The author would like to thank Vera Blok, painting conservator and teacher at the University of Amsterdam.

Suzan de Groot (Conservation Scientist), Ineke Joosten (Conservation Scientist) and Klaas Jan van den Berg (Senior Conservation Scientist) are based at the Cultural Heritage Agency of the Netherlands (RCE). They supervised and performed the technical research. Eric Domela Nieuwenhuis (curator, RCE) for introducing the subject of this research and providing access to the paintings, Zeph Benders (conservator, RCE) for all his help in the studio and the permission to take paint samples and scrapings, Kate Seymour (SRAL) for her help in writing this article and Emilie Froment (UvA) for correcting the French abstract.

Introduction

1Willem Witsen (1860-1923) was a painter from Amsterdam. His household effects came into the possession of the State of the Netherlands when his wife died in 1943. They are now part of the Collectie Nederland, which is administered by the Cultural Heritage Agency of the Netherlands (RCE). Part of these household effects is a collection of 57 paintings that were present in his studio when Witsen died. They vary widely in size, painting technique, type of support, and condition. The panels are made of wood or veneer. The stretched canvases are attached to a frame or occasionally cardboard, while the unstretched canvases are rolled up. Most paintings in this collection are thought to be unfinished or sketches in paint. Their registration at the RCE contains inventory numbers, but the paintings do not have official titles.

2 There are white phenomena present on the paint surfaces of 24 paintings within this collection, creating white spots or whitish, matt hazes (Table 1). These 24 paintings are all unvarnished. The white phenomena form an aesthetic problem as these disturb the image, change the colour balance, and reduce the contrast (Fig. 1). Furthermore, they may also present a conservation problem. Investigation of the phenomena will increase the understanding of the conservation conditions of these paintings.

3 The group of 24 unvarnished paintings by Willem Witsen containing these white phenomena presents a unique case study because it involves a large number of paintings, all made by the same painter but on different supports, probably never treated, and stored together in the same locations since at least 1923. Correlations between the studied problem and a certain property of the paintings are statistically more reliably established when the paintings have more in common and the number of objects is higher. Analysis of the white phenomena will therefore not only increase the understanding of the conditions of these particular paintings, but also contribute to greater insight into white phenomena on painted surfaces in general.

4Examination of the paintings shows that the white phenomena consist of various white deposits. Five types were identified: small spheres, both needle and thread shaped crystals, grit-like particles, and homogeneous whitish layers. All of the types of phenomena noted consist of various lead compounds or, in the case of the spheres, pure starch. The discovery of the starch spheres was very unexpected. Research into the phenomena and the original materials of the affected paintings may increase the understanding of the formation process of the white phenomena, leading to more appropriate decisions for treatment and preventive conservation of the paintings. This article is focused on the characterisation of the white phenomena and the determination of their formation process.

Formation of white deposits on painted surfaces

5White deposits on painted surfaces are a known problem and have been the subject of an increasing number of publications in the last fifteen years (Koller and Burmester 1990; Hinde et al 2011; Van Loon et al 2011; Sawicka et al 2013). They have been found on paintings and painted objects from all periods, and occur in conjunction with all kinds of supports, grounds and painting techniques. There are several terms for these phenomena that seem to be used interchangeably, including but not limited to the following: efflorescence, blooming, white haze, chalking, and blanching. In most cases, the deposits consist of crystallised material, such as salts, free fatty acids or metal soaps (Koller and Burmester 1990; Hinde et al 2011; Van Loon et al 2011; Sawicka et al 2013). This research has proved that the source of the phenomena is from within the painting, sometimes in combination with reactions with materials from the atmosphere.

