Color Analysis and Microfade Testing of the 1918 Curtiss Jenny U.S. Airmail Stamp Thomas Lam1, Susan Smith2, Scott Devine2, and Edward P. Vicenzi1 Abstract The Inverted Jenny is the best recognized stamp in the United States and is displayed in the National Postal Museum’s Gems of American Philately exhibition. A careful approach to exhibition management, including a conservative lighting plan, has allowed the museum to permanently exhibit some of its most precious objects such as the Jenny without subjecting the objects to unnecessary light exposure. One goal of this study is to determine if the muse- um’s current exhibition practices are effective in protecting such rare philatelic objects given we know very little about the Jenny’s blue and red inks or how they interact with exposure to light.  Since the Inverted Jenny could not be used for variability, as well as the need to consider the impact testing, surrogate stamps were purchased and their of exhibit lighting versus daylight/sunlight on stamp colors analyzed and verified using color reference coloration. Ultimately, both the stamp paper and the materials designed specifically to resist fading. Two colors of the Jenny were determined to be stable with independent systems were used for the color mea- respect to the lighting plan developed for the Invert- surements and were found to be comparable de- ed Jenny at the National Postal Museum. spite differences in the instrumentation. One of the replicate stamps was subjected to accelerated light Introduction exposure under controlled conditions and its colors measured at the end of the testing period to simulate Less than fifteen years after the Wright brothers’ the impact of long-term exhibition lighting.  first successful sustained flight by a piloted aircraft, the US Post Office Department offered special deliv- The measurements demonstrated that there is ery airmail service and the Bureau of Engraving and not a single-color value that represents the red, blue, Printing (BEP) issued a striking red and blue 24¢ and uncolored regions even within a single Jenny. Curtiss Jenny postage stamp. The stamp is com- These regions faded, or experienced shifts in color, monly referred to as “the Jenny” after the US Army’s at different rates with exposure to light; the stamp Curtiss aircraft that carried mail on the inaugural paper changed at the same rate as the blue regions, flights in May 1918. The stamp is not only recognized with the red regions shifting in color at a slower pace. as the first US airmail stamp, but also the first stamp These color shifts act upon the spread of red, blue, issued worldwide for regular airmail service and the and paper colors measured before accelerated labo- first bi-colored airmail stamp (Kirker 2006). The Jen- ratory tests were performed, highlighting the impor- ny is best known for the single sheet of 100 stamps tance of the understanding the initial range of color in which the plane was printed upside down. As a result of this printing error, the inverted Jenny is the best recognized stamp in the United States. 1Smithsonian Institution, Museum Conservation Institute, Despite the stamp’s popularity and although Suitland, MD 20746 2Smithsonian Institution, National Postal Museum, Wash- much is known about its design and printing, we ington, D.C. 20002 are unaware of information regarding the material 52 / Institute for Analytical Philately Color Analysis and Microfade Testing of the . . . Jenny US Airmail Stamp properties of the inks used and their susceptibility of paper-based objects. Current museum lighting to light exposure. It has been known for some time standards were developed in the 1970s with a focus that the BEP’s Engraving Division’s Stamp History on reducing visible light to only what is needed to for the 24¢ Air Mail Stamps (Airplane), Series 1918, sufficiently view the object, and eliminating ultravi- contains a level of uncertainty regarding ink formu- olet light altogether when exhibiting light sensitive lation and nomenclature within the records (Engrav- objects such as paper and textiles (Thomson 1978). ing Division 1960). Given the short period of time The National Postal Museum (NPM) limits expo- during which the stamps were printed, roughly two sure of paper-based objects to 54 Candela steradian/ months, it is reasonable to assume however that the square meter (the equivalent of 5 footcandles (fc) ink formulations and materials did not change during or 54 lux) for no more than three to six months at a production. time assuming eight to ten hours of exhibition per The period during which the Jenny was produced day. To exhibit rare philatelic materials for longer was a difficult one at the BEP, as it was in the US as a periods, NPM employs various strategies including whole. World War I created challenges for the BEP, wall mounted pullout frames, motion sensors, and including its inability to take delivery of Prussian SmartGlass technology, all of which ensure that Blue and red (referred to as “standard red” or “print- objects are illuminated only when being viewed by a ing red S225” in documents) dry color or powder visitor. These specific technologies, combined with pigments from Germany (Gebr. Heyl & Co. 1917, custom-designed exhibit spaces and a program of Treasury Dept. to Speaker of the House 1917) and ongoing monitoring, allow for the creation of cus- difficulties “obtaining steel of the proper degree of tomized lighting plans for a range of philatelic and hardness” (N. Underwood to James Wilmeth 1918) postal history objects. to replace worn printing plates. The war also neces- Balancing light exposure with the need to make sitated staffing shifts to print ‘the more important objects available for short term display as well as [war] bonds” ( James Wilmeth to Wilfred A. French long term exhibition is a complex formula that 1918). There was also the lack of sufficient machin- relies on understanding the material composition ery to handle the printing needs, leading to the use of of the object, the nature of the light source, and a retired hand-roll printing press (the spider press) to careful management of the