A Dark Discoloration on Aristide Maillol’s Outdoor Lead : Identification, Formation, and Further Research

Manuela Toro* Ineke Joosten Joy Bloser Lynda Zycherman University of Amsterdam National Heritage Laboratory The Museum of The Amsterdam, The Netherlands Rijksdienst voor het Cultureel New York NY, USA New York NY, USA [email protected] Erfgoed (RCE) [email protected] [email protected] Amsterdam, The Netherlands *Author for correspondence Tonny Beentjes [email protected] University of Amsterdam Amsterdam, The Netherlands [email protected]

Abstract Two lead sculptures by Maillol, one which exhibits a dark red research into the sculptures’ casting history, past treatment, discoloration of the surface and one which does not, are held outdoor weathering patterns, and collections care practices in the collections of the Kröller-Müller Museum, Otterlo, and was compared. XRF and SEM-EDX analysis did not point out the Museum of Modern Art (MoMA), New York, respectively. substantial differences between the objects. Their treatment In an effort to understand better the formation of the histories and environmental conditions revealed possible corrosion layer, the two works were analyzed and compared future paths of research. in a collaborative effort between authors at the University of Amsterdam and MoMA. Technical analysis using scanning Keywords electron microscopy with energy-dispersive x-ray spectroscopy Maillol, lead, outdoor , lead dioxide (SEM-EDX) and x-ray fluorescence (XRF) was performed, and

Introduction

The French sculptor Aristide Maillol (1861–1944) cast a exhibit any dark discoloration. L’Ai r (3/6, cast in 1962 selection of his sculptures in lead together with founder by Georges Rudier Foundry), held at the Kröller-Müller Eugène Rudier throughout the early 20th century. In Museum (KMM) in Otterlo, the Netherlands, has also recent years, an unexplained, dark red discoloration has been on display in the outdoor sculpture garden since been noted on a cast edition of Maillol’s L’Ai r (The Air) its acquisition in 1962, and a dark red discoloration displayed outdoors, while another cast edition of Maillol’s appears to follow its water runoff patterns. The casting La Rivière (The River), also displayed outdoors, has histories, past treatments, outdoor weathering patterns, retained the expected blue-gray appearance of an oxidized and ongoing preventive care of the two sculptures were lead surface. The physical change in surface appearance compared and both submitted to technical analysis by suggests a chemical change is occurring on the surface scanning electron microscopy with energy-dispersive of some objects outside the expected behavior of lead x-ray spectroscopy (SEM-EDX) and x-ray fluorescence oxidation in an outdoor environment. Research on lead (XRF). X-ray diffraction (XRD) analysis on samples from discoloration of outdoor sculptures in Queluz, Portugal, L’Ai r identified a mixture of two polymorphs of lead(IV) supported by the World Monuments Fund suggests this oxide as the corrosion products present in the discolored discoloration is a mixture of lead(IV) oxide and a lead areas (Joosten and van Hoesel 2016). oxycarbonate. The complete corrosion mechanism, This investigation studied the corrosion mechanisms of however, was not identified (Bernard et al. 2010). the discoloration forming on the surface of select Maillol La Rivière (no. 2, cast in 1948 by Alexis (Eugène) Rudier lead sculptures and considered how their different casting Foundry), held at the Museum of Modern Art (MoMA), histories, varying preventive care, and treatment proto- New York, has been on display in the outdoor sculp- cols may affect or possibly prevent the discoloration of ture garden since its acquisition in 1948 and does not lead over time. 2 ICOM-CC | METAL 2019 | NEUCHÂTEL, SWITZERLAND CORROSION STUDIES