6 The paint surface can appear white if components from within the painting, such as free fatty acids, metal ions, metal soaps or salts, migrate to the surface and crystallise, sometimes after further reaction with gaseous material from the atmosphere (Ordonez and Twilley 1997). Metal ions, free fatty acids and lead soaps are known to have the ability to move through the paint layers as a result of climate fluctuations or moisture gradients (Boon et al 2007; Sawicka et al 2013). Oil paint consists of pigments bound in an organic medium, often a drying oil. Additives, such as extenders or siccatives, may also be present, especially in more modern paints. Migratory components in oil paint are free fatty acids and metal ions, which may react with each other. Lead carboxylate (lead soap) is a reaction product of the saponification of free fatty acids with lead white, which can occur if the paint is medium rich (Noble et al 2008). The free fatty acids may originate from added metal stearates which are quite common in 20th Century paint (Tempest et al 2013). Inter-layer reactions are also possible. Lead carboxylates can react further to form various lead salts on the paint surface (Boon et al 2007).

7 To date there are very few articles about the treatment of lead containing deposits on the paint surface. The lead compounds that are mentioned in the literature, such as lead carboxylates and sulphates, cannot be put into solution in organic solvents, but they do react to reagents. Aqueous gels containing chelators, with adjusted concentration, pH and application method, can be designed to remove specific salts (Van Loon et al 2011; Sawicka et al 2013).

Experimental methods

8For the analysis of the white phenomena, all 24 unvarnished paintings were examined without magnification, and under the stereo-microscope. Solubility tests were conducted to give further information. As a result of these observations, eight paintings were selected for sampling and further research.

10After examining their surfaces, the samples were embedded in epoxy resin and polished, so that their stratigraphy could be analysed. The white phenomena were investigated again with optical microscopy, SEM-BSE and EDX. Additional analyses of the paint and ground layers included the same analytical methods, and aimed at determining whether or not the white material originates within the painting or derives from an outside source.

12Finally, all observations and measurements were compared.

Results & Discussion

13The observations that were made without magnification vary by painting. Some paintings have areas that are brighter and more matt than the surrounding healthy paint, while others have opaque white or light blue areas or little dots (Table 1). The phenomena are present in specific locations, sometimes in combination with a particular colour (Fig. 2), or they are distributed randomly on the paint surface. A distinction can also be made between little dots and larger areas (Fig. 3). None of the tested phenomena are soluble in water, iso-octane, acetone, ethanol or isopropanol.

14 The following five types of phenomena can be distinguished on a microscopic level: small spheres (Fig. 4), needle-shaped crystals (Fig. 5), thread-shaped crystals (Fig. 6), grit-like particles (Fig. 7), and a homogeneous layer. The latter is not composed of different particles, but appears as a white, matt, semi-transparent layer on top of the paint layer (Fig. 8). Each painting showing the white phenomena contains one or more of these types. Comparing the observations did not present direct correlations between the type of support or condition of the painting and the distribution and appearance of the white areas or the type of phenomenon (Table 1).

15 SEM-BSE/EDX analyses show that the grit-like particles, the needle- and thread-shaped crystals, and the homogeneous layer contain lead compounds (Fig. 9). FTIR analyses of scrapings of the different phenomena result in spectra relating to lead carboxylates and three types of lead salts derived from a small, saturated acid such as acetic acid, propanoic acid or succinic acid (X, Y, and Z in Table 2). A direct correlation between the observed phenomenon and the chemical composition of these lead compounds can only be established in the case of the needle-shaped crystals. Possibly, all these phenomena are different stages of the same process, in which lead carboxylates migrate to the surface and then react further to form different lead salts. The exact compositions of the three similar lead salts derived a small, saturated acid are unknown, but could be of interest because they possibly have not been published before in this field of research. However, more research using gas chromatographic techniques should be performed to establish final proof.

16 SEM-BSE/EDX analysis shows that the spheres, surprisingly, contain only organic material (Fig. 10). FTIR analysis shows that they consist of pure starch (Table 2). The thread-shaped crystals that were on the spheres appear to consist of cellulose. Examination of existing publications about white phenomena on painted surfaces did not yield any previous accounts of starch spheres or other types of starch particles.