light level. The research print this air mail special delivery stamp. With all the discussed here addresses the fading properties of ongoing issues during that period, it is surprising that postage stamps by using a set of color-based spec- the decision was made to create a bi-color intaglio trometry tools to quantitatively measure colors and stamp, which was a labor and supply intensive pro- evaluate the nature of the color change imposed by cess. The turn around time from the creation of the exposure to museum lighting. One of the goals of this dies from May 4 – 10 to the printing of plate proofs study is to determine if that the already conservative on May 10 and 11, to the delivery of the first stamps approach being taken at the National Postal Museum to the main Washington DC post office on May 13 is not significantly damaging collection material. A represented an accelerated production schedule second goal is to show that the current exhibition (Engraving Division 1960). The patriotic color se- practices are effective in preserving these objects for lection for the Jenny together with the stamp’s novel future generations, in part by estimating long term purpose and the high cost at 24¢ was eight times the cumulative illumination exposure. general cost of a 3¢ (USPS 2022); underscore the importance of the project of America’s first postage stamp for airmail. Materials and Methods The research and exhibition value of historic Materials stamps is often dependent upon the condition of The 24¢ Curtiss Jenny postage stamp was cho- the paper substrate and subtle nuances of color in sen for this study for reasons related to the stamp’s the printing inks and dyes, so extended or recur- bi-colored design, the availability of surrogate copies ring exhibitions of rare and unique objects like the for testing, and the notoriety of the famous “inverted Inverted Jenny present a challenge. Exposure to light Jenny.” In this study, research was performed on six causes irreversible damage to some materials. Visible Jenny stamps (Figure 1A) with a seventh stamp on light damage may include aesthetic changes such as cover (Figures 1B and C). Similar to other philatelic fading or color shifts, while prolonged exposure may studies, information from plate proofs can strengthen lead to structural damage including embrittlement philatelic research (DeBlois and Harris 2011, Charles Thomas Lam, Susan Smith, Scott Devine, Edward P. Vicenzi Institute for Analytical Philately / 53 Figure 1. Curtiss Jenny stamps and additional materials used in this study. A) Images of the six objects examined, each stamp is 2.0 cm x 2.5 cm (with the red frame being 1.9 cm x 2.2 cm) B) An enlarged view of the Jenny on cover with regions of interest for the substrate paper, blue, and red regions highlighted in black-, cyan-, and white-unfilled circles. C) Image of Jenny 7 cover and the vendor-supplied white reference tile. D) Blue vignette for 24¢ stamp; plate number 8493 (National Postal Museum, Object number 0.242263.15824) printed on paper of height x width: 27.94 x 46.04 cm (11 x 18 1/8 in.). E) 24¢ carmine frame for USA Scott C3; plate number 8492 (National Postal Museum, Object number 0.242263.15825 printed on paper of height x width: 28.58 x 35.56 cm (11 1/4 x 14 in.). 54 / Institute for Analytical Philately Color Analysis and Microfade Testing of the . . . Jenny US Airmail Stamp 2017, Charles 2020). The blue and red plate proofs been. This allowed for the same locations as the VSC from BEP plates 8493 (Figure 1D) and 8492 (Figure color measurements depicted in Figure 1B to more 1E) were made available for analysis from the Smith- easily be evaluated. MFT is a spectroscopy technique sonian’s National Postal Museum (NPM). These developed by Whitmore et al. (Whitmore, Pan et al. proofs are from the only two plates used for this 24¢ 1999) as a non-contact and “virtually non-destruc- stamp. The proofs have experienced limited exposure tive” method for determining small increments of to natural and artificial lighting given their continu- reflectance color shift on cultural heritage materials ous storage at the BEP and NPM for the last 100 plus (Ford and Druzik 2013). Using a xenon light source, years. Therefore, as trial impressions for a stamp that this system generates a spectral distribution in the began production immediately following their print- visible wavelength range from ≈ 370 nm to ≈ 760 nm ing, they likely represent the closest approximation (Whitmore, Pan et al. 1999), which has been filtered to the original stamp colors available to researchers through parts from a “Fading Test System” (Newport and the philatelic community. Oriel Corporation). For MFT setup, a custom head serves multiple purposes as it: 1) enables the light Color measurement source from a fiber optic cable held in the center of The Foster and Freeman Video Spectral Com- this mount to be readily focused on the target, 2) parator (VSC8000/HS) at the NPM was used to provides a fixed position for the spectrometer collec- determine quantitative color. Six locations in each tion fiber that carries the detected signal is routed to of the substrate paper, blue, and red regions were the spectrometer at the one end, and 3) has a digital measured for each stamp as shown in Figure 1B. endoscope magnifier positioned at the opposite end To obtain color information on the unpigmented that provides a microscale view of the target area region of the stamp a vendor-supplied white refer- (Figure 2A). Prior to collection of each series of ence was used. For the red and blue color analysis, color fading experiments, dark and white referenc- the paper substrate served as the white reference. es were collected to mitigate potential drift of the Using the stamp paper as the white reference allows reflected light signal. Manual insertion of a neutral for the interpretation of the ink color in contrast to density filter into the optical path reduced the light the surrounding substrate given that the substrate intensity and allowed for the safe navigation to re- itself – the stamp paper – can influence/influences gions of interest using the microscope image, and