Foundry history a natural oxidation layer in lead, but a dark red color can be seen on most of the surface of L’Ai r in the form Alexis Rudier set up his foundry in in 1874 and of stains and drips (Figure 1). Its thin composition has a after his death in 1897, his son Eugène ran the foundry. patina-like effect on several areas of the sculpture. Upon Maillol formed a close working relationship with Eugène close examination, white powdery spots can also be seen Rudier that lasted more than 39 years. After Maillol’s in the darkened areas (Figure 2). death in 1944, Eugène continued to work with Maillol’s estate until his death in 1952. Despite Eugène’s successful L’Ai r was cared for in the same way as outdoor bronzes foundry, he inscribed all of his work under his father’s until the early 1990s: with light cleaning with soapy name, “Alexis Rudier.” water and applications of a beeswax coating once a year. Treatment documentation does not indicate exactly After Eugène Rudier’s death, his cousin, Georges Rudier, what kind of soap, water, beeswax, or tools were used started a foundry despite Eugène’s wish that the Rudier or in what way. The period in which these treatments foundry close upon his death. Georges retained Eugène’s stopped seems to correspond with the approximate date workers and began to produce work for the Maillol on which the dark red stains were first spotted. Since estate and the Musée Rodin and for other contemporary the wax coating would have isolated the sculpture from sculptors of the time. However, both the Maillol Museum water during most of its life, this further confirms the and Musée Rodin ceased working with Georges around impact of water on the object, as do the drip-like stains. 1976 among rumors of suspicious activity at the foundry At this point, there seem to be no traces of beeswax left (Lebon 2014, 133–41). on the sculpture.

History of the sculptures, their condition, and Condition reports begin to mention the dark corrosion treatments stains around 1993, but no further descriptions are provided until 2001. Photographic documentation has Maillol’s L’Ai r depicts a reclined female nude first real- provided some indication of how and when the red stains ized in stone in 1938 for a monument commissioned by started to appear and how quickly they developed. A the city of Toulouse. The second version was cast after major increase was observed between 2001 and 2005 his death in six lead editions and a number of bronze (Mahu 2016). editions. The lead edition 3/6 was cast in 1962 and is installed outdoors in the KMM’s gardens. The corrosion products in Maillol’s L’Ai r are not only distracting to the eye, but also pose risks to the sculp- ture’s structure and integrity if the corrosion is active. Examples of dark red corrosion on lead sculptures continue to be found elsewhere, yet other lead sculp- tures, such as MoMA’s La Rivière, do not yet exhibit this condition.

Figure 1. Aristide Maillol, L’Air, 3/6, 1939, cast 1962. KMM, 2018

Early documentation of the KMM’s L’Ai r shows a dark gray surface polished to an even sheen. As it is installed in the museum’s outdoor garden, the object is often touched by visitors. Today, the sculpture is of a dull gray- blue color overall, which is the expected appearance of Figure 2. Condition detail of L’Air. KMM, 2018 A DARK DISCOLORATION ON ARISTIDE MAILLOL’S OUTDOOR LEAD SCULPTURES: IDENTIFICATION, FORMATION, AND FURTHER RESEARCH 3

Maillol’s La Rivière was originally commissioned as a monument to Henri Barbusse in late 1938. Eugène Rudier cast the work in an edition of two, plus the artist’s proof, in 1948. Two subsequent editions of six, in bronze and in lead, were cast in the 1960s by Georges Rudier. MoMA acquired La Rivière no. 2 from the original set of casts in early 1949, and subsequently installed the work outdoors, where it has been on view almost continuously since 1953 (Figure 3).

Figure 5. Condition detail of La Rivière. MoMA, 2018

tive treatment on the sculpture occurred in 1952: it was washed with a mild soap and water, rinsed with benzine, and polished with a proprietary metal polish. The sculp- ture was then installed over the reflecting pool in 1953 and was regularly rinsed with water, occasionally with a mild detergent, and waxed periodically into the 1970s.

Treatment protocol since the early 1980s has been to rinse the sculpture twice a year with water, and wash Figure 3. Aristide Maillol, La Rivière, no. 2, 1938–43, cast 1948. Installation with a mild detergent and light mechanical action once view in the garden. Image by Soichi Sunami, MoMA, prior to 1953 every few years. It is not within touching distance of the public. The sculpture has developed dark black accretions Photographic documentation of MoMA’s La Rivière just in between the water run-off patterns on the sculpture. after acquisition shows the sculpture with a middle-gray There is no indication of dark red or purple staining in tone and the surface polished to an even sheen (Figure 4). these runoff areas. While it largely retains this middle-gray tone overall, dark discoloration where water drips from the sculpture’s Overview of lead corrosion undercut surfaces is becoming more prominent (Figure 5). When exposed outdoors, lead usually corrodes until it develops an adherent film of insoluble lead compounds that protect the surface from further attack. This behavior differs from what has been observed on L’Ai r . The sculp- ture started to show brown/red stains on the surface that developed and darkened in the relatively short time span of approximately twenty years. The corrosion products that formed on it were identified in 2016 as plattnerite

