Research: Care and Conservation

The foundation of the Recoloring Faded Fur project was its partnership with Yale University’s Institute for the Preservation of Cultural Heritage (IPCH) and the Peabody Museum of Natural History to develop best practices for recoloring faded mammal taxidermy. The project goals included testing and devising recoloring methods that would minimally alter the texture or sheen of hair and fur and be as reversible or re-treatable as possible.  

The Conservation of Feathers project partnered with IPCH and UCLA/Getty Interdepartmental Master’s Program in the Conservation of Cultural Heritage (UCLA/Getty) to conduct additional conservation research specifically on bird taxidermy and feathered objects. The project goals included devising and testing cleaning and recoloring methods for feathers that do not negatively impact feather structure, appearance, or surface chemistry, as well as investigating consequences from the use of contemporary pesticides on feathers. 

Results of both projects provide a clearer picture of how each approach impacts preservation, and inform better decision-making in treatment and preventive care.​ Though efforts at the AMNH have focused specifically on conservation strategies for taxidermy, the insights and approaches developed in this project can be adapted by those considering these materials and techniques for use in other collections.

Both IMLS-funded projects relied on cross-disciplinary partnerships between conservators and scientists with varying forms of expertise, helping to bridge the institutional gap between natural history, art, and history museums and collections. 

AMNH Conservators 

Lisa Elkin, Chief Registrar and Director of Science Conservation; Principal Investigator & Project Director, IMLS National Leadership grant MG-30-13-0066-13, Recoloring Faded Taxidermy, 10/1/2013–9/30/2016 & IMLS National Leadership Research Grant, MG-60-18-0031-18, Continuing Conservation Research Challenges: The Impact of Cleaning and the Preservation and Restoration of Color on Historic Taxidermy, 9/1/2018–8/31/2022. 

Julia Sybalsky, Conservator, Science Conservation; co-Project Director, IMLS NL MG-30-13-0066-13 & IMLS NL MG-60-18-0031-18

Judith Levinson, Conservator Emeritus, Anthropology Conservation; co-Project Director, IMLS NL MG-30-13-0066-13 & Participating Conservator, IMLS NL MG-60-18-0031-18

Project Conservators

Michaela Paulson, 2020–2022
Renee Riedler, 2018–2020
Fran Ritchie, 2015–2017

Project Interns

Devon Lee, 2022
Nicole Feldman, 2020–2021
Adrienne Gendron, 2020
Leslie Vilicich, 2019–2022
Logan Kursh, 2016–2017
Caitlin Richeson, 2015–2016
Kelly McCauley Krish, 2015
Ersang Ma, 2014–2015  

Project Partners  

Paul Whitmore, former Director of Aging Diagnostics Lab, Institute for the Preservation of Cultural Heritage at Yale University, 2013–2017
Ellen Pearlstein, Professor, UCLA/Getty Interdepartmental Program in the Conservation of Cultural Heritage, 2018–2022
Katherine Schilling, Associate Research Scientist in Chemical and Environmental Engineering and Associate Conservation Research Scientist at the Institute for the Preservation of Cultural Heritage at Yale University, 2018–2022
Vincent Beltran, Assistant Scientist, Getty Conservation Institute, 2018–2022
Aniko Bezur , Director of Technical Studies Lab, Institute for the Preservation of Cultural Heritage at Yale University, 2013–2017

External Advisory Committee Members 

Michael Anderson, former Museum Preparator at the Yale Peabody Museum
Timothy Bovard, Taxidermist, Los Angeles County Museum of Natural History
George Dante, Master Taxidermist and Head of Wildlife Preservations
Catherine Hawks, Museum Conservator, National Museum of Natural History
Jane Pickering, Director of Harvard Peabody Museum of Archaeology and Ethnology; former Executive Director, Harvard Museums of Science and Culture
Stephen Quinn, Diorama Historian and Artist; former Senior Preparator, American Museum of Natural History
Richard Kissel, former Director of Public Programs at the Yale Peabody Museum
Corina Rogge, PhD., Andrew W. Mellon Research Scientist at the Museum of Fine Arts, Houston
Catherine Sease, former Senior Conservator, Yale Peabody Museum
Tim White, Director of Collections & Operations at the Yale Peabody Museum

Previous testing by BASF established the lightfastness of Orasol dyes in two resin films, but results differed according to binder, demonstrating that for an investigation of lightfastness in dyes to be germane to one’s application, it must take place in the material system of their intended use. Lightfastness testing conducted at the AMNH included similar products from different manufacturers and suppliers, applied in solvent alone to quartz plate: Orasol dye manufactured and supplied by BASF; Orasol dye manufactured by BASF and supplied by Kremer Pigments; and chemically equivalent Sorasolve dye manufactured by First Source Worldwide LLC and supplied by Museum Services Corporation. The group of materials chosen reflected an awareness that production methods and storage practices have potential to result in minor chemical difference in the final product.