an the color. In other words, measuring the color of the example of the filtered spot is highlighted and shown substrate enables the study to account for it. For- in Figure 2B. With the neutral density filter removed, ty-nine measurements from the north position of the the illuminated spot is on the order > 1 mm (Figure red elliptical “24” region at the bottom right corner 2C). Recent developments in the assessment of the of stamp impressions in the red proof were obtained. light fading spot size indicate the full width at half the While fifty measurements were collected from stamp maximum (FWHM) to be < 1 mm (Świt, Gargano et impression in the blue proof. Additionally, 100 mea- al. 2021). Given the MFT measurement is dependent surements in the top center unpigmented sky region upon the working distance, a custom spacer ≈ 1.1 cm were made on multiple regions of both proofs. For in height was additionally used to achieve a constant stamp 7 (Figure 1B and C), an additional series of value, and this height above the target generated eighteen data points were collected using the vendor a focused spot for this optical geometry. With the provided white reference. Color analyses collected focused spot on a region of interest, the visible reflec- using the VSC represent point spectra obtained tance signal sent to the spectrometer was recorded while viewing the stamp at a constant horizontal field and stored using Spec32 acquisition software. Using width of ≈14.1 mm, and a circular region of interest a custom script (Laudato Beltran, Pesme et al. 2021), for reflected light directed to the spectrometer was the International Commission on Illumination approximately 0.11 mm in diameter with a spectral (CIELAB) color space coordinates L*, a*, b* and col- resolution of 3 nm. or difference by ∆Ε*ab (also referred to as ∆Ε * 76) were exported from the acquisition software for a near real Experimental color fading time display (Robertson 1977). CIELAB color space Microfade testing (MFT) was performed on coordinates of L* is expressed on the z-axes and Jenny 7. Ats the only Jenny stamp on a cover, its represents the lightness to darkness value, the a* is weighted surface kept it flatter and in a more stable for red/green axis, and b* for the yellow/blue axis as position than individual stamps alone would have shown in Figure 3A. Thomas Lam, Susan Smith, Scott Devine, Edward P. Vicenzi Institute for Analytical Philately / 55 Figure 2. (A) Picture of custom mount used for Microfade testing. (B) Image obtained using the endoscope mag- nifier with the neutral density filter in place (C), and with the neutral density filter removed during data acquisition. Within this direction of the stamp in B and C there are 11 perforations along the 2 cm distance which is ≈ 0.18 cm per perforation. Due the challenges of accurately measuring the method can be informational. Successful color dose for a focused spot on the millimeter- to sub-mil- analysis has also been demonstrated using other limeter-scale, coupled with run-to-run optical align- approaches, e.g. Red, Green, Blue (RGB) (Cibulski ment variations of the MFT that are different from 2015, Cibulski 2015, Cibulski 2017); hue, saturation, one MFT instrumental setup to another , a qualita- and brightest (HSB) (Cibulski 2015, Mckee 2015, tive assessment of the relative kinetics of color fading Hisey 2020); hue, saturation, and luminance (HSL) was performed by comparing the stamp materials (Cibulski 2015, Cibulski 2017); CIE-L*(uv) plots to different grades of International Organization for (Allen and Lera 2012, Caswell 2012, Lera, Giaccai et Standardization (ISO) blue wool standards that have al. 2012, Hofmeyr 2020), CIE-XYZ (Cibulski 2015), documented light sensitivities (Materials Technolo- or analysis Romney Model in Munsell space ( Judge gy) (Lowe, Smith et al. 2017). In this study, ISO blue 2015). wool 1 and ISO blue wool 2 light fastness standards However, in quantitatively comparing the color purchased from SDL Atlas USA (Rock Hill, SC) were difference among multiple colors, the color differ- used for the comparison. In the MFT data collection ence by ∆Ε*ab formula does not provide a complete of the blue wool, the topography (height variation) analysis as it assumes that the CIELAB color space of the weave is taken into consideration. Three mea- is perceptually uniform in the 3-dimensional (3D) surements were collected for both blue wool 1 and color space, which it is not. The color difference by blue wool 2. For each region of interest of the stamp DE2000 or ∆Ε*00 formula addresses this limitation. to be evaluated, the test was performed for run dura- The color difference by ∆Ε*00 takes into account the tions of 300 seconds.(Materials Technology) perceptual non-uniformities of the CIELAB color space by implementing five mathematical corrections Color space to aspects of color attributes within the formula. The CIE-L*ab color model (Cibulski 2015, Judge (See glossary for details; Luo, Cui et al. 2001). Upon 2015) and use of color difference by ∆Ε*ab (see glos- request, the authors can provide an Excel worksheet sary, equation 1) calculated from CIE-L*ab are both to convert color difference ∆Ε*ab values to the color color analysis expressions reported in the analytical difference ∆Ε*00 values (Kuzio 2018). For this study, philately literature (Hofmeyr 2020). For color anal- color difference by ∆Ε*76 calculations have been ap- ysis involving a single type of color and discussion plied to both VSC and MFT data. It should be noted regarding a specific color, the ∆Ε*ab color difference that color difference by ∆Ε * 00 and color difference by 56 / Institute for Analytical Philately Color Analysis and Microfade Testing of the . . . Jenny US Airmail Stamp Figure 3. A) Schematic representation of color space showing end-member colors for a*, b*, and L* intensities with concentric gray circles represent increasing chromaticity values outward from the origin. B) 3D plot of average values of L*, a*, and b* for the paper substrates. C) 2D plot of the average values of a* and b* along with a* and b* uncertainties for the “white” data based upon population