(β-PbO2) and scrutinyite (α-PbO2) by SEM-EDX and XRD analysis (Joosten and van Hoesel 2016). These compounds are polymorphs of lead(IV) oxide (also

known as lead dioxide, PbO2) which have been found Figure 4. Aristide Maillol, La Rivière, MoMA. Image by Thomas Griesel, on lead water-supply pipes. The corrosion mechanisms MoMA, 2014 by which plattnerite and scrutinyite form at the surface of the sculpture are yet to be fully explained. A complete treatment history for MoMA’s La Rivière (697.1949) is not available, but notes in the object’s record Other cases of outdoor lead sculptures showing dark red provide a general scope of treatment. The most interven- stains include that of the Palace of Queluz in Portugal, 4 ICOM-CC | METAL 2019 | NEUCHÂTEL, SWITZERLAND CORROSION STUDIES where plattnerite was also found (Bernard, Costa, and Tests Joiret 2010). The sculptures at Queluz were treated in Visual analysis: The sculptures were observed to under- 2012 and have since developed the dark red corrosion stand the patterns in which the stains occur (when layer again, indicating that the process is still active, present), how water runoff behaves, and the patterns of possibly as a cyclic or continuous mechanism. sunlight exposure. Extensive research has been done regarding the behavior Drip experiment: In order to confirm water runoff of lead in water-supply systems, some including the patterns on the sculptures’ surface, drops of deionized formation of lead dioxides. In recent years, it has been water were released using a pipette and observed as they proposed that lead dioxide forms from the oxidation of lead carbonate hydrocerussite (Bernard, Costa, and flowed, pooled, and evaporated from the surface. Joiret 2010); however, opposing research suggests “. . . XRF analysis: XRF analysis of the KMM’s L’Ai r was

PbO2 formation does not require lead(II) carbonates as performed in 2016 by Arie Pappot (Rijksmuseum precursors or intermediate phases” (Wang et al. 2010). Amsterdam).4 XRF measurements of MoMA’s La Rivière Other influential factors in the formation of lead dioxide were taken by Joy Bloser.5 Both collected spectra were 3 include the effect of dissolved inorganic carbon (DIC) , analyzed by Arie Pappot with the same software cali- pH fluctuations, and free chlorine concentration (Wang brated with lead standards. Quantitative analysis was et al. 2010). not performed on the MoMA’s La Rivière.

