Daniela R. Leonard, An Exercise in Fugitivity: Investigating the fading of red lake pigments, e-conservation Journal 5, 2017
Available online 15 March 2017
An Exercise in Fugitivity: Investigating
the fading of red lake pigments
Daniela R. Leonard
In 2009 at the Hamilton Kerr Institute, University of Cambridge, the author undertook a third year student project to study the appearance, handling and fading of red lakes in combination with a variety of traditional artists’ pigments. Four red lakes were made following 16th century recipes, and simplified paint systems were devised based on examples found in 14th-17th century European paintings. The preparation of paint samples highlighted the difficulties involved in comparing pigments with vastly different physical characteristics. Three sets of test panels were painted, one to be artificially aged in a light box, one to be naturally aged, and one as a control. Color measurements were taken of each sample in order to quantitatively compare the paints before and after fading. The project ran for 14 months, after which time only the artificially aged samples exhibited significant color changes. The results generally followed expectations: the red lakes in combination with white pigments tended to fade the most, followed by those with blue, while those with red, green and black pigments had minor to no change; the relative lightfastness of the different types of lakes tested depended in part on the paint system in which they were used; finally, the lakes applied as glaze layers performed better than those in mixtures, and thicker layers faded less than thinner layers. Although there were many uncontrolled variables within this project, useful information was gained from hands-on experimentation with traditional artist materials..
Traditional red lake pigments were highly valued for the beautiful glazes they produce, and for this reason were often applied by artists in the upper layers of paintings, despite their sensitivity to light. In part because of their fugitive nature, organic colorants have been the subject of a fair amount of study by conservators and scientists, particularly in connection with their use in textile dyes. Investigations into the fading of lake pigments have generally focused on layers or mixtures with white, but artists used pigments in countless permutations, and behavior with white is not necessarily representative of behavior with other colors. In 2009, the author conducted a student project at the Hamilton Kerr Institute, University of Cambridge (HKI) in order to learn more about the behavior of red lakes in a variety of paint combinations [1, 2]. The intention was to expand on past studies of red lakes by considering how a painter’s choice of materials and method of application affect both the immediate visual appearance of a work and subsequent aging.
2. Aspects of Fading
To make a lake pigment, an inorganic substrate is dyed with colorants extracted from raw dyestuffs derived from plant or insect sources. The physical structure and chemical degradation of these combined materials are not entirely understood, but certain factors are known to affect lightfastness. The durability of any paint depends first on the sensitivity of the pigments to the types of light to which they are exposed . The source of the dyestuff, not only the type of insect or plant but also the environmental conditions in which they are harvested, will determine the type and quantity of colorant that can be extracted . The extraction process itself, along with proper washing to remove impurities, will also play a role in the durability of the final color .
Historically, the quality of red lakes available to artists would have been determined primarily by the colormen that sold them, but a painter’s technique also affects how well a work holds up over time. The artist controls the combination of pigments, their relative concentration and distribution within the paint, the thickness of glaze layers, and some physical aspects like particle shape and size (influenced by the level of grinding). All of these elements can affect what fraction of incident light is scattered and what is absorbed. The total amount of light absorbed by a paint layer determines the rate of fading of a given pigment. Theoretically, red lakes will fade more slowly when combined with pigments other than white, based on total light absorption and scattering in the paint .
Past studies of light-sensitive pigments have demonstrated that the physical loss of organic colorant in paint generally follows first-order kinetic behavior, meaning that under constant light exposure a constant fraction of pigment will be lost, regardless of concentration . However, as has been shown both theoretically and empirically, the human eye is less sensitive to color loss in highly pigmented paints than in midtones, meaning that we may not detect fading in a more concentrated paint, even when the total physical loss of pigment is actually greater . In fact, when a glaze is very highly pigmented we will perceive an increase in chroma with exposure to light (in other words, the color becomes brighter), which means that we may not recognize certain detectible visual changes as ‘fading’ . Similarly, light damage to a red lake pigment that is layered or mixed together with white pigment may be easily identified as fading, whereas damage to the same lake combined with a red or green pigment may not.