statistics. ∆Ε*ab are not equivalent, and generally differ by a fac- seen by the spectrometers of the VSC and MFT by tor 2, where the color difference by ∆Ε*ab is typically using the L*, a*, and b* inputs. This software tool the larger value (Ford and Druzik 2013). also converts color values between a variety of color To properly evaluate the accuracy of the color models, including those found within philatelic analysis carried out on the VSC and MFT, color literature e.g. RGB Adobe 98, HSLCIE-L* ab, and measurements were also collected on red, green and CIE-L*(uv) (EasyRGB). blue light-fast ceramic tiles obtained from Hale Color Consultants, Inc., (Baltimore, MD) with reference Statistical evaluation of color data tile colors traceable to the National Bureau of Stan- A variety of statistical approaches are available to dards (now the National Institute of Standards and determine which color measurements are equivalent Technology). Color visualization was performed to one another within uncertainty, or conversely, using an online software tool to simulate the colors have differences that are significantly greater than expected by random chance. The statistical spread Thomas Lam, Susan Smith, Scott Devine, Edward P. Vicenzi Institute for Analytical Philately / 57 of an average value of randomly distributed data is mensional (3D) plot of the mean color values in termed the standard deviation (σ) of the population. terms of CIE L*, a* and b* for measurements taken Since the deviation can be positive or negative, the from the paper regions of the Jenny blue and red spread of the data that results in plus or minus one proofs, as well as for the paper locations described or two standard deviations lies within the bounds of above for the seven Jenny stamps. (Figure 1B). L* random chance to within 68 or 95 percent confidence values of both the blue and red proofs are larger than respectively. In this study, the standard deviations those for the stamps. The differences in L* were not reported represent the spread of the population of unexpected since the proofs are on card (Robert A. all color measurements and were calculated using Siegel Auction Galleries 2017) rather than on stamp an MS Excel 365 spreadsheet. The evaluation of the paper. Figure 3C is two-dimensional (2D) plot of simple statistical output was made using analysis of the mean values for a* and b* that yields a top-down variance (ANOVA), as well as through a compari- projection of figure 3B with associated uncertainties son between each stamps’ a* and b* color values to to better describe the distribution of color values. the Jenny blue and red proof “controls” using the Considering the spread of the mean values for the Holm-Sidak method (see glossary) with a cutoff for paper color of the seven stamps, Jenny 5 and Jenny 7 dissimilarity at > 95% chance. The ANOVA computa- are > 1 σ different in the b* value, or the blue-yellow tion was performed using Sigma Plot v14.5 (Inpixon) dimension (Figure 3A), relative to the cluster formed software. by the other five Jenny stamps. As seen in Table 1, the simulated colors from the Results measured color values for the stamp papers do differ. Color Analysis of the Stamp In the examination of the paper regions the blue and red proofs (Figure 1D and 1E), the color difference Uncolored Paper is noticeable since we know they are not the same as the paper used in the stamp production. Given The uncolored regions of the Jenny are defined as the proof papers were not used in production of the areas free of any print mark or patterns and represent stamp, they do not represent a “control” for color the stamp paper substrate. Figure 3B is a three-di- analysis of the stamp paper, and therefore no color Table 1. Averages and standard deviations of the population for L*, a*, and b* within the uncolored regions of the substrate paper for the proofs and stamps, along with their simulated color. 58 / Institute for Analytical Philately Color Analysis and Microfade Testing of the . . . Jenny US Airmail Stamp Figure 4. A) 3D plot of average values of L*, a*, and b* for the blue data B) 2D plot of the mean values of a* and b* with the a* and b* uncertainties represent the standard deviations of the population for the blue data. difference by ∆Ε*00 values were reported. The gray- simulation for Jenny 5 stands out as an outlier among ish blue simulated for the blue proof paper with a the stamp papers, which is consistent with the visual smaller b* value and the reddish tint in the red proof observation that it was significantly “brighter” than with a larger a* value are qualitatively consistent the other six stamps and was indeed found to have with the colors observed in the paper substrate in the largest L* value. Figure 1D and 1E by naked-eye inspection. The color Table 2. ANOVA analysis for the blue colors with Holm-Sidak comparisons between the stamps and the blue proof control. Thomas Lam, Susan Smith, Scott Devine, Edward P. Vicenzi Institute for Analytical Philately / 59 Table 3. Averages and standard deviations for L*, a*, b* measurements, and ∆Ε*₀₀ of the blue data for the proof and the Jenny stamps with their simulated color. Blue Regions est ∆Ε*00 difference value in comparison to the proof. The blue proof serves as a de facto color control Jenny 1 had the greatest ∆Ε * 00 difference value, which based upon its controlled storage conditions with re- also had the largest average L* value. spect to light exposure. The average L* values of the blue proof were smaller than that for all the stamps, Red Regions except Jenny 6 (Figure 4A). For the blue regions, Applying the same methodology used for the blue Jenny 2, 3, and 7 are most similar to the blue proof proof above, the red proof can also be understood as a both in terms of b* and a* data (Figure 4B). Figure control for the red frame printed in 1918. In this case, 4B also shows that Jenny 5 is > 1 s different in the the red proof has L* values intermediate to the distri- b* when compared to the blue proof. The analyses of bution of means values for the Jenny stamps (Figure variance of the blue data, along with comparisons of 