Environmental factors Sampling and microscopy: L’Air had seven samples taken on site according to accessibility and appearance, Since an oxidative environment must be present for embedded in a polyester casting resin, and polished. formation of the red corrosion layer to occur, it is worth Polyester was chosen as an embedding material due to comparing the outdoor climates of the sculptures with its minimal shrinkage upon drying, minimizing stress on and without staining. Queluz, Otterlo, and New York the soft metal samples. The cross sections were observed City are similar in some respects and quite distinct and photographed under a Leica DM2500 M optical in others. New York City has the most extreme high microscope with ZEN Lite 2.3 software. La Rivière had and low temperatures of the three cities, the highest three samples taken from non-visible locations exhibiting humidity levels, regular snowfall, and the highest drip patterns. Samples were taken at the edge of drip average rainfall. Queluz has the warmest climate and formations to include both the black and white crust 1 lowest average rainfall of the three cities. Air quality in formation on the lead substrate. Samples were embedded New York City has been historically poor, but has been in a polyester resin and polished to match protocols improved to levels roughly equivalent with Otterlo and used for analysis of the KMM’s L’Ai r. Cross sections were Queluz for the past two decades.2 Although specific imaged using a Leica DM IRM research microscope with particulate matter was not investigated in this paper as Phase One H20 and Capture One software. a possible cause of corrosion, it may shed more light on the differences in corrosion behavior between these SEM-EDX: This technique allowed the investigation three locations. of the morphology and elemental composition of the cross sections, and made it possible to assess possible Local microclimates were also noted. Neither MoMA stratigraphies. L’Ai r samples were analyzed by Ineke nor Queluz cover their sculptures during the winter Joosten (RCE).6 La Rivière embedded cross sections months, leaving them fully exposed to the weather all were analyzed by Joy Bloser (MoMA), courtesy of the year round. The KMM builds an open air shed around Conservation Center at the Institute of Fine Arts, New its sculptures to shield the work from November to York University.7 April. While it is difficult to attribute lead dioxide formation to a single environmental scenario, it is Test results interesting to consider why the dark red stains do not appear to form on the sculpture left uncovered in New Visual analysis and dripping experiment: Observations of York City. Water dwell time on the sculptures may be L’Ai r confirmed the red stains follow the pattern in which an important factor. water falls on the sculpture. However, they increasingly A DARK DISCOLORATION ON ARISTIDE MAILLOL’S OUTDOOR LEAD SCULPTURES: IDENTIFICATION, FORMATION, AND FURTHER RESEARCH 5 develop in the areas where the water runs, not where the water pools. White spots (identified in 2016 as lead carbonates) are only visible within the red stains (lead dioxides) (Joosten and van Hoesel 2016), and not in the gray surface (lead oxides). It was also observed that, given its outdoor location, the sculpture receives sunlight for long periods of time at different angles throughout the year. The drip experiment also showed that the white spots concentrate in the areas where the water pools. For La Rivière, drips of water flowed down the white channels of the drip striations on the undercut surfaces of the sculpture. The sculpture receives direct sunlight primarily only at midday due to its location in the garden and shadows cast by the surrounding buildings. It does receive reflected light from the surrounding building surfaces and reflecting pool below it. Figure 6. Diagram of microscopic, SEM, and EDX analysis of L’Air XRF: Qualitative analysis confirmed the two sculptures (sample 2), 2018 are both cast from lead alloys containing lead (Pb), anti- mony (Sb), and tin (Sn). This is a common casting alloy. Comparison of percent composition between the two sculptures was not possible given experimental limita- tions. As the two objects were cast by different founders at different times, it is expected their compositions are not identical. The 2016 measurements of L’Ai r were compared with lead standards, and quantitative analysis concluded the lead alloy composition contained 4–6% of antimony (Pappot 2016).

Sampling and microscopy: During the sampling process for L’Ai r , the dark red layer appeared to be harder and more brittle than the underlying lead alloy. Microscopic analysis did not allow discerning the morphologies of the corrosion products, although differences in color within the corrosion layers were evident. The red and white products were randomly mixed in a thin, well-adhered layer at the surface (Figure 6). Sampling for La Rivière revealed the white and black drip channels to be slightly more brittle than the lead substrate. Observed under Figure 7. Diagram of microscopic, SEM, and EDX analysis of La Rivière magnification, the white and black corrosion products (sample A), 2018 appear as thin, well-adhered crusts (Figure 7).

