White paints
Careful inspection of Calder’s Half-Circle revealed that different white paints are present on the object’s surface. For instance, an uncharacteristically thick, high-gloss, multi-layer paint was applied on the entire exterior of the base with perpendicular brush strokes. Rounding all the sharp edges of the support and hiding the texture of the wood, this paint displays heavy drips and multiple times overpainted lacunae throughout the surface. On the other hand, thin, brush-applied white layers were visually identified as possible original paint for the base exterior, as was a different kind of paint used for the shelf and woodblock holding the pulley in the interior of the base. Slight shade variations in Calder’s whites are not surprising, as the artist was known to often repurpose various bits of materials, which may have been already painted. Both the shelf and woodblock for the pulley do not seem to be integral part of the wooden case: they appear to have been machine cut and manufactured for other purposes, then chopped up to parts, and incorporated into the sculpture.
Two single-layered samples, S7 and S8, of what was visually recognized as likely original white paint were removed from the interior of the artwork’s wooden base on the proper right and left sides, respectively, for scientific analysis. According to Raman and FTIR data, both samples contain mixtures of zinc white, barite and/or lithopone, calcite, and gypsum, which are accountable for the white color observed (Figure 2). Analysis of the binding media with Py-GC/MS primarily found glycerol and a series of fatty acids, the evaluation of whose ratios represents a common method to differentiate among various oil media as well as describe modern adulterated oils and alkyds. The most significant fatty acids found via the transesterification of triglycerides and esterification of free fatty acids are palmitic (P) and stearic (S) acids, as well as the diacids azelaic (A), suberic (Sub), and sebacic (Seb) acids. The ratio of A/P reflects whether the oil analyzed might constitute a drying, semi-drying, or non-drying oil, while P/S provides information on the exact identity of the drying oil [8]. While traditional drying oils (e.g. linseed, walnut, poppy seed) have a fairly strict range of ratios, oils that have been adulterated with heavy metal soaps or other agents in order to achieve certain physical or mechanical properties can display widely ranging ratios. In samples S7 and S8, the fatty acid ratios (P/S = 1.3 and 2.2, A/P = 0.86 and 0.89, respectively) are consistent with the presence of a traditional drying oil binder, possibly linseed oil (Figure 3). These results clearly indicate, for samples S7 and S8, the use of a relatively limited array of pigments, extenders, and binding media that were all available in the early 1930s, when the artwork was created. Based on this observation, a hypothesis may be put forward that such a white layer could in fact represent an original paint applied by Calder to the piece’s wooden base.
On the other hand, two additional samples of single-layered white paint, S9 and S10, taken from the internal shelf and woodblock in the artwork’s mechanism, were found to be composed of lead white and calcite (S9), and of barite and/or lithopone, calcite, gypsum, and quartz (S10). With fatty acid ratios calculated as P/S = 1.4 and A/P = 1.6, results for white paint sample S10 show the presence of a drying oil modified by the addition of driers. All of the materials detected in samples S9 and S10 were also well known and widely used in the first few decades of the 20th century, which, again, may support authenticity of the paint layers examined.
In general, the use of different pigments in the white paints - zinc and lead whites, among others - may be explained by taking into consideration Calder’s inclination towards a spontaneous use of materials and odd bits of wood at hand to create his works [1]. It is worth noting that, in three of the white paint samples mentioned above, the detection of trace amounts of abietic acid derivatives in the binding media pointed to the presence of a diterpenoid resin belonging to the Pinaceae family, which may be interpreted as either studio debris or as a deliberate addition to the oil to modify its properties. Additionally, zinc and other metal carboxylates, as well as calcium oxalates, were identified in many of the white paints examined. As often reported in the literature, the metal carboxylates might have been intentionally added to the oil binder to adjust properties such as pigment suspension, gelation, lubrication, and plasticizing [9-11] and/or originate from the chemical reaction between the oil and certain pigments in the paint layer. The oxalates, on the other hand, may result from the degradation of organic materials and their reaction with calcium-containing pigments and/or particulate dirt [12].