5A). The top-down projection of figure 5A shows each stamp with the blue proof serving as a control, Jenny 7 was most similar to the red proof (Figure 5B). are shown in Table 2. Jenny 2, 3, and 7 were indeed Jenny stamps 1, 2, 5, and 6 have statistical overlap determined to be statistically indistinguishable to the with the red proof in the b*, or yellow-blue dimen- blue proof. In other words, the spread of blue color sion. Conversely, relative to the red proof, Jenny 2-6 measured within Jenny 2, 3, and 7 falls within the red data are > 1 s in the a*, or red-green dimension, range of blues measured in the proof to within a 95% with Jenny 4 exhibiting the greatest difference in a*. confidence interval. ANOVA results indicate Jenny 7 was most similar Though the graphical visualizations of the color to the red proof among all stamps in both a* and b* data were useful, a table of values along with simulat- dimensions (Table 4). ed colors demonstrates that Jenny 6 has the smallest Consistent with Figure 5B, the tabulated values * calculated color difference by ∆Ε* with a value of for ∆Ε00 value also show that Jenny 7 exhibited the 00 * 3.86 and is qualitatively consistent with the simulated smallest ∆Ε00 color difference value of 2.51 relative to color for the blue proof. In this case, the ∆Ε* calcu- the red proof (Table 5), and that the simulated color 00 lation takes into account the smaller and more simi- of Jenny 7 was also qualitatively consistent with that lar L* value of Jenny 6 in relation to the blue proof for the red proof (Table 5). Table 5 also shows Jenny * discussed above. Though Jenny 5 had the largest color 3 had the greatest ∆Ε00 difference value of 7.4 relative difference in b*, it was not the stamp with the great- to the red proof, and this difference is largely a func- tion of the low L* values measured for Jenny 3. 60 / Institute for Analytical Philately Color Analysis and Microfade Testing of the . . . Jenny US Airmail Stamp Figure 5. A) 3D plot of average values of L*, a*, and b* for the red data B) Plot of the mean values a* and b* with the a* and b* population standard deviations for the red data. Table 4. ANOVA analysis for the red colors and Holm-Sidak comparison. Thomas Lam, Susan Smith, Scott Devine, Edward P. Vicenzi Institute for Analytical Philately / 61 Table 5. Averages and standard deviations of L*, a*, b*, and ∆Ε*00 of the red data for the proof and the 7 stamps with their simulated colors. Color Validation unexpected. Importantly, studies using multiple systems should seek to document differences relative Given the two instruments’ different optical to a standard if the comparisons of the data sets are setups, spectrometers, and detectors, the color made. Table 6 shows such a comparison between differences obtained from the instruments were not the VSC and MFT measurements relative to color Table 6. Mean and the population standard deviation results comparing the VSC and the MFT color measurements of the red, green and blue reference tiles, with the simulated colors and images of the actual RGB color represen- tation of the tiles for comparison. 62 / Institute for Analytical Philately Color Analysis and Microfade Testing of the . . . Jenny US Airmail Stamp Figure 6. Plot of color difference measured by ∆Ε*00 for the uncolored paper regions, six blue regions, six red regions for Jenny 7 in relation to blue wool 1 and blue wool 2. tile references. The average difference for VSC spot Microfade Testing measurements had ∆Ε*00 color difference values of < 2.36 for all reference tiles. While MFT data yielded All MFT runs of the regions of interest shown in comparable ∆Ε* color difference values for the red Figure 1B for Jenny 7 were performed for 300 sec-00 reference tile, ∆Ε*00 color difference values of slightly onds, with the results plotted in reference to the ISO * > 5 were obtained for the green and blue tiles. A pos- blue wools. The ∆Ε00 color difference values reported sible reason for the slightly elevated ∆Ε* color dif- represent the color change after 300 seconds of illu-00 ference values could be related to the low L* values mination. For this experimental session, the ISO blue for the green and blue tiles. Since the MFT measure- wool 1 had an average ∆Ε * 00 color difference of 1.55 ment is generated from a reflected color signal from ± 0.03 (n=3); the ISO blue wool 2 had a ∆Ε * 00 color a focused white light spot, the absorption of darker difference of 0.87 ± 0.01 (n=3) as shown in Figure 6. colors results in a smaller reflected light signal/noise From the MFT results, the uncolored stamp paper, ratio sent to the spectrometer. Such absorption will blue, and red regions all change at a rate < both blue also take place for the VSC measurement, but with a wool 1 and blue wool 2 (Figure 6). The six uncol- smaller relative loss of signal given the dispersion of ored paper regions were the most sensitive to light source light over the entire light chamber compared exposure, relative to red and blue regions, and had to the focused MFT source. Therefore, the marginal an average ∆Ε * 00 color difference of 0.49 ± 0.04. The increase in ∆Ε* color difference of the MFT is likely red regions had an average ∆Ε * 00 color difference of 00 systematic in nature, and due to inherent optical 0.33 ± 0.05 (n=6), which is statistically different from differences between the two color measurement the change noted in the white/uncolored regions of systems. the paper. The blue regions had average ∆Ε * 00 color difference of 0.44 ± 0.09 (n=6), and the blue region Thomas Lam, Susan Smith, Scott Devine, Edward P. Vicenzi Institute for Analytical Philately / 63 Figure 7. The plots of a* and b* changes for MFT as a function of time for the paper material, red, and the blue regions for Jenny 7. fading behavior was intermediate between rate of object, which has been reported to be equivalent to change for white/uncolored paper and red regions. a ∆Ε*00 of ≈ 1.6 (Ford and Druzik 2013); it should be The stamp faded more slowly than blue wool 2, noted that JND/JNF is an approximation with other which suggests exhibition lighting could be designed perception ranges reported