain quartz, it is not a ‘quartz ground’ like those found in the paintings by Rembrandt. The ground layer in Dogs in a landscape differs in the sense that it also contains a large component of lead white.
31 K. Helwig, “Iron oxide pigments: Natural and synthetic,” in Artist’s Pigments. A Handbook of
Their History and Characteristics, volume 4, ed. B. H. Berrie (Washington: National Gallery of Art,
2007), 39.
32 Eastaugh et al., Pigment Compendium, 113 33 Stols-Witlox, A Perfect Ground, 52. 34 Ibid.
35 K. Groen, “Grounds in Rembrandt’s workshop and in paintings by his contemporaries,” in A
Corpus of Rembrandt Paintings. Vol IV, ed. Ernst van de Wetering (Dordrecht: Springer, 2005),
318-334.
36 Groen, “Grounds in Rembrandt’s workshop,” 328.
37 K. Groen, “The origin of the materials used for the preparation of the Night Watch and many other canvases in Rembrandt’s workshop after 1640,” Art Matters – Netherlands Technical Studies
23 Dubbeldam, UvA, 2019 It remains difficult to assess what is the source of quartz and possibly feldspar in the painting’s ground layer. The ground contains at least one of the earth pigments umber or sienna, which could have carried it as a contaminant. In addition, Beeldemaker may have used a sandy clay, such as potter’s earth, that was mixed with lead white and earth
pigments and a small amount of black pigment to adapt its lightness and colour. The use of clays in the ground could have been appealing to because of the affordability of the material.
2.4 Paint layer
2.4.1 Materials
The study of cross-sections in visible light reveals that the paint layer consists mainly of a greyish white mass. Apart from a few red pigment particles at the surface of the sample and a single large black pigment particles, the pigments in the paint layer can hardly be distinguished. When studied in ultraviolet light and in a backscattered electron image however, the presence of angular jagged particles is revealed (fig. 2.13).
The shape of the jagged particles is characteristic for crushed particles of glass, the SEM-EDX distribution maps of silicon and oxygen indeed clearly define the jagged particles, confirming their glassy nature (App. VI.a – VI.d). The same observation was made in the paint sample from the pendant painting (App. VI.e). It was assumed that Beeldemaker used the blue glass pigment smalt to paint his skies. Yet, in all of the cross-sections that were studied, not a single particle has retained a visible trace of their blue colour. It was shortly considered that Beeldemaker may instead have purposely added a colourless glass to his
Fig. 2.13 cross-section DM/970/465#4A. top left: BF 500x, top right: UV 500x, Bottom: BSE 800x. SDD detector. NovaNanoSEM450. BSE image: Ineke Joosten (RCE).
24 Dubbeldam, UvA, 2019 paint, the addition of which has been advised in historical treatises from the 15th to 17th centuries to improve the drying properties of the paint.38 However, smalt is a potassium silicate glass that was coloured blue by the addition of cobalt oxide (CoO) and the presence of cobalt that was detected by EDX confirmed that the glassy particles in the paint were indeed the pigment smalt (fig. 2.14).39
For a potassium silicate glass, the potassium content in the smalt is surprisingly low. This is in fact indicative of the discolouration of the pigment, which involves the leaching of potassium from the pigment (see section 2.4.2.1). The EDX spectrum further reveals the presence of magnesium (Mg) and a trace of iron (Fe), these elements are known to occur as impurities in smalt. 40
The relatively low peak for cobalt may indicate that Beeldemaker opted to use a paler variety of smalt. Different grades of smalt were available which depended on the cobalt content and the particle size of the smalt.41 The smalt particle size is typically medium to very coarse, the pigment loses its colour considerably when it is finely ground.42 According to the particle
38 K. Lutzenberger, H. Stege & C. Tilenschi, “A note on glass and silica in oil paintings from the 15th -17th century,” Journal of Cultural Heritage 11 (2010): 371-372.
39 B. Mühlethaler and J. Thissen, “Smalt,” in Pigments: A Handbook of Their History and
Characteristics, Volume 2, ed. Ashok Roy (Washington: National Gallery of Art, 1993), 113.
40 Mühlethaler and Thissen, “Smalt,” 115. 41 Ibid., 113.
42 Eastaugh et al., Pigment compendium, 581.
Fig. 2.14: Cross-section DM/970/465#4A. Top left: BSE. Top right: smalt phase. Bottom: EDX spectrum corresponding to the smalt phase (see also App. VI.f). SDD detector.