In order to investigate the behavior of red lakes in paint systems that represent actual artistic practice, a variety of inorganic pigments were selected based on examples found in European paintings from the 14th-17th centuries. However, as the project progressed through the planning stages it became apparent that the idea of comparing the light sensitivity of different paints, while seemingly simple in conception, raises problems for experimental study. The first hurdle was the complexity of artists’ choice of pigments. The examples of red lakes used in paintings were found in combination with multiple pigments. Reconstructing these paints would not have allowed for a comparison of the individual effects of each inorganic pigment on fading. Therefore, the components were simplified to one or two red lakes applied as glaze layers over an inorganic pigment, and to one red lake mixed with one inorganic pigment in different ratios (1:3, 1:1 and 3:1 by weight). The red lakes tested were brazilwood, cochineal, lac lake and madder, all of which were made at the HKI following 16th century recipes. The inorganic pigments tested were azurite, lead white, smalt, ultramarine, verdigris, vermilion, and vine black. The medium used, also made at the HKI, was a linseed drying oil with a cobalt siccative.
Another consideration was pigment concentration. As has been noted, the visual perception of fading is dependent on the concentration of colorant, even while the rate of the physical loss of pigment is not. Consistency was maintained in the samples by measuring the weight of pigment and the volume of medium. There was found to be some variation in pigment volume and oil absorption between the different lakes. In particular, a mistake made with the recipe for madder resulted in a pigment that was noticeably lacking in colorant. At equal weights, the madder layers were significantly paler and also relatively medium rich when mixed in oil, while the cochineal and brazilwood layers were slightly underbound, resulting in a matte appearance in some of the samples.
Maintaining equal weights of inorganic pigment across the various mixtures tested was a greater challenge. Artists’ pigments are not only different in color, but also in a variety of other physical characteristics like particle size, oil absorption and tinting strength. With the exception of madder, the red lakes made for this project were fairly similar, but the difference between vine black and vermilion was significant. The mercury in vermilion is much heavier than the carbon in vine black; therefore, the equal weights used in comparative mixtures of these two pigments resulted in notably different volumes of paint. The total quantity of each mixture was applied to the same 2.5 cm2 sample area, meaning that the layer thickness of the mixtures was variable.
The thickness of a paint passage is not a factor for the fading of opaque layers, but it is for glazes, which have been found to fade from the top down . The relatively high concentration of lake pigments in the mixture samples, along with variations in the volume and covering power of the inorganic pigments, meant that many mixtures were semi-transparent. This was true of mixtures made with lead white, azurite, smalt, ultramarine, and to some extent vermilion, as well as almost all of the mixtures with madder. An effort was made to apply the paints evenly, but it was considered necessary to use a brush (rather than a drawdown) in order to better understand how the handling qualities of traditional materials affect the resulting paint layer. Differences in oil absorption meant that the vermilion and lead white mixtures had an overly runny texture while the vine black mixtures were quite stiff and matte. A matte surface poses problems for measuring color change; therefore, a thin coat of dammar varnish was applied to all of the samples to fully saturate the paints.
Three sample sets were made, one to be artificially aged in a light box, one to be hung on a wall at the Fitzwilliam Museum in Cambridge, and one to be kept in the dark as a control. The artificially aged samples were exposed to continuous light of approximately 5500 lux, while the museum samples were exposed to an average of about 140 lux during daylight hours. It was not possible to control the climate for the artificially aged or control samples. No attempt was made to measure the physical loss of colorant within the test samples, but visual color changes (ΔE) were calculated from color measurements taken with the use of a Minolta CR-221 colorimeter, which provides numerical values for colors based on the CIELAB color space. The full project ran for a total of 14 months and color measurements were taken at irregular intervals, as time allowed.
Left to right:
Figure 1: Samples of cochineal, lac lake, madder and brazilwood (C, L, M, B respectively), individually layered in horizontal and vertical stripes over a lead white underlayer. Control samples after 14 months. (Photo: Hamilton Kerr Institute, University of Cambridge)
Figure 2: Samples of cochineal, lac lake, madder and brazilwood (C, L, M, B respectively) individually layered in horizontal and vertical stripes over vine black, azurite, verdigris and vermilion underlayers (clockwise from top left). Control samples after 14 months. (Photo: Hamilton Kerr Institute, University of Cambridge)
Figure 3: Samples of cochineal, lac lake, madder and brazilwood (C, L, M, B respectively) individually mixed with lead white in ratios of 1:3 / 1:1 / 3:1 (from left to right). Control samples after 14 months. (Photo: Hamilton Kerr Institute, University of Cambridge)
Figure 4: Samples of cochineal, lac lake, madder and brazilwood (C, L, M, B respectively) individually mixed with ultramarine, smalt, verdigris and vermilion in ratios of 1:3 / 1:1 / 3:1 (clockwise from top left). Control samples after 14 months. (Photo: Hamilton Kerr Institute, University of Cambridge)
At the end of 14 months, no significant color changes were noted in either the control or the naturally aged samples (Figures 1-4). Slight lightening and color shifts noted in the control samples can likely be attributed to sinking of the varnish, though the lack of climate control may also have had a small impact. All of the artificially aged samples became lighter and less red, to varying degrees, over the course of the project (figures 5-8). The red lakes combined with lead white demonstrated some of the greatest color changes, followed by the lakes with smalt, ultramarine and azurite. The vermilion samples had smaller color shifts, while verdigris and vine black had minor or no changes. Overall, the extent of visual color change in the layers and mixtures was fairly comparable, even though the mixtures contained higher concentrations of lake pigments (figures 9-10).