Modest differences in the color of the dye film when substituting different solvents suggests differences in the way that the deposit interacts with light, so the study also considered what influence, if any, the solvent has on lightfastness: each of twenty-nine dye colors were prepared as 1% w/v solutions in five solvents (acetone, ethyl acetate, ethanol, isopropanol, and propylene glycol monomethyl ether) and airbrushed onto inert quartz plates.

Accelerated aging was conducted in a Q-SUN Xe-3 Xenon Test Chamber in UV-included and UV-filtered conditions. One set of ISO Blue Wool (BW) standard cloths was exposed alongside dye samples in each test cycle so that a BW equivalence could be assigned to each dye application. 

A small number of the dyes tested are highly fugitive (BW 3 and below). However most Orasol dyes have intermediate (BW4) or good (BW 5 or 6) lightfastness. Blue, brown, and black dyes consistently demonstrated values of BW6–7. There was greater variation within red, orange, and yellow color groups, with most dyes classified as BW4–5 and at least one dye in each group classified as BW6–8. With a very few exceptions, the choice of manufacturer, supplier or solvent had no impact on lightfastness beyond the margin of error. 

Dyes deposit very differently onto hair compared to quartz, and the thickness of a deposit impacts its lightfastness. In addition, fibers interact with light differently than quartz, changing the total light exposure delivered to an overlying dye film. And as furs age, they produce reaction products that may affect the chemical behavior of an applied dye. For all of these reasons, a second accelerated light aging study was conducted to see if the dyes’ lightfastness is changed when it is applied to fur.

Twenty-three dyes, a group that excluded colors found to be fugitive testing of the dyes alone, were airbrushed onto samples of white arctic fox and domesticated white fallow deer fur. The fox was chosen to represent fine, smooth-haired fur-bearing mammals, while the deer provides hollow guard hairs.

With a couple of exceptions, testing conducted on fur reflected the results of initial lightfastness testing on quartz. One red dye (CI Red 160 / Orasol Red 330) performed significantly worse on fur, and is not appropriate for use in recoloring fur. One blue (CI Blue 70 / Orasol Blue 855) and one brown (CI Brown 42 / Orasol Brown 2GL) both dropped 2 BW steps when they were applied to fur.

A third accelerated light aging study investigated whether Orasol dyes affect degradation behavior in keratin. Samples of needle-felted bison hair were masked during aging to create areas with different levels light exposure. After aging, FTIR spectroscopy was used to compare markers of oxidation in dyed and undyed furs, a technique that has been successfully used elsewhere to analyze wool materials. Peaks associated with the amino acid cysteine and its oxidation product cysteic acid were evaluated, and the peak-height ratio was charted to understand how it changes with increasing exposure. The formation of these oxidation products, particularly cysteic acid, correlates with loss of strength and implies breakage of the disulphide bonds that stabilize keratin.

In comparison with controls, the spectra showed the dyes did not increase the rate of deterioration (i.e. did not produce higher peak ratios). In fact the average dyed sample appeared less oxidized than controls with the longest exposure, which may be the result of a light-shielding effect produced by the dye. These efforts support the conclusion that Orasol dyes do not contribute to an increase in the degradation of fur, and if anything, can have a protective benefit.

However, in some colors, a reduction in dye solubility was observed after aging. This was  particularly true of the blue dyes. In about half of the other colors, residual insoluble color was slight or just barely visible after washing, which does not allow for reversibility, but does allow re-treatability.

A concurrent real-time benchmarking study was used to develop estimates for how long recoloring treatments might last in a diorama before dyes fade, correlating accelerated aging data with performance in real time under diorama lighting conditions. Three sets of Blue Wool standards were exposed inside the American Bison diorama at the AMNH, their reflectance spectra were recorded quarterly for approximately one year, and then this data was compared with data from the aging studies. 

In general, BW1 and BW1 grade materials exposed in the very bright UV-filtered bison diorama are expected to fade perceptibly in about 2 years; BW2 in about 6 years; BW3 in about 20 years; and BW4 in about 60 years. Estimated time to a noticeable fade is much shorter if UV included, underscoring the importance of UV filtration to the treatment’s longevity. In contexts where exhibit restoration is expected to take place several times per century, like the AMNH, this study indicates that even dyes of intermediate lightfastness, BW5 or perhaps even BW4, may have acceptable longevity in some recoloring applications, as long as lighting is UV-filtered.