and the topic of “accept- to handle up to a maximum of 1.2 Mlux∙h below able” color change is considered endlessly debatable the “Just Noticeable Difference/Fade” ( JND/JNF) (Michalski and Dignard 1997). For reference, if the (Lowe, Smith et al. 2017) threshold for the life of the stamp paper and colors faded more rapidly than blue 64 / Institute for Analytical Philately Color Analysis and Microfade Testing of the . . . Jenny US Airmail Stamp wool 2, a more conservative exhibition light level of a MFT is a useful tool for the curation of artifacts, maximum of 0.4 Mlux∙h would have been advised for but it is important to consider its limitations (Ford the life of the object (Lowe, Smith et al. 2017). and Druzik 2013). MFT is designed to provide data L* values of all the test regions were found to be for materials classified as light sensitive, meaning ma- constant during the tests. For the stamp paper, a* and terials that would be able to handle 3.6 Mlux∙hrs of b* both decrease during the MFT (Figure 7) indicat- lighting at a maximum. MFT cannot predict nonpho- ing a slight decrease in red and a shift from yellow to tochemical changes, such as long term polymer deg- less yellow respectively. For the blue regions, very radation or biological attack (Ford and Druzik 2013). slight increases in a* (green to less green) with a In MFT, the term ‘reciprocity’ is used to correlate noticeable decrease in b* (increasing blue). While accelerated laboratory color fading with that expect- for the red regions, a* is constant while b* decreases ed for objects experiencing lower illumination levels indicating a small shift from yellow to less yellow used during exhibition display, and an assumption (Figure 7). is made that color shift depends only upon the total exposure dose. Such a reciprocal relationship can Discussion fail when chemical reactions occur owing to induced local heating or dehydration reactions (Ford and Color difference by ∆Ε*00 is a more robust repre- Druzik 2013). Even the MFT color measurement sentation of expressing color differences relative to itself has been reported as a source of uncertainty color difference by ∆Ε*ab , because it corrects for the (Ford and Druzik 2013), which is why the color com- perceptual non-uniformities of CIELAB values of parisons presented here are regarded as a focal point the CIELAB system (Luo, Cui et al. 2001), and has of this study. values roughly half the magnitude of an equivalent * To evaluate differences between color measure-∆Εab value (Ford and Druzik 2013). In particular, the ments systems, a set of VSC data was collected using analysis of the blue colors showed Jenny 6 yielded a dedicated white reference tile (used in the MFT the smallest color difference as it had the lowest * portion of the study) for Jenny 7, the stamp whose ∆Ε00 value. However, based upon the mean values colors most closely approximate the blue and red of for a* and b* one would have identified Jenny 2, 3 the Jenny proofs. The ∆Ε* color difference for the and Jenny 7 as being most similar to the Jenny blue 00 colors and paper have a difference value of 5.48 and proof, not Jenny 6. The explanation for these seem- have a consistent simulated color output as shown in ingly disparate results stems from the fact that the Table 7, and are qualitatively similar, particularly for differences in L* values are only accounted for when * the uncolored paper and blue regions of the stamp. color difference by ∆Ε00 is computed. Even when the * As described in the color validation section above, ∆Ε00 calculation is performed, it should be noted that differences between two data sets derived from two there is not a single-color value that represents any different measurement setups are not unexpected. of the Jenny’s regions - red, blue, or uncolored- even The blue values from MFT for the stamp in the case within an individual stamp. Instead, there exists a of Jenny 7 were better than the blue tile in Table 6, range of color values that are defined by statistical which is likely because of the higher L* of the blue spread in measurements. Recognizing that L*, a*, color in the stamp. Though paper substrate was used and b* coordinates for these stamps cluster about a as the white reference in the VSC color measure- value, rather than defining a unique set of values, is ments and yielded color information about the pig- important to understand. Appreciating that there ment free of the substrate; a white tile reference was is a distribution of color values for an object, rather used for comparison of the VSC and MFT data since than a solitary color, is most important with respect the MFT measurement contains the color pigment to MFT results since the goal of the technique is to on the substrate. This is needed to eliminate a poten- measure a shift in color where the starting color value tial systematic error in the difference calculations. is not uniquely fixed but is locally diffuse. Finally, The similarity in the L*, a*, and b* values obtained we note that the non-linear nature of color fading in for the two-color measurements systems, and their general may mean that the Jenny stamps experienced simulated output, allow for a reasonable level of con- more rapid loss of color in the years following print- fidence in comparing the experimental color changes ing relative to the 100+ year old object examined in with the color data for the Jenny stamps. this study. For the a* and b* data for both the blue and red regions of the Jenny stamp, the ANOVA results Thomas Lam, Susan Smith, Scott Devine, Edward P. Vicenzi Institute for Analytical Philately / 65 Table 7. Mean and population standard deviations for L*, a* and b* comparisons of VSC and MFT for the different colors of Jenny 7 with VSC data using a dedicated tile for a white reference as opposed to stamp paper. showed Jenny 7 to be the most similar to the blue and The previously described NPM limits on expo- red proofs. As a result of their statistical similarity, sure of paper-based objects to 54 Candela steradian/ Jenny 7 and the blue and red proof color data have square meter (5 footcandles (fc) or 54 lux) for no been