Examination of cross sections with optical microscopy and scientific analysis by means of SEM/EDS, Raman, and ATR-FTIR provided crucial information on the stratigraphy of Calder’s painted surfaces in Half-Circle. In particular, cross sections S1, S4, and S5 - all removed from white areas of the exterior of the base likely to have been repainted - display an overall similar, incredibly complex stratigraphy that comprises up to eleven distinct paint layers. Close inspection of the polarized light and ultraviolet (UV) light microphotographs, along with BSE images, revealed that these paint layers are characterized by an array of different colors (ranging from white and cream tones to various gray shades), UV-induced auto-fluorescence emissions, as well as pigment particle size and morphology, which are reflected by different elemental and chemical compositions (Figure 4). Materials detected include a wide array of white pigments and additives, including titanium white, zinc white, barite and/or lithopone, calcite, gypsum, bassanite and/or anhydrite, dolomite, magnesite, silica, alumina, magnesia, as well as talc and various other silicates. In addition, a few iron(III) hydroxide inclusions as well as bone or ivory black particles were identified in most of the gray paint layers. In all three cross sections, titanium white is present in the form of tetragonal rutile, one of the three polymorphs of titanium dioxide along with tetragonal anatase and orthorhombic brookite. Interestingly, in cross sections S4 and S5, analysis with Raman spectroscopy detected a characteristic luminescence emission pattern that has been recently attributed to Nd3+ ions substituting into the orthorhombic alkaline earth sulfates of composite titanium dioxide pigments produced by co-precipitation with BaSO4 or CaSO4 [13]. In addition to delivering fundamental details related to pigment formulation and methods of manufacture, this observation bears interesting implications for dating. Indeed, although experimental rutile pigments were patented in 1931 in Czechoslovakia, Germany, and the United States, the first commercially viable methods for production were developed in 1937 and industrial manufacture began in 1938-39. Moreover, the anatase and calcium sulfate composite, introduced in the United States in 1925, was phased out in the early 1940s after rutile and calcium sulfate composites with increased hiding power and chalk resistance were introduced [14]. In cross sections S4 and S5, a rutile co-precipitated pigment was detected in most paint layers, including the bottom layers (layers 1 and 2) (Figure 5); on the other hand, the lowermost layer (layer 1) of cross section S1 was found to be similar in morphology and chemical composition to layers 3/4 in S4 and layer 3 in S5, indicating that the stratigraphy in sample S1 is likely incomplete. Based on these observations, it can be hypothesized that all paint layers in cross sections S1, S4, and S5 were applied within later repainting campaigns, possibly to cover lacunae, which is in accordance with the initial visual assessment of the corresponding sampling sites.
Binding media analysis with Py-GC/MS provided additional insight into the possible dates of application of the paint layers found in these cross sections. Samples S2 and S3, corresponding to cross sections S1 and S4, respectively, contain a drying oil modified by a possible addition of a non-drying oil or metal palmitates (P/S = 1.3 and 1, A/P = 0.45 and 0.65, respectively), alongside small amounts of a diterpenoid resin belonging to the Pinaceae family. In addition, the relatively high amounts of acetic acid and benzene liberated in the analysis of both samples are indicative of the presence of a polyvinyl acetate (PVAc) binder in one or more of the paint layers. Significant quantities of phthalates and phthalic anhydride were also identified in these two samples’ chromatograms. The detection of the latter compounds could be either related to the oil and resin, thus indicating the use of an alkyd paint, i.e. late formulation enamels based on ortho-phthalic acid, or assigned to an early formulation of PVAc with phthalates used as external plasticizers [15]. As an additional alternative, it cannot be ruled out, based on the data collected, that the phthalates identified might bear a twofold attribution and be present in both components. Py-GC/MS analysis also found, in samples S2 and S3, relatively high amounts of styrene, which could point to the presence of styrene-modified alkyds or styrenated oils (Figure 6) [15, 16]. On the other hand, the exact source of the trace levels of acrylic compounds detected in S3 remains unclear. Sample S6, a scraping of white paint corresponding to cross section S5, was found to contain a small oil component and Pinaceae resin, which may suggest the presence of an early enamel paint [11, 17]. Due to the low intensity of the fatty acid peaks in the chromatograms, their ratios, in this case, cannot be used to draw conclusions on the type of oil present. As observed for samples S2 and S3, PVAc was also identified here, as well as phthalates and styrene, whose presence, as discussed above, may be associated with one or more of the binding media found in this sample. Furthermore, traces of acrylic compounds appear to be present in one or more of the paint layers. One final connection between S2, S3, and S6 is a tightly packed group of unique compounds detected between 9.1 and 10 min in the TMAH chromatograms - predominantly branched fatty acids - which would appear to support the premise that these samples represent a particular paint formulation that is different from others analyzed on the sculpture.