25 Dubbeldam, UvA, 2019 size classification for pigments as proposed by Feller and Bayard, the smalt particles in Dogs in a landscape (5-10 µm) can be considered large to coarse.43
The EDX spectrum of the phase that surrounds the smalt particles, presents besides the high counts for lead - indicative of the presence of lead white - also a significant portion of silicon. It furthermore contains traces of Na, Mg, Al, Ca and Fe (fig. 2.15). The phase is continuous with part of the ground layer, suggesting that the composition of the paint film is partly similar to that of the ground. The admixture of a small amount of a white clay such as kaolin (Al4[Si4O10](OH)8) 44 to the smalt and lead white may serve as an explanation for the presence of silicon and the other trace elements in the paint layer.
43 R.L. Feller and M. Bayard, “Appendix. Terminology and Procedures Used in the Systematic Examination of Pigment Particles with the Polarizing Microscope,” in Artists’ Pigments. A
Handbook of their History and Characteristics, volume 1, ed. R.L Feller (Washington: National
Gallery of Art, 1986), 286.
Particles < 0,3 µm = very fine, 1,0-0,3 µm= fine, 3–1 µm= medium, 10-3 µm= large, > 10 µm= coarse.
44 Eastaugh et al., Pigment Compendium, 214.
Fig. 2.15: Cross-section DM/970/465#4A. top left: BSE. Top right: phase outside the smalt particles is continuous with part of the ground layer. Bottom: corresponding EDX spectrum (see also App. VI.f). The spectrum reveals the presence of lead white but additionally the peak for silicon and traces of Na, Mg, and Al suggests that it may also contain clay. SDD detector. NovaNanoSEM450 (Thermo Scientific). Pathfinder Software. Analysis: Ineke Joosten (RCE).
26 Dubbeldam, UvA, 2019 In ultraviolet light, a very thin paint layer (≈5 µm) with a slightly different fluorescence compared to the main paint layer can be recognized (fig. 2.16). This thin layer of paint was used to give the paint layer a pink tone. In the layer, jagged smalt particles and a few red pigment particles are present. The high level of mercury (Hg) and sulfur (S) detected in these red pigments reveals that it concerns the mercury sulphide (HgS) pigment vermilion (fig. 2.17).
Figure 2.17: Cross-section DM/970/465#4A. top left: BSE. Top right: BF. Bottom; EDX spectrum of circled red pigment particle. A high concentration of Hg and S is revealed,
identifying the pigment as vermilion (HgS) (see also VI. a). SDD detector. JEOL JSM-5910LV. Noran System 7 Software. Analysis: Maartje Stols-Witlox (UvA).
Fig. 2.16: Cross-section DM/970/465#4A. UV. 500x. Leica DM2500 microscope. Along the top of the paint surface a thin paint layer (indicated with bracket) with slightly different fluorescence is detected. It contains smalt and vermilion.
27 Dubbeldam, UvA, 2019 2.4.2 Condition
The whitening of the paint surface, which is the focus of this thesis is the most obvious degradation phenomenon related to the paint layer, this phenomenon will be elaborately described in chapter 3. Here will be discussed some of the other aspects of the paint layer’s condition that may directly or indirectly provide insight into the formation of the whitening. 2.4.2.1 Discoloration of the smalt pigment
One of the main alterations in the paint layer has already been introduced above, this is the discoloration of the smalt pigment. In the cross-sections that have been prepared from the painting, the pigments have lost its blue colour completely. Because the prepared cross-sections represent only a tiny fraction of the paint surface it is not known whether the smalt pigment has discoloured to this extent all across the paint surface. The paint still presents a pale blue colour so perhaps some pigment particles have at least partly retained their blue colour.
It is difficult to assess to what extent the pigment discoloration has visually altered the perception of the colours of the paint layer. The low signal for cobalt measured with the EDX reveals that potentially a pale grade of smalt was used. Furthermore, smalt – as characteristic for a glass – is highly transparent, the pigment’s refractive index (1,46-1,55) is very close to that of linseed oil (1,484). The pigment has a weak tinting strength and thus a poor hiding power.45 Mixed with lead white, the smalt containing paint must originally already have been characterised by a relatively subdued blue colour which was appropriate for painting skies and backgrounds. Along the top of the painting a more vivid blue colour is visible, possibly a different type of blue pigment has been used for this part of the painting but this has not been further investigated (fig. 2.18).