SEM-EDX: L’Air sample analysis showed that there was to observe antimony-rich areas in the bare metal which no evident difference in microstructure between the are not linked to the oxide layers. Important to note is red (lead dioxides), white (lead carbonates), and dark that the sulfur and lead peaks overlap at nearly 2,340 eV8, gray (lead oxide) areas. However, it was possible to making it very difficult to determine whether sulfur is distinguish the oxide layer from the bare metal. The present. Overall, distinguishing the red and white areas elemental analysis confirmed that there was an increase was extremely difficult, both chemically and visually. of carbon and oxygen in the corroded layers, and no trace The main difference in morphology was that the red elements were present at significant levels. It was possible areas looked more fractured than the white. Whisker-like 6 ICOM-CC | METAL 2019 | NEUCHÂTEL, SWITZERLAND CORROSION STUDIES formations on the samples were first noticed seven days surface of L’Ai r . At which speed these processes may after the polishing date, suggesting that they may have be occurring, or whether one is predominant over the developed during that time. While a few whiskers are other, is still under investigation. found on the bare metal, most are highly concentrated A surprising find from L’Ai r ’s analysis was the obser- near the surface, at the interface between the metal and vation that white powdery spots appeared first on the the corrosion layer (Figure 6). La Rivière sample anal- fresh metal of the sculpture after the samples were ysis confirmed XRF findings of an antimony-lead alloy. taken (Figure 8). Based on the previous documentation Elemental mapping showed discrete particles of antimony of L’Ai r , it was not possible to judge if lead carbonates surrounded by a lead-rich matrix. No discernible differ- appeared on the artwork before the red layer formed. ence between the metal surface and the oxide layer was Whether the white spots of lead carbonate will degrade detected using SEM-EDX (Figure 7). into lead dioxide can only be confirmed in the future. Discussion The current condition of La Rivière provides a good baseline for the future observation of lead carbonate The drip-like pattern of the oxidation layers confirmed and lead dioxide development on lead-antimony alloys that both sculptures are reacting to the runoff of precip- displayed outdoors. itation, one more drastically than the other. In the case of L’Ai r , the difference in color between the places where water runs and those where it sits indicates that lead reacts differently in flowing water versus stagnant water conditions. This observation coincides with research suggesting flow destabilizes the corrosion scales of lead pipes mechanically or chemically within a few hours (Xie 2010). Mechanical destabilization of the surface due to flow could also be linked to the formation of whiskers observed in L’Ai r ’ s samples, allowing corrosion to develop further in a continuous manner by creating cracks and exposing fresh surfaces. However, whisker formation has not been observed on the object itself, and it may not be possible given their scale. Figure 8. Sample #3 location a month after sampling showing white corrosion (Dino-Lite digital microscope at approx. 50×), 2018 Visual examination and SEM-EDX analysis suggest a layer of white lead carbonates is present in the areas XRF analysis did not single out any substantial differ- of water runoff on the MoMA’s La Rivière, similar in ences that could explain the formation of lead dioxide. appearance to L’Ai r , although no red corrosion layer or However, additional research into the percent compo- staining is visible on the sculpture. As La Rivière did sition of trace elements could provide more detailed not show clear traces of lead dioxide, XRD analysis information. Past waxing treatments in both sculptures, was not deemed essential at this stage of the research. more recently at the KMM, may have influenced the Raman analysis of cross-section samples is scheduled development of lead dioxide in L’Ai r ; nevertheless, this in order to understand any surface changes better. The does not explain why that was not the case in La Rivière. assorted red and white corrosion layers on L’Ai r suggest Evaluation of the treatment history of other examples that the corrosion build-up is not layered, but rather would be beneficial. a well-combined mixture of products. Bernard, Costa, and Joiret (2010, 341–44) proposed the red color may So far, the results seem to circle back to environmental be an oxidation product of hydrocerussite, while Wang conditions and water chemistry. Freeze-thaw cycles, et al. (2010, 8950–56) concluded both lead carbonates snow, and winter exposure may stunt the formation and lead oxides can be a precursor of the formation of of lead dioxide according to the weather patterns lead dioxide. It can therefore be suggested that more observed in New York. Whether water temperature than one mechanism is likely to be taking place at the also has a role requires further research. While pH A DARK DISCOLORATION ON ARISTIDE MAILLOL’S OUTDOOR LEAD SCULPTURES: IDENTIFICATION, FORMATION, AND FURTHER RESEARCH 7 is certainly a part of lead dioxide formation in lead measurement protocol (at 40 kV and 10 kV) and treated water-supply pipes, pH experiments on the surface of with PyMca software. objects remain problematic in the field of conservation. 