grouped as a shaded area in Figure 8A and 8B more than three (≈ 0.04 Mlux∙h) to six months (≈ respectively. From the additional light exposure from 0.08 Mlux∙h) at a time assuming eight to ten hours the MFT, blue regions experienced a slight increase of exhibition per day falls within the maximum of in a* by 0.16 units, while b* decreased 0.57 units, 1.2 Mlux∙h for the lifetime of the object. In fact, the while for the red regions a* was constant and b* de- current lighting plan designed for the National Postal creased by 0.64 units. The vectors representing these Museum’s Gems of American Philately exhibition, changes are depicted in Figures 8A and 8B. Those which opened in 2013, features a gradient plan in vectors are represented as arrows initiated from the which visitors enter the William H. Gross Stamp grouped shaded areas of the proofs plus Jenny 7, Gallery at a light level of 54 Candela steradian/sq m and proof plus Jenny 2, 3, and 7 for the red and blue (5 fc) and slowly transition to even lower light levels regions respectively. Note that the vector is scaled of 21.5 -32.3 Candela steradian/sq. m (2-3fc) in the by up a factor of 10 x to show the direction of the Gems of American Philately exhibition. As an added color change expected for the stamp when displayed measure to protect the philatelic objects exhibited under museum exhibition lighting conditions. Also in Gems of American Philately, the exhibit cases are note that the representation of the color shift arrow equipped with motion sensors that activate only is not quantitative and is provided as a graphical aid. when a visitor steps directly in front of a recessed The history of the lighting conditions Jenny 1-7 have wall mounted exhibit case. Once activated, the fiber suffered is not known to us, but it is important to optic lights inside the exhibit case are engaged for note that the MFT setup is designed with UV and only 20 seconds at a time. Daily monitoring is re- IR cutoffs beyond which natural lighting will extend quired to track visible damage and to ensure that (Whitmore, Pan et al. 1999). Specifically, long term all elements of the lighting plan are functioning as exposure to lighting with a UV component, namely designed. Broken window shades, motion sensors, sunlight, would yield results the MFT is not designed and lighting components must be identified and to explain. Fading from sunlight exposure is then a repaired immediately to ensure that light levels do potential explanation for the reported a* values > 1 not exceed established limits. Working closely with σ in comparison to the proofs. This is most clearly building facilities staff and museum security to cover shown for the red region data (Figure 8B), whereas cases or temporarily remove objects from exhibition the blue regions showed a significant degree of over- until problems can be resolved is an essential aspect lap with the blue proof (Figure 8A). of managing long term exhibition of light sensitive materials and a key factor in the successful imple- mentation of any lighting plan. 66 / Institute for Analytical Philately Color Analysis and Microfade Testing of the . . . Jenny US Airmail Stamp Figure 8. Direction of potential color shift on the a* and b* plot of the blue and red color data for the Jenny A) blue regions and B) red regions. Blue and red shaded areas represent the grouping of Jenny 2, 3, and 7 with the blue and Jenny 7 with the red proof data respectively. Blue and red arrows represent the vector of MFT color fading (see text) and the yellow shaded regions represent a color potentially owing to exposure to sunlight. Thomas Lam, Susan Smith, Scott Devine, Edward P. Vicenzi Institute for Analytical Philately / 67 Table 8. Tabulated calculations of daily exposure duration and projected decadal light exposure based on viewing frequency which is well under the 1.2 Mlux∙h limit. This careful approach to exhibition management the 1918 blue vignette proof and the color values for and lighting plan design has improved the visitor the stamps examined.  However, color fade testing experience by making some of the most precious offered no insight into differences observed for the objects in the national collections permanently avail- color change between the red frame in the red proof able for viewing without overexposing the objects relative to the red color measured in the examined to unnecessary light. Such an approach is consistent stamps. The significant difference between the two with the calculations presented in Table 8. Even disparate color change vectors for the red regions of using an assumption of 1000 views/day lighting at the Jenny suggests that UV-daylight exposure may be 26.9 Candela steradian/square meter for 10 years, responsible for color differences observed between this would result in a maximum lighting exposure the proof and the vast majority of the stamps.  There of 0.55 Mlux∙h, which is well below the 1.2 Mlux∙h exist many opportunities to extend such laboratory limit. Because 1000 views/day is likely on the high color analysis to other stamps of interest to the phil- end of visitor frequency, data for 100 views/day, 250/ atelic community, and we hope this study provides a day and 500/day were also tabulated. These new methodological framework for those future efforts to data and calculations affirm that the conservative predict color changes in exhibition settings.  