While enamels were introduced in the late 19th century, all other binders mentioned above became commercially available in America at different times during the first half of the 20th century. The first oil-modified alkyd resins were produced in 1927, when DuPont began its research into decorative paints based on alkyds. However, it is not until shortly before the Second World War that alkyd-based paints were introduced into the American market, where they made a significant impact starting from the late 1950s. Generally, alkyd resins received only limited attention by manufacturers of artists’ paints, with the notable exception of Winsor & Newton’s Griffin paints in 1970, and it is mainly in the form of house paint that they are found in works of art. In an attempt to overcome their susceptibility to hydrolysis, alkyds started to be modified with a variety of molecules, including styrene and acrylics, in the 1940s. Undocumented for use in decorative paints, styrenated alkyds were reportedly employed for low-cost fast-drying enamels, baking coats, and undercoats. Similarly, research on styrenated oils came to fruition in the late years of the same decade. First introduced in the 1930s, PVAc became available as a water-borne emulsion in the late 1940s and was initially available upon modification with external plasticizers, such as dibutyl phthalate or tricresyl phosphate, which, however, caused paint films to become weak and brittle. PVAc-based paints did not gain popularity in the United States until the mid-1950s, when the vinyl acetate monomer became financially competitive and resins were internally plasticized by copolymerizing vinyl acetate with a softer monomer. By that time, PVAc films were also being modified with alkyds in an attempt to improve their scrub resistance, freeze-thaw stability, and pigment-wetting properties. As with the alkyds, PVAc has had fairly limited use in the artists’ paint market; the first artists’ PVAc emulsion paint, reportedly developed by Burden Co. in 1945, never achieved success among artists due to issues with the effective distribution of pigments in the medium. Acrylics solutions and emulsions, on the other hand, were first introduced in the late 1940s and mid-1950s, respectively, and have been widely available in the form of artists’ paints since their first introduction [15, 16, 18, 19]. Based on this historical information and on the pigments’ results previously discussed, and given that the presence of PVAc in the bottom layer (layer 1) of cross section S4 was unambiguously confirmed by ATR-FTIR, the earliest date of application of the numerous paint layers observed in cross section S4 may be placed in the late 1940s. Application of the white paint layers found in the three cross sections examined within later repainting campaigns (post 1940) is also supported by the existing scholarship on Calder, as the use of both titanium white and PVAc-based media appears to be uncharacteristic for this artist [5-7].
Red paints
Wrinkled, thickly applied red paint with glossy appearance and multiple overpainting on large losses suggested that the sphere’s red color is not authentic. Though Calder used different shades of bright orange-red in his works, the red color of Half-Circle appears somewhat deep, and certainly does not correspond to the thin brushstrokes seen on the artist’s original reds. This paint showed a tendency to gum up and crumble, and does not reveal the simple construction of the intersected sheet metal circles and their mechanical joins as observed in original surfaces.
Cross sections S11 and S12, removed from the sphere’s red flange and circle, display a nearly identical stratigraphy that includes seven paint layers consisting of an assortment of synthetic organic red pigments, a lead chromate-based pigment that is present as two distinctively orange layers located in the top portion of both cross sections, a cadmium sulfoselenide-based pigment in between them, as well as a variety of extenders, such as barite, gypsum, calcite, dolomite, silicates, silica, and alumina (Figure 7). Four of the synthetic organic reds identified, i.e. PR1, PR3, PR4, and PR49, belong to the molecular class of β-naphthols, which are based on the coupling of a substituted aniline ring with β-naphthol and are among the oldest of synthetic organic pigments. While PR1 was the only coloring material detected in layer 2 of both cross sections, binary combinations of the other synthetic organic reds, sometimes along with inorganic pigments, appear to be predominant in all other paint layers, as follows: PR4 and PR49 in layers 1 and 3; a mixture of PR3 and PR4 with a lead chromate-based pigment in layers 5 and 7 (Figure 8); and a cadmium sulfoselenide-based pigment, alongside traces of lead chromate, in layer 6 (Figure 9). Interestingly, layer 4 was found to contain PR83 - the synthetic counterpart of natural dye alizarin (1,2-dihydroxyanthraquinone) - precipitated onto an alumina substrate (Figures 8 and 9); accordingly, such layer in both S11 and S12 shows a translucent appearance that is typically associated with lake pigments. Among the β-naphthols, PR49, discovered in the late 19th century by Austrian chemist Paul Julius and also known as Lithol Red, was the first to be manufactured as a pigment lake and was widely available by the early 1920s; originally precipitated onto an inorganic carrier material such as barium sulfate, it was later used in its pure form [20]. In the present case, X-ray elemental maps collected by SEM/EDS display an abundance of calcium (Ca) throughout layers 1 and 3 of both samples S11 and S12, suggesting that the PR49 pigment used here might belong to the Ca type (PR49:2), the most commonly found along with the Ba type (PR49:1) since the beginning of the manufacturing process (Figure 9). Unfortunately, this