45 J. Plesters, “A Preliminary Note on the Incidence of Discolouration of Smalt in Oil Media,”
28 Dubbeldam, UvA, 2019 In large parts of the sky and background, the colour of the brown ground layer shines through the paint layer. There seems to be a some loss of colour balance, especially in the upper right corner of the sky (fig. 2.18). Although the colour of the ground may have been partly intended to shine through the ground layer, its effect has become exaggerated. The loss of the blue colour of the smalt in combination with an increased transparency of the paint with aging may explain a visual darkening of the paint layer.
The leaching of potassium from the smalt pigment has been identified by many as the mechanism by which smalt loses its colour. 46 This process induces a structural change of the cobalt ion from a tetrahedral coordination (blue) to a octahedral coordination (light pinkish).47 Spring et al., report for several paintings the potassium oxide (K2O) concentrations from both well preserved smalt particles and degraded particles. The well preserved particles contain approximately 10-15 wt% K2O whereas the discoloured particles contain only 0,5-3,5 wt% K2O.48 The discoloured smalt pigmentsin Dogs in a landscape as
46 J. Boon et al., “Imaging Microspectroscopic, Secondary Ion Mass Spectrometric and Electron Microscopic Studies on Discoloured and Partially Discoloured Smalt in Cross-sections of 16th Century Paintings,” CHIMIA 55, no. 11 (November, 2001): 952-960; L. Robinet, M. Spring and S. Pagès-Camagna, “Investigation of the Loss of Colour in Smalt on Degradation in Paintings using Multiple Spectroscopic Analytical Techniques,” in ICOM Committee for Conservation 16th triennial conference Lisbon, 19-23 September 2011: preprints, ed. J. Bridgland and C. Antomarchi (Almada:
Critério, 2011), 1-8; M. Spring, C. Higgit and D. Saunders, “Investigation of Pigment-Medium Interaction Processes in Oil Paint containing Degraded Smalt,” National Gallery Technical Bulletin 26, (2005): 56-70.
47 Robinet, Spring and Pagès-Camagna, “Loss of Colour in Smalt,” 1-8.
48 Spring, Higgit and Saunders, “Pigment-Medium Interaction Degraded Smalt,” 70. Fig. 2.18: Detail sky and background. DM/970/465.
Adriaen Cornelisz Beeldemaker. Dogs in a landscape. 1697. Oil on canvas. 44,7 x 49, 1 cm. Dordrecht, Dordrechts Museum. Photograph: Ivana Jerdonekova.
29 Dubbeldam, UvA, 2019 measured in sample #4A clearly fall under the category of degraded smalt as only a K2O concentration of 1,9 wt% could be detected (App. VI.f see phase 4).
The migration of potassium out of the smalt pigment can only take place in the presence of water; water diffuses into the glass and through ion exchange reactions alkali metals leave the glass as hydroxides. In the absence of water, either as a liquid or a vapour, the discoloration of smalt test samples does not occur.49 It is known that Dogs in a landscape has been exposed to high quantities of water during its last restoration in 1971 (see chapter 4) this may have played a big part in the discoloration of the pigment. On the other hand, it is possible that the discoloration had already been induced prior to 1971 through the exposure to moisture in the environment.
Sample #4A showed an additional phenomon that must be related to the degradation of the smalt pigment. Around the smalt pigments, two phases are recognized, each with a clear increasing peak height of lead relative to silicon (fig. 2.19).
49 I. Cianchetta et al., “Discoloration of the smalt pigment: Experimental studies and ab initio calculations,” Journal of Analytical Atomic Spectrometry 27, no 11 (November 2012): 1942. Fig. 2.19: Cross-section DM/970/465#4A. Lead rich phases (phase 18 and 19) are present around the smalt particles (phase4). See also App. VI.f SDD detector. NovaNanoSEM450 (Thermo Scientific). Pathfinder Software. Analysis: Ineke Joosten (RCE).
30 Dubbeldam, UvA, 2019 A similar observation has been made in cross-sections of discoloured smalt paints in paintings by Pedro Campaña (1503-1586) and a painting by Joachim Beuckelaer (ca. 1533 – ca. 1574), here lead rich rims were recognized around the smalt particles in an electron backscattered image.50 It was noted that the enrichment of lead at the pigment surface must be related to the degradation of the smalt since it was not seen around two well preserved particles in the painting by Beuckelaer. 51 This enrichment of lead on the surface of the smalt, was not detected in sample #6A, which has been taken from the paint affected by the whitening (App.VI.g). Therefore, this enrichment of lead around the smalt particles may not be related to the whitening phenomenon on the paint surface.