In future approach being taken at the National Postal Museum efforts, we plan to turn our attention to character- has been successful in protecting collection material izing the compositional nature of the blue and red during the nearly ten years that the Gems of Ameri- inks used to print the Jenny to gain a material-based can Philately exhibition has been open. Daily gallery understanding for the color differences observed. checks and periodic condition surveys of the objects on view have additionally revealed no visible changes Acknowledgements in color or appearance. Allyce McWhorter from Foster and Freeman is Conclusions thanked for introductory color measurement training and general VSC8000HS instrumentation resource. High accuracy color measurements for the We are grateful to Dan Piazza from the National Post- bi-colored 1918 Curtiss Jenny airmail stamp were al Museum for interesting discussions on the topic achieved and verified using color fast reference of 1918 Curtiss Jenny US airmail stamp. We also materials. These measurements were consistent with thank Beth Heydt and Kelly Cooper from NPM who a second independent system used to measure color assisted in making the vignette for 24¢ stamp; plate changes following accelerated light exposure under number 8493 (National Postal Museum, object num- laboratory conditions, according to a rigorous meth- ber 0.242263.15824 and 24¢ stamp carmine frame: for od to compare color space coordinates. Additionally, USA Scott C3; plate number 8492 (National Postal digital simulations of these color values allow one to Museum, object number 0.242263.15825) available qualitatively assess such comparisons. for color analysis. Museum exhibition lighting conditions for the inverted Jenny should not exceed 1.2 Mlux·h over the References lifetime of its public exhibition to ensure that color Allen, J. A. and T. Lera 2012. The U.S. 1851 3¢ Stamp changes remain below the just noticeable threshold Color, Chemistry, and Changes, The Institute for of 1.6 ∆Ε* 00 color difference units.  Experimental light Analytical Philately exposure of the Jenny indicated the blue portions of the stamp underwent a color change broadly con- Caswell, L. R. 2012. Reflectance Spectroscopy of sistent with the small differences observed between Colored Overprints, The Institute for Analytical Philately. 68 / Institute for Analytical Philately Color Analysis and Microfade Testing of the . . . Jenny US Airmail Stamp Charles, H. K. 2017. Exploring Color Mysteries in the Correspondence Files 1881 - 1949). Container United States Large and Small Numeral Postage 44 (1918: Paper: Internal Revenue THRU 1918: Due Stamps using X-ray Fluorescence Spectrome- Publications). Folder titled “Postage Stamps: try, The Institute for Analytical Philately. Colors 1918” U. S. National Archives and Re- Charles, H. K. 2020. 1894 and 1895 Series First Bu- cords Administration (NARA). reau Postage Due Stamps: Questions of Color, Flu- Judge, R. H. 2015. The Admiral Issue of Canada: A orescence, and Early Use: Part I. IV, The Institute Colorimetric and XRF Study of the Carmine 2¢ for Analytical Philately. Issue, The Institute for Analytical Philately. Cibulski, J. M. 2015. Resolving the Scanner Dependen- Kirker, J. (2006). “24c Curtiss Jenny single.” Re- cy in Color Matching, The Institute for Analytical trieved 15 October 2022, from https:// Philately postalmuseum.si.edu/object/np- Cibulski, J. M. 2015. Towards a Stamp-Oriented Color m_1980.2493.10009#copy_link. Guide: Objectifying Classification by Color, The Kuzio, O. (2018). Excel Sheet to Calculate DE2000 Institute for Analytical Philately. from CIE L*, a*, b* Cibulski, J. M. 2017. THe Colors of the German Crown Laudato Beltran, V., et al. (2021). Microfade Tester: and Eagle Series: A Tutorial on the Objective Light Sensitivity Assessment and Role in Light- Dermination of Color Varieties, London, UK, The ing Policy. Los Angeles, CA, J. Paul Getty Trust. Institute for Analytical Philately. Lera, T., et al. 2012. A Scientific Analysis of the First DeBlois, D. and R. D. Harris (2011). The Colors of Issues of Chile 1853-1862, London Printing, The Martin Van Buren: An Engraved Postage Stamp Institute for Analytical Philately. (1938-1959). The Prexie Era-The Newsletter of the Lowe, B. J., et al. 2017. “Light-ageing characteristics USSS 1938 Presidential Era Study Group 52. of Māori textiles: Colour, strength and molecular EasyRGB. Retrieved 10/14/2022, 2022, from change.” Journal of Cultural Heritage 24: 60-68. https://www.easyrgb.com/en/convert.php#in- Luo, M. R., et al. 2001. “The Development of the CIE putFORM. 2000 Colour-Difference Formula: CIEDE2000.” Engraving Division, Bureau of Engraving and Print- Color Research and Application 26(5): 340-350. ing. 1960. “Stamp History 24¢ Air Mail Stamps Materials Technology,. (12/22/22). “The Blue Wool (Airplane).” Scale.” 2022, from http://www.uvweather- Ford, B. and J. Druzik 2013. “Mirofading: the state of ing.com/uv_scale.html. the art for natural history collections.” Collections Mckee, A. 2015. Paper and Color Varieties of the Forum 27(1-2): 54-71. People’s Republic of China “Workers and Soldiers” Invoice from Gebr. Heyl & Co. Manufacturers of Definitive Set of 1955-1961, The Institute for Ana- chemical colors to Government of the United lytical Philately States of North-America, Bureau of Engraving Michalski, S. and C. Dignard 1997. “Ultrasonic mist- and Printing. 15 January. RG 0318 (Bureau of ing. Part 1, experiments on appearance change Engraving and Printing) ) A1 12 (Central Cor- and improvement in bonding.” Journal of Ameri- respondence Files 1881 - 1949). Container 35 can Institute for Conservation 36(2): 109-126. (1917. Travel THRU 1917: ACCTS: Govt. Prop. Letter from N. Underwood in Ink Making Division Sold) Folder titled “ACCTS: COLORS” U. S. to Director James Wilmeth. 24 July 1918. RG National Archives and Records Administration 0318 (Bureau of Engraving and Printing) ) A1 (NARA). 12 (Central Correspondence Files 1881 - 1949). Hisey, R. 2020. Single Pixel Colorimetry and Optical Container 44 (1918: Paper: Internal Revenue Densitometry in Philately. , The Institute for Ana- THRU 1918: Publications). Folder titled “Post- lytical Philately. age Stamps: Colors 1918” U. S. National Archives Hofmeyr, J. 2020. A Quantitative Color Analysis of and Records Administration (NARA). the US 3¢ 1861 Issue, Virtual, The Institute for Robert A. Siegel Auction Galleries, I. (2017). The Analytical Philately. Don David Price Award-Winning Collection of Letter from Director James Wilmeth to Mr. Wil- the 1918 24-cent Jenny Air Post Issue Featuring fred A. French. 30 July 1918. RG 0318 (Bureau the Inverted Jenny Position 28 and the Original of Engraving and Printing) ) A1 12 (Central Robey Sale Letter. New York, Robert A. Siegel Thomas Lam, Susan Smith, Scott Devine, Edward P. Vicenzi Institute for Analytical Philately / 69 Auction