Maggie Barkovic and Olympia Diamond presented a case study that outlined the decision-making process that lead to the successful treatment of darkened, dirt-infused water stains on the bare canvas portion of a large-scale acrylic dispersion painting: Composition, 1963, by Justin Knowles. The authors attributed the treatment’s success to the combination of extensive evaluation of Knowles’ materials and aesthetic aims and the understanding of new, innovative cleaning techniques designed for acrylic dispersion paintings (with the help of Brownyn Ormsby, TATE, and Maureen Cross, Courtauld Institute of Art). This presentation served an excellent compliment to Jay Kruger’s presentation Color Field Paintings and Sun-Bleaching: An approach for removing stains in unprimed canvas, which discussed the treatment of acrylic solution and oil paintings on bare canvas.
Composition is a privately-owned work that was brought to the Conservation and Technology Department at the Courtauld Institute of Art for treatment in 2013. The large-format work is a two-dimensional acrylic painting with brightly colored geometric forms juxtaposed against an unpigmented acrylic sized canvas. The painting had sustained disfiguring water stains along the top and bottom edges which disrupted the aesthetic reading of the image, rendering it unexhibitable.
Context
In the first step of the conservation process, Barkovic and Diamond assessed how the water stain affected the aesthetic interpretation of the painting. They explored where this painting fit into the artist’s oeuvre: it was part of a series of early, pivotal works where Knowles explored his initial ideas of spatial tension using non-illusionistic geometric compositions that incorporate negative space in the form of unpainted canvas. The authors carried out technical examinations of four other paintings from this early stage in his career, finding that Composition was painted in a comparable manner to his other early works: a fine linen canvas was stretched on a wooden stretcher and then sized with an unpigmented (pEA/MMA) acrylic dispersion coating. Then, Knowles used pencil and pressure-sensitive tape to demarcate where he would paint the geometric forms with acrylic dispersion paints. Though he applied a transparent acrylic “size” layer over the linen/negative space, he still considered the visible canvas “raw” and unprimed. Through the examinations and research on Justin Knowles’ personal notes, the authors assessed that the characteristics and color of the linen canvas were equally important to the interpretation of the work as the paint colors. As such, the canvas should be treated and the water stains removed if at all possible.
Replicas
Second, the authors explained that they needed to identify the components of the water stain (with no prior knowledge of water-staining incident) in order to test cleaning methods. Replicas were made using linen and the same unpigmented acrylic polymer that Knowles most likely used. The replicas were then stained with dirty water. Using XRF spectroscopy and empirical testing as a guide, a visually accurate and equally tenacious water stain was made with iron, calcium, and organic “dirt” components from aged linen. The test replicas were aged in a light box for two years to allow the stain to photo-oxidize and bond with the fabric and size layers.
Testing
Third, the authors needed to determine how to treat the water stain with the presence of the unpigmented acrylic dispersion size layer, which swelled in water and was affected using polar solvents. Their goal was to remove the stain or reduce the appearance of the stain to make successful inpainting possible. The authors looked to successful textile and paper conservation treatments for possible methods. The initial cleaning and/or retouching tests included the use of solutions with various pH values, conductivities, chelating agents, surfactants, bleach (sodium borohydride), the application of toasted cellulose powder, and pastel retouching.
The authors thoroughly explained the various test groups, but a recapitulation of all of these various solutions is outside of the scope of this blog post. In general, higher pH values (around 8) and higher conductivity values (above 2.5 uS) allowed for better cleaning efficacy. Perhaps more notably, the chelating agent DTPA (diethylene triamine pentaacetic acid) greatly outperformed TAC in cleaning efficacy. This is likely because DPTA is a much stronger chelator that is much more suitable for sequestering iron and calcium (which XRF showed to be present in the stain). DPTA could be used safely because the acrylic size layer was unpigmented. Finally, the use of agar (rather than free solution) was found to be useful in the reduction of the stain. The agar gel allowed for greater control of the solution distribution onto the stain and dirt absorption into the gel. The most effective cleaning agent, which was eventually used to clean the painting, was made from a higher concentration of agar gel at 5% (w/v), using Boric Acid 0.5% (w/v), DTPA 0.5% (w/v), TEA, at pH 8, 2.4 mS.
Evaluation of Successful Treatment
While a successful treatment methodology was developed through empirical testing, an investigation into the effects on the surface morphology of an unpigmented acrylic dispersion size layer was thought necessary due to the different absorbencies among the test canvases, observed differences in retention times for the agar gel, and concerns about the higher pH required to reduce the stain. The lack of pigmentation and hard surface features made changes caused by cleaning more difficult to perceive, measure and contextualize, so changes in surface gloss and stain reduction were evaluated with a spectrophotometer and subjective observations by conservators. The impact of the cleaning methodology on the surface of the size layer and canvas fibers were examined with dynamic Atomic Force Microscopy (AFM) and high resolution digital microscopy. A preliminary investigation into possible residues from cleaning was also investigated using FTIR-ATR spectroscopy.
The number of samples for AFM was too small to draw concrete conclusions without more testing and utilizing additional analysis such as FTIR-ATR; however, a general trend was observed that an increase in the gel concentration from 2.5% (w/v) to 5% (w/v) appeared to reduce the time in which fiber flattening occurred. In addition, FTIR-ATR showed a decrease or complete removal of migrated surfactant from the acrylic size layer surface in all treated samples regardless of the agar concentration in the gel, and along with the swelling of the acrylic layer, was considered by the authors an acceptable risk with this treatment. IR bands corresponding to agar or the additives in the cleaning solutions were not detected.
Final Treatment
As mentioned previously, the cleaning agent that was eventually used to clean the painting was made from a higher concentration of agar gel at 5% (w/v), using Boric Acid 0.5% (w/v), DTPA 0.5% (w/v), TEA, at pH 8, 2.4 mS. The agar was hand-cut to perfectly align with the stain patterns on the canvas and weighted with sandbags to increase the gel-canvas contact. Using this method, stains were greatly reduced. However, a few, minor discolorations remained after the cleaning. Further tests were carried out to determine the best inpainting method for these residual discolorations. Dry pigment with Lascaux Jun Funori, Aquazol 50, Aquazol 200, watercolour and gum arabic and Paraloid B72 were all tested for optical effects, handling properties, and reversibility. The Aquazol 50 series was found to be the most effective overall and was used to inpaint the remaining discolorations.
Conclusion
The authors concluded by restating that the success of the treatment would not have been possible without the combination of art historical and material understanding of Knowles’ work and research into new cleaning methodologies for acrylic dispersion paint films. They thanked their project advisors Maureen Cross, Courtauld Institute, and Bronwyn Ormsby, Tate, and many others for their generous support and guidance throughout the project.