Left to right:
Figure 5: Samples of cochineal, lac lake, madder and brazilwood (C, L, M, B respectively) individually layered in horizontal and vertical stripes over a lead white underlayer. Artificially aged samples after 14 months. (Photo: Hamilton Kerr Institute, University of Cambridge)
Figure 6: Samples of cochineal, lac lake, madder and brazilwood (C, L, M, B respectively) individually layered in horizontal and vertical stripes over vine black, azurite, verdigris and vermilion underlayers (clockwise from top left). Artificially aged samples after 14 months. (Photo: Hamilton Kerr Institute, University of Cambridge)
Figure 7: Samples of cochineal, lac lake, madder and brazilwood (C, L, M, B respectively) individually mixed with lead white in ratios of 1:3 / 1:1 / 3:1 (from left to right). Artificially aged samples after 14 months. (Photo: Hamilton Kerr Institute, University of Cambridge)
Figure 8: Samples of cochineal, lac lake, madder and brazilwood (C, L, M, B respectively) individually mixed with ultramarine, smalt, verdigris and vermilion in ratios of 1:3 / 1:1 / 3:1 (clockwise from top left). Artificially aged samples after 14 months. (Photo: Hamilton Kerr Institute, University of Cambridge).
Reflectance of the samples was not measured for this project, but the results of calculated color changes make intuitive sense in terms of expected scattering and absorption. As previously noted, many of the mixtures were semi-transparent and visually similar to the glaze layers, but mixtures have more potential to scatter light internally, which may result in additional fading. White pigments will scatter the most light, while blue pigments will absorb more, but reflect primarily higher energy blue wavelengths that can be more damaging to surrounding organic red pigments than green or red wavelengths. Black pigments will absorb all wavelengths, reducing damage to the lakes.
In terms of visible color shifts, the degree to which fading of a red component affects change also depends on the hue and relative tinting strengths of the pigments in the paint. When combined with white or blue pigments, the lac lake used in this project generally demonstrated the least color change and brazilwood the most, but in other pigment combinations the type of lake was less relevant. The difference between vermilion and the blue pigments, particularly ultramarine and smalt, was notable. Observing the samples at the end of the project, the lake pigments clearly faded to a large degree in all cases (figures 4 and 8), but in terms of calculated color changes as expressed by L*, a*, b* readings with a colorimeter, the loss of red in the vermilion samples was mitigated by the presence of the inorganic red pigment, while a loss of red in the blue samples was concurrent with an increase in blue. The color change calculations demonstrate that the different lakes mixed with vermilion behaved similarly, while changes in the blue samples were much more varied (figures 11-12). In a few cases, such as lac lake mixed with ultramarine at 1:3, dramatic shifts towards blue resulted in greater overall color changes than the combined loss of red and increased lightness in corresponding lead white samples (figure 10). It is important to keep in mind that total color change as expressed numerically by ΔE calculations is a simplified abstraction that does not provide us with a full understanding of what has occurred within the paint.
As expected, most of the mixtures with higher concentrations of lake pigments retained their color better than those at lower concentrations, and double layers faded less than single layers. The notable exceptions were the samples with madder lake, where the mixtures at 3:1 and the double layers registered slightly greater changes. These results confirm that the madder made for this project lacked colorant within the lake/substrate matrix, since paints with only traces of colorant will have less color to lose, and visually change less than paints with higher pigment concentrations .