In the preliminary study, more than 80 methods described in survey data were evaluated for their potential to cause damage through a series of open-ended cleaning tests conducted on commercial poultry feathers and taxidermy mounts. This included methods relying on brushes, cloths, sponges, suction, pressurized air, water, solvents, solvent mixtures, or surfactants. This initial phase of experimentation on sample feathers helped to understand how alternatives within a category compare. Outcomes informed an assessment of which approaches deserved further attention, and the ultimate selection of 23 representative techniques for in-depth, systematic study. Selections balanced criteria including prevalence in the survey data, availability, cost, health and safety, ease of use, etc.

The short-term impact study explored the short-term impacts of 23 wet and dry cleaning methods on feathers of differing morphology, biopigmentation, and condition states, noting whether each one caused structural damages, and on what scale; and whether it caused changes in color, surface energy, or UV-induced fluorescence  (a visual indicator of oxidation in keratin).​ Because every type of dirt is unique in its response to cleaning, testing focused on damage outcomes rather than the efficacy of each method.

1. Vacuum + glass pipette​
2. Vacuum + Vellux​ blanket (Martex)
3. Vacuum + photo puffer​ (Giotto Rocket Air Blower)
4. Feather brush (into vacuum)​
5. Squirrel hair brush (into vacuum)​
6. Synthetic fiber brush (into vacuum)​
7. Sofft sponge​ (a pastel applicator similar to a cosmetic sponge manufactured by PanPastel)
8. Soot sponge​ (Absorene)
9. Dust Bunny Magnetic Fabric​ (GWJ Company)
10. Groom/Stick​ Molecular Trap/Paper Cleaner (Picreator)
11. Vacuum + Guardsman Dusting Cloth​
12. Deionized water, pH 5.5, at room temp​
13. Synthetic saliva ​(In Situ)
14. Ethanol​
15. Deionized water and ethanol, in 30:70​ solution
16. Deionized water and isopropanol, in 50:50 ​solution
17. Gamsol ​(Gamblin)
18. Dawn Ultra dishwashing liquid, 1% in deionized water (Procter & Gamble)
19. Orvus WA Paste cleaner, 2% in deionized water (Procter & Gamble)
20. Synperonic A7, 2% in deionized water (Croda)
21. Synperonic A7, 2% in Ethanol (Croda)
22. Surfynol 61, 1% in Gamsol (Evonik, Gamblin)
23. Vulpex, 1% in Gamsol (Picreator, Gamblin)

All methods were tested on white mute swan feathers, which lack melanins that contribute strength and protect against light-damage, making them more fragile than most. If a technique has potential to cause structural damage, that damage is more likely to occur in white feathers. Their lack of biopigmentation made it possible to look for color change in the keratin alone. In addition to the swan feathers, wet methods were tested on pink scarlet Ibis feathers, naturally colored with light sensitive carotenoid pigments; and iridescent brown-black common raven feathers, naturally colored with melanin and iridescent structural color in the barbules. This made it possible to look for any bleeding or fading of the biopigments. 

Testing of each method included both fresh, unmodified feathers, and feathers that were aged in an accelerated aging chamber until they were slightly discolored, noticeably weakened and brittle to the touch, and bore chemical markers of oxidation similar to those we have measured in naturally aged historic taxidermy using FTIR. This made it possible to understand the impact of each method at it was applied to feathers representing a range of condition states commonly encountered in AMNH collections.

Though this study investigated the damage associated with cleaning rather than the efficacy of each method, it was nevertheless necessary to work with soiled feathers in order to ensure that cleaning tests were realistic. There is no way to know how may strokes of a squirrel hair brush, in what direction, and with what amount of pressure, are needed to clear dust from a feather until you try it. Standardized commercial soils were used to ensure consistency. However soils on the feathers were incompatible with capturing accurate measurements of color change, viewing minute structural damages, and differentiating damage resulting from the cleaning method from damage occurring in soiling. So testing paired a soiled and unsoiled example of each feather type. Fore example, first the soiled aged pennaceous swan feather was cleaned. Then the unsoiled aged pennaceous swan feather was “cleaned” in exactly the same way. 

Detection and documentation of damages caused by cleaning relied on images of each feather taken before and after using multiband imaging (visible light, ultraviolet-induced fluorescence, and reflected ultraviolet), digital photomicroscopy, short videos, notes captured during cleaning, spectrophotometry, and a simple test measuring the wettability of the feather surface.

The long-term impact study looked at long-term impacts of 9 methods which were deemed to have the greatest potential for leaving residues behind (3 sponges and molecular traps, and 9 detergents). The selected methods were applied to fresh, white pennaceous mute swan feathers and feathers were aged in the chamber for the equivalent of roughly 17 years alongside controls. Testing relied on spectrophotometry and FTIR spectrometry to evaluate whether these techniques are associated with accelerated degradation compared to controls, specifically monitoring color change and the development of chemical markers of oxidation.