5 XRF analysis was performed using a handheld Bruker Electrochemical research could also shed more light Tracer III-SD XRF spectrometer at 40 KeV, 1.5 μAmps, on the subject. and 75 seconds with P1 XRF software; files were exported as .csv. Analysis was performed on six locations. Conclusion and suggestions for further research 6 Analysis was performed using SEM (NovaNano 450 FEG from FEI) combined with EDX (Thermo Fisher In summary, the results show that the surfaces of L’Ai r Scientific UltraDry silicon drift detector), and analyzed and La Rivière are reacting in various ways to oxygen and with Pathfinder software. water in their environment. XRF and SEM-EDX analysis 7 did not point out substantial differences between the Analysis was carried out using a Hitachi TM3000 objects. Their treatment histories and environmental scanning electron microscope; electron dispersive spec- conditions suggest that the application of a protective troscopy was performed with a Bruker XFlash MIN SVE coating such as wax may prevent the formation of some using Quantax 70 software. corrosion layers by limiting the sculptures’ contact with 8 X-Ray Data Booklet Table 1–3. http://xdb.lbl.gov/ water. Further research into the removal of the corrosion Section1/Table_1-3.pdf . when desired is also suggested, including mechanical, chemical, and electrochemical methods. References Bernard, M.-C., V. Costa, and S. Joiret. 2010. On unex- Acknowledgments pected colour of lead sculptures in Queluz: Degradation The authors would like to thank Arie Pappot (Rijksmu- of lead white. Corrosion Engineering, Science and Tech- seum Amsterdam), Susanne Kensche (KMM), Virginia nology 45(5): 341–44. Costa (Institut National du Patrimoine, ), Luc Joosten, I. and A. van Hoesel. 2016. L’Air, Aristide Megens (RCE), René Peschar (University of Amsterdam), Maillol, 1930, Kröller Müller Museum. Onderzoek Roger Griffith, Ellen Moody, and Megan Randall (MoMA) naar corrosieproducten. Rijksdienst voor het Cultureel for their advice and support. Erfgoed, Amsterdam. Notes Lebon, É. 2014. Dictionary of art bronze founders: France 1890–1950: English translation. London: Sladmore 1 “Comparison of the Average Weather in New York Editions; Perth, Australia: Marjon Editions. City, Queluz, and Otterlo.” Weather Spark. https:// weatherspark.com/compare/y/23912~32002~52680/ Mahu, I. 2016. KM 127.576 Maillol – L’Air. Conser- Comparison-of-the-Average-Weather-in-New-York- vation report. Kröller-Müller Museum, Otterlo, The City-Queluz-and-Otterlo (accessed 28 October 2018). Netherlands. 2 “New York City trends in air pollution and its health Pappot, A. 2016. Corrosion phenomenon on L’Air by consequences.” NYC Health, September 26, 2013. https:// Aristide Maillol (KM 127.576). Conservation report. www1.nyc.gov/assets/doh/downloads/pdf/environ- Rijksmuseum, Amsterdam. mental/air-quality-report-2013.pdf (accessed 20 October Wang, Y., Y. Xie., W. Li., Z. Wang, and D. Giammar. 2010. 2018). Formation of lead(IV) oxides from lead(II) compounds. 3 Dissolved inorganic carbon (DIC): the sum of inor- Environmental Science & Technology 44(23): 8950–56. ganic carbon species in a solution; these include carbon Xie, Y. 2010. Dissolution, formation, and transformation dioxide, carbonic acid, and carbonates. https://en.wiki- of the lead corrosion product PbO2: Rates and mecha- pedia.org/wiki/Total_inorganic_carbon (accessed 06 nisms of reactions that control lead release in drinking October 2018). water distribution systems. Ph.D. dissertation, School of 4 XRF analysis was performed on eight spots using a Engineering & Applied Science, Washington University in handheld Olympus Delta X professional with a two-beam St. Louis, USA. http://openscholarship.wustl.edu/etd/38 . 8 ICOM-CC | METAL 2019 | NEUCHÂTEL, SWITZERLAND CORROSION STUDIES

Authors

Manuela Toro is a post-master’s metal conserva- tor-in-training at the University of Amsterdam. Her MA thesis focuses on the theme of this paper.

Joy Bloser is the David Booth Fellow in Sculpture Conservation at the Museum of Modern Art, New York (2018–2020).

Lynda Zycherman is the Conservator of Sculpture at the Museum of Modern Art, New York.

Ineke Joosten is a senior researcher in conservation and restoration at the Cultural Heritage Laboratory of the Cultural Heritage Agency of the Netherlands.

Tonny Beentjes is program leader in metal conservation and researcher at the University of Amsterdam.