hypothesis could not be confirmed by Raman spectroscopy, as spectra of the various salts (Na, Ca, and Ba) of Lithol Red display nearly identical patterns, in which even band shifts in the region around 1200 cm-1 [21] fall within the calibration range and, thus, cannot be used to differentiate among the three lake pigments. Albeit commonly employed in the production of low-cost printing inks for its brightness and bleed resistance, Lithol Red is extremely fugitive and its poor lightfastness has been identified as the cause of the dramatic color changes observed in artworks from the 1950s and 1960s, most notably paintings by Clyfford Still in The Met’s collection [22] and a mural cycle by Mark Rothko, known today as his Harvard Murals [21]. The other three β-naphthol reds identified in Calder’s cross sections are also generally prone to fading when exposed to light, and some of them were originally developed for applications that did not only include artists’ pigments. For instance, PR1, or Para Red, first synthesized in 1880 by Holliday in England, is reported to have poor lightfastness and is not used in artists’ paints because of bleeding; it was initially employed industrially in metal finishes and printing inks, but has been since replaced with more durable colors. PR3, also known as Toluidine Red, was first manufactured in 1904 by Lucius and Brüning in Germany, but reached its peak of popular use in the 1970s; also characterized by reduced resistance to light and weather, it is primarily employed for industrial coatings, in addition to wax crayons, pastels, and watercolors. PR4, or chlorinated Para Red, was first produced in the early 1900s along with its isomer PR6 (Parachlor Red). Despite its poor lightfastness and sensitivity to solvents, PR4 has been used extensively in artists’ materials, including paints, colored pencils, and wax crayons [23, 24]. While the presence of additives and binders, among other factors, may affect the physicochemical properties of a paint, the overall unsatisfactory lightfastness of several of these pigments may partly justify the recurrent repainting observed on red areas of Calder’s Half-Circle.
Binding medium analysis for samples S11 and S12 (all red paint layers combined) indicates another possible example of an early enamel, consisting of a mixture of oil, a diterpenoid resin from the Pinaceae family, and the polyhydric alcohol pentaerythritol (Figure 10), a distinctive cross-linking agent used in place of or sometimes in mixture with glycerol from the mid-1940s [16]. While the fatty acid ratios (P/S = 0.85 and 0.99, A/P = 1.9 for both) suggest a drying oil, the presence of both marker compounds arachidic and behenic acids points to the addition of sunflower or safflower oil [25, 26], used in paints since 1949 [16]. However, in the case of modern oils, the addition of specific fatty acids to modify the oil’s properties cannot be ruled out. Due to the inherent limitations of bulk analysis with Py-GC/MS in case of samples displaying a complex stratigraphy, it is difficult to draw conclusions on the exact dates of application of the numerous paint layers observed in these two cross sections. Indeed, while some of the pigments identified in S11 and S12 reached their peak of distribution in mid-20th century or later, they were all available in the early 1930s when Half-Circle was created. ATR-FTIR analysis of the stratigraphy of these samples did not detect binding media other than oil and resin in the bottom layer (layer 1); yet, based on the data collected, it is not possible to trace the precise distribution of pentaerythritol and sunflower / safflower oil – the only two components that may provide clues as to dating - within the layer structure. If one assumes the latter to be located in the upper portion of the cross sections, it is conceivable to expect for the first layers as being original, since enamel commercial paints were previously documented as Calder’s paints [5-7]. On the other hand, it is likely that the paint layer(s) containing pentaerythritol and sunflower / safflower oil were applied as part of a later intervention.
Black paint
The black paint on Half-Circle’s rod element is matte and appears to have been applied as a thin single layer directly on the metal; however, the lack of typical brush marks and its nearly pristine condition suggested that, while visually comparable to an original paint from the 1930s, it is likely a later repaint performed in an educated manner. With virtually no losses, wear, flaking or any signs of even partially failing adhesion of the paint to the steel support, such black paint likely constitutes a complete repaint carried out in the distant rather than recent past.
Black paint sample S13, visually thought to be a repainting layer, contains a carbon-based pigment and/or bone/ivory black with calcite and gypsum dispersed in an early alkyd binder, based on the fatty acid ratios (P/S = 0.77, A/P = 1.1) as well as the presence of phthalic anhydride, bis(2-ethylhexyl) phthalate, pentaerythritol, and Pinaceae resin. While the fatty acid ratios are consistent with a drying oil, in this case, too, the presence of arachidic and behenic acid suggests that sunflower or safflower oil may have been added. Again, it should be noted that, in the case of modern oils, the addition of specific fatty acids to modify the oil’s properties cannot be ruled out. Moreover, the detection of low amounts of elaidic, aleuritic, and butolic acids appear to indicate the use of shellac [27]. While the date of introduction of alkyd paints in America [15, 18, 19] suggests that this paint layer may have been applied in or after the late 1930s, the presence of both glycerol and pentaerythritol as polyhydric alcohols pushes the earliest possible application date back to the second half of the 1940s [16].