1. Sofft sponge​ (a pastel applicator similar to a cosmetic sponge manufactured by PanPastel)
2. Soot sponge​ (Absorene)
3. Groom/Stick​ Molecular Trap/Paper Cleaner (Picreator)
4. Synperonic A7, 2% in deionized water (Croda)
5. Synperonic A7, 2% in Ethanol (Croda)
6. Orvus WA Paste cleaner, 2% in deionized water (Procter & Gamble)
7. Dawn Ultra dishwashing liquid, 1% in deionized water (Procter & Gamble)
8. Surfynol 61, 1% in Gamsol (Evonik, Gamblin)
9. Vulpex, 1% in Gamsol (Picreator, Gamblin)

Survey results generated a list of approximately 12 different general methods and materials that had been used for restoring lost color. 

1. Oil paints 
2. Acrylic Paints 
3. Watercolor/Gouache 
4. Solvent-based conservation/restoration paints or custom-made pigmented resins 
5. Industrial Dyes 
6. Over-the-counter human hair dye (ex. Clairol) 
7. Other dyes (i.e. Rit) 
8. Water-dispersible pigments (ex. Kremer XSL-Pigments) 
9. Dry pigments 
10. Pastels 
11. Pens/Markers 
12. Inks

Through further consultations with colleagues, literature and product research, and we acquired over 30 products within these categories and carried out an initial exploration of how they behave when applied to feathers.  That assessment took into account the range of colors available, the colorant’s sheen and opacity, its handling for blending or creating crisp hard lines, and its affinity for the keratin surface (i.e. whether it beads up or creates an even film, adheres well, or powders off).

This initial effort led to the selection of a small group of promising colorants for further in-depth exploration. All three are pigment-based colorants with existing ASTM ratings demonstrating generally good lightfastness; and with data sufficient to support the informed selection of specific colors for use in treatment.

QoR Modern Watercolor are pigment-loaded paints bound in Aquazol, a resin soluble not just in water but in ethanol and other solvents.

QoR colors can be applied by brush or airbrush, lending themselves to reproducing color fields as well as multi-hued hard-line coloration. A key benefit of QoR colors is their solubility in ethanol, allowing them to be used without water, which can cause barbule deformation in degraded feathers at any concentration. (Though to a lesser extent, ethanol too can cause structural damage, so caution is still required.) There is a wide palette of colors that can be blended on or off the feather, or diluted and layered for luminous color with a matte or satin finish. And due to the presence of a binder, they do not transfer when dry. 

Limitations of QoR colors include the fact that in our tests they were not fully reversible, though they could be reduced with the application of solvent, leaving avenues for retreatment. If diluted to a wash, they tended to wick up the vane, making it more difficult to control a hard line of color. If too concentrated, they sometimes adhered barbules together, making colored feathers more difficult to groom. Translucent colors lacked sufficient opacity to fully cover stubborn soiling or discolored substrates.  

Colorfin’s PanPastels are soft, highly concentrated pastels packed in a pan instead of a stick.

Pastels lend themselves to soft fields of uniform or graduated color, and are particularly useful for maintaining the velvety appearance of plumulaceous feathers. There is a very wide palette that can be blended in-situ on the feather using brushes or sponge applicators. They are applied dry, and pigment particles are mechanically held between barbules rather than adhered by a binder, so barbs/barbules do not clump, and the feathers remain soft. If used on original feathers, this avoids exposing degraded keratin to solvents (including water), and their opacity permits coverage of stubborn soiling or discolored substrates.  

Full removal of PanPastels was not possible in our tests, though with effort, they could be somewhat reduced by vacuuming through a cleaning cloth, offering possibilities for future retreatment. PanPatels were not suited to producing hard lines, and they tended to transfer in handling (i.e. grooming, cleaning, etc.).  

Kremer’s XSL-Pigments are micronized pigments treated with a dispersing agent that allows them to be mixed with water to make high-intensity solutions without an added binder.

Kremer’s XSL-Pigments can be applied by brush or airbrush to reproduce color fields as well as multi-hued hard-line coloration on both original and fill feathers. The palette of colors available is more limited than other materials tested. They must be dissolved in water before any solvent carrier is added, meaning that they present a greater risk if used on degraded feathers. XSL-pigments were sometimes found to be difficult to control, giving uneven coverage and tending to become dull when built up. Barbs/barbules tended to clump, though they could be manually repreened. Attemptes to remove XSL-pigments dry were not successful; wet removal was only marginally better, but presents greater risk of damage.