2.4.2.2 Bubbles and microscopic deformations in the paint layer.
A peculiar phenomenon that was encountered on the paint layer and it may prove important for understanding the cause of whitening of the paint surface. A lot of small ‘bubbles’ are present in the paint layer of the sky of both Dogs in a landscape and its pendant (fig. 2.20).
These occur in areas affected by the whitening and not affected by the whitening. Although these bubbles look similar to craters that may be encountered in paint films containing aggregated lead soap protrusions, the bubbles encountered in Dogs in a landscape are all empty. Additionally, in other places the phenomenon takes on more amorphous shapes (fig. 2.21). It assumed that these structures are the result of the softening and boiling of the paint as a consequence of the exposure of the painting to excessive heat during its wax-resin lining. The exposure to heat was not always easily controllable with traditional irons
50 Spring, Higgit and Saunders, “Pigment-Medium Interaction Degraded Smalt,” 66. 51 Ibid.
Fig. 2.20: DM/970/465. Dino-Lite photograph. Left: 55 x. Right: 245x.. Bubbles have formed in the paint layer.
31 Dubbeldam, UvA, 2019 and it could easily lead to burning of the paint. The exposure of the paint to excessive heat may have played a role in the formation of the whitening on the paint surface.
Fig. 2.21: DM/970/465. Dino-Lite photograph. Magnification 265x. Microscopic amorphous deformations in the paint.
32 Dubbeldam, UvA, 2019
3. CHARACTERISATION OF THE WHITENING PHENOMENON
3.1 Visual characteristics of the whitening and its distribution across the paint surface.
When the distribution of the whitening across the paint surface is considered, one of the most important observations that can be made is that the whitening has occurred only in the background of the painting, where distant mountains and the sky have been depicted (fig. 3.1 and App. I.h). This observation could be of great significance, it suggests that the phenomenon may be related the pigments that have specifically been used to paint this area. In the left upper corner of the sky, the whitening is largely absent. In this part, the composition of the paint may differ slightly compared to the rest of the sky, the paint is lighter and includes stronger pink tones and a yellow paint. The ultraviolet light photograph taken after varnish removal reveals however that the yellow areas consist mostly of overpaint (App. I.g). Potentially more whitening or another form of damage is present underneath this paint. In any other sense, the whitening does not conform to the painted composition, the whitened areas are seen to cross the boundaries in which the mountains morph into the sky and clouds.
A striking feature of the whitening is the pattern in which it has manifested itself across the paint surface. The phenomenon has created a pattern of parallel vertical stripes. (fig 3. 1, App. III.a). Interestingly, this pattern of the whitening had already been detected in an infrared reflectogram of the painting that was taken before the start of the 2018-2019 restoration (App. I.f). It shows a pattern of light stripes across the surface, the nature of which was not yet understood at that time. The pattern detected in the infrared reflectogram
Fig. 3.1: Adriaen Cornelisz Beeldemaker. Dogs in a landscape. 1697. Oil on canvas. 44,7 x 49, 1 cm. Dordrecht, Dordrechts Museum. Right: mapping of the paint surface affected by whitening. See also App. III.a. Photograph: Ivana Jerdonekova, Mapping: Sofie Dubbeldam.
33 Dubbeldam, UvA, 2019 coincides very nicely with the pattern of the whitening that was uncovered during the 2018-2019 restoration. The same type of alteration in or on the paint that has caused the whitening must thus have caused the infrared radiation to be more strongly reflected. The difference in reflection cannot be explained by the presence of the old retouches that were applied on top of the whitening, these had been applied much more broadly than the fine pattern of stripes detected in the infrared reflectogram (App. III.e). In ultraviolet light, the whitened paint surface looks slightly lighter but it does not have a significant fluorescence (App. I.g).
The manifestation of the whitening in this pattern of vertical parallel stripes could be related to the structure of the canvas. As has been shortly discussed in section 2.1 the study of the painting in raking light reveales that the surface of the painting is uneven (App. I.d – I.e). Only in the vertical direction and not the horizontal direction of the painting parts of the paint surface are slightly raised. Possibly the structure of the canvas has played a role in the distribution of the whitening across the surface in this vertically striped pattern.
This potential relation between the whitening and the structure of the canvas is further supported by the observation that also on a smaller scale, in some areas, the pattern of the whitening is reminiscent of the structure of a canvas (fig. 3.2). When viewed in raking light, the structure of the canvas is indeed quite dominantly present in the paint layer. The whitening seems to occur mostly at the interstices of the canvas.