Uncertainties remain due to uncontrolled variables in the setup. For example, although an attempt was made to grind the lakes evenly during their preparation, cross-sections taken at the end of the project revealed variations in particle size, particularly in the cochineal samples. This could be significant because small pigments fade more quickly than large ones . Another variable was the brush application of the paints. In order to compensate for uneven brushstrokes, five color measurements were taken within each sample square so that an average could be calculated. At the end of the project, the brushstrokes became noticeably more prominent in many samples (figures 5-8).
Left to right:
Figure 9: Calculated color changes (ΔE) in samples of lac lake applied as a single layer over inorganic pigments.
Figure 10: Calculated color changes (ΔE) in samples of lac lake mixed with inorganic pigments in a ratio of 1:3.
Figure 11: Calculated color changes (ΔE) in samples of red lakes mixed with vermilion.
Figure 12: Calculated color changes (ΔE) in samples of red lakes mixed with ultramarine.
Red lakes have long been known to fade, but the extent of visual change that occurs as a result of light damage has multiple factors, including many that are controlled by the artist. The origin of the dyestuff and elements of the extraction process determine physical durability, while the presence of other pigments in the paint affects the internal reflectance of light. Layer thickness of semi-transparent glazes and pigment concentration affect the visual perception of fading, along with the relative tinting strengths of red lakes in combination with other pigments. Differences between the four lakes tested were most detectible in samples containing white or blue pigments, in which lac lake tended to retain the most color and brazilwood the least. The largest color shifts were generally measured in the lead white samples, followed by ultramarine, smalt, and azurite, whereas combinations with vermilion, verdigris and vine black resulted in much smaller changes. A possible avenue for further research would be to investigate the prevalence of different lakes in real paintings. If artists were aware that any lake in combination with red, green or black was likely to provide a durable paint, perhaps they would elect to use lower quality, less expensive lakes for such passages, saving the higher quality lakes for combinations with white or blue pigments.
While the constraints defined by this project resulted in paints that were, in various ways, outside the norm of traditional artists’ use, the process through which the samples were made was nevertheless informative, not least for highlighting that a direct comparison of different pigments is not particularly meaningful if the results become too far removed from artists’ actual practice. The simple act of painting out the samples was a useful reminder that the physical characteristics of paint are intimately tied to technique. Scientific analysis can provide us with important information about the material components of a paint sample, but reconstructions of paints can help us to understand the meaning behind that information. Hands-on experience working with traditional artist materials can lead to a better understanding of artists’ decisions and how the physical and visual qualities of paints influence the ways in which they are applied. Once a painting has been built up of various mixtures and layers laid out by the artist, the susceptibility of the work to future light exposure has been determined. Further study of the behavior of red lakes in paint systems that are representative of their use in real paintings may provide a broader understanding of how artists’ choices influence the light-sensitivity of their works.
 D. R. Leonard, The Fading of Traditional Paints, Third Year Project, Hamilton Kerr Institute, 2009
 D. R. Leonard, Reconstructing red lakes in traditional paint systems, in: L. Wrapson, J. Rose, R. Miller, S. Bucklow (eds.), In Artists Footsteps: The Reconstruction of Pigments and Paintings, Archetype Publications Ltd, 2012, pp. 141-149
 P. Whitmore, C. Bailie, Further Studies on Transparent Glaze Fading: Chemical and Appearance Kinetics, Journal of the American Institute for Conservation 36(3), 1997, pp. 207-230, doi: 10.1179/019713697806124402
 J. Kirby, R. White, The Identification of Red Lake Pigment Dyestuffs and a Discussion of their Use, National Gallery Technical Bulletin, 17, 1996, pp. 56-80
 J. Kirby, M. Spring, C. Higgitt, The Technology of Red Lake Pigment Manufacture: Study of the Dyestuff Substrate, National Gallery Technical Bulletin, 26, 2005, pp. 71-87
 R. Johnston-Feller, R. L. Feller, C. W. Bailie, M. Curran, The Kinetics of Fading: Opaque Films Pigmented with Alizarin Lake and Titanium Dioxide, Journal of the American Institute for Conservation 23(2), 1984, pp. 114-129, doi: 10.2307/3179474
 R. Johnston-Feller, C. W. Bailie, An Analysis of the Optics of Paint Glazes: Fading, Studies in Conservation 27(1), 1982, pp. 180-185, doi: 10.1179/sic.1982.27.Supplement-1.180
 D. Saunders, J. Kirby, Light-induced Colour Changes in Red and Yellow Lake Pigments, National Gallery Technical Bulletin, 15, 1994, pp. 79-97
Daniela R. Leonard