Table 2: Results of FTIR analyses of the scrapings of the white phenomena

Painting (inv.no.)	Type of phenomenon	FTIR results
w190	Grit-like particles	Lead carboxylate
w1818	Thread-shaped crystals	Lead carboxylate
w314	Homogeneous layer	Lead salt derived from a small, saturated acid (X)
w1829	Needle-shaped crystals	Lead salt derived from a small, saturated acid (Y)
w1829	Little spheres with thread-shaped crystals	Spheres: starch Thread-shaped crystals: cellulose
w1827	Needle-shaped crystals	Lead salt derived from a small, saturated acid (Y)
w271	Thread-shaped crystals	Lead salt derived from a small, saturated acid (Z)
w1822	Little spheres with thread-shaped crystals	Spheres: starch Thread-shaped crystals: cellulose

17 There is no doubt that the lead present on the paint surfaces originates from the painting itself instead of the surroundings. The grounds of all paintings sampled contain lead white, in some cases in combination with chalk (Fig. 11). Since these ground layers contain much more lead white than the paint layers, it is presumed that the lead originates from the lead white pigments in the grounds. The paint layers on top of the ground layers all seem to be very medium rich, because they contain relatively few pigments and can therefore be expected to be a good reservoir of free fatty acids. The lead from the grounds most probably reacted with these free fatty acids to form lead carboxylates, while a smaller proportion of lead reacted with free fatty acids from the ground layer itself. This is exemplified by one example with crystallised lead compounds on the surface, where no paint layers were present. If the lead is in contact with a medium rich paint layer, however, there is a much greater chance that the free fatty acids of this paint layer were saponified instead of those in the ground layer. Furthermore, migration is easier in layers with relatively few pigments. Dark paints often have a low pigment concentration, which may explain why some white phenomena follow darker areas (Noble and Van Loon 2010).

18 For many years, this collection was stored in the studio of Willem Witsen, where it was exposed to potentially strong regular as well as occasional climate fluctuations (Molinari 2013: 49-51). This contributed to the migration of the lead carboxylates to the surface. Once the lead carboxylates reached the paint surface, they crystallised or reacted further to form a lead salt derived from a small saturated acid. The composition and the origin (the painting itself or the surroundings) of the reagents the lead carboxylates reacted with remain unknown.

19The starch spheres are observed on all rolled canvases, but also on paintings on panel. Their origin remains unknown. It is expected that larger particles, such as these starch spheres, cannot migrate through a paint layer without creating cracks unless the paint is not completely cured yet. This, and the shared storage history of the paintings, might be an indication that the starch is an external material that deposited on the paint surface. However, it is also possible that the starch spheres are original components of the painting itself, since they have been observed below the paint layer with stereo-microscopy (Fig. 12). In the scope of this research, it was not clarified whether the spheres can be found in the ground or paint layers. The presence of starch in the original painting materials and the capacity of starch spheres to migrate through a paint layer could be very interesting subjects for further research.

Conclusions

20Five types of phenomena could be observed on the 24 investigated paintings by Willem Witsen with a stereo-microscope. One of them, the spheres, consists of pure starch. The other types consist of lead carboxylates and lead salts derived from a small, saturated acid, probably containing lead from the ground layers. Further research on the ground and paint layers is needed to establish the exact origin of the different materials that caused the white phenomena. Further research into the possibility of the migration of starch spheres could also lead to interesting results.

21 Climate fluctuations in the past probably contributed to the formation of the white phenomena on the paintings by Witsen. Well regulated climate conditions are therefore of major importance to the conservation of this collection: to inhibit any further formation of the white phenomena on the studied paintings as well as to prevent the formation of these phenomena on the unaffected paintings. Each painting will need its own treatment proposal, since there is wide variety in both the used painting materials and the types of phenomena. The knowledge that was gained regarding the compositions and origins of the white deposits on the paint surfaces could, when combined with known and published treatments of similar white phenomena, form a good starting point in the search for an appropriate treatment for each individual painting.