The whitening does not present itself in this canvas like pattern everywhere, in other areas the whitening is continuous and more closed. (fig. 3.3) A more in depth discussion on how the structure of the canvas may relate to the pattern of the whitening across the paint surface will be included in section 4.3.
Fig. 3.2: DM/970/465. Dino-Lite photograph. 25x magnification. Left: The whitening has adopted a canvas like pattern. Right: same area in raking light from below. The structure of the canvas is quite dominantly present in the paint layer, the whitening appears to occur mostly in the interstices of the canvas.
34 Dubbeldam, UvA, 2019 Within the vertically oriented stripy pattern of the whitening there remains also a certain randomness to the pattern of the phenomenon. The whitened stripes vary in width and are sometimes interrupted along their length. Within the striped pattern the whitening takes on varying and unpredictable shapes. In some places the border between the affected and unaffected paint is abrupt, whereas in other places this transition is more gradual. The fact that whitened and non-whitened areas exist directly next to each other within the same colour areas, reveals that factors other than the composition of the paint alone must have played a role in the formation of the whitening.
Rounded shapes of unaffected paint can be detected in areas that have been affected by the whitening, it is unclear how these may have formed but they may later provide an important hint for understanding the underlying cause that led to the formation of the whitening (fig. 3.4).
Fig. 3.3: DM/970/465. photograph. 25x
magnification. An area of the whitening that has not adopted a canvas like pattern but is more closed and continuous.
Fig. 3.4: Detail whitened sky. Adriaen Cornelisz Beeldemaker. Dogs
in a landscape. 1697. Oil on canvas. 44,7 x 49, 1 cm. Dordrecht,
Dordrechts Museum. Right: mapping of part of the paint surface that shows the rounded shapes of unaffected paint that occur within the whitening pattern.
35 Dubbeldam, UvA, 2019 Although the term whitening is a good general term to describe the overall phenomenon, the actual colour of the affected paint surface may vary. Overall, the surface affected by the whitening has an off white and slightly grey tone. Those areas where the unaffected paint has a stronger pink colour, the whitened paint surface also presents itself with a pinker tone (fig 3.5). The study of the cross-sections has revealed that the pink colour of the paint is imparted by the presence of a thin paint layer containing smalt and the red pigment vermilion. The adoption of a pink tone of the whitening may mean that the whitening phenomenon is covered by this thin vermillion containing paint layer, suggesting that it concerns a degradation within the paint layer and not on top of the paint layer.
3.2 Microscopic study of the whitening.
The study of the whitening phenomenon microscopically also revealed that the phenomenon is likely a form of degradation of the paint film itself rather than the result of residues of a material left on the surface from a past restoration. The whitened paint surface is interrupted by the aging cracks of the paint layer (fig. 3.6). If the whitening were to be caused by residues from a material not inherent to the painting technique, then the whitening would be expected to cover the aging cracks in the paint.
Another observation that corroborates the fact that the whitening concerns a degradation of the paint layer itself is the presence of several paint losses. In these losses the delamination has taken place at the interface between the ground and paint layer, revealing the dark coloured ground. Interestingly, the paint losses have been confined only to the areas that have been affected by the whitening and must thus relate to the condition of the paint in these areas (fig. 3.7).
Fig. 3.5: Normal light. Detail. The whitening adopts a pinker tone when it crosses into a pinker area.
Fig. 3.6: DM/970/465. Dino-Lite AM4113zT photograph. 55x magnification The whitening is interrupted by aging cracks in the paint layer.
36 Dubbeldam, UvA, 2019 The borders of the paint losses are quite irregular and it seems that these losses did not occur along the aging cracks in the paint (fig. 3.8) All this suggests that the coherence within the paint film has weakened and that some form of physical disintegration and embrittlement of the paint may have taken place. Additionally, there must have been a decrease in the adherence between the paint and ground layer.
Fig. 3.7: DM/970/465. Dino-Lite AM4113zT photographs. a. 45x, b. 50x, c. 50x, d. 40x, e. 50x, f. 50x. Paint losses have occurred that are confined to the areas affected by the whitening. The losses reveal the brown coloured ground layer.
Fig. 3.8: Hirox 3-D digital microscope photograph. Scale-bar: 2mm. Paint loss in whitened areas. The borders of the loss are jagged and do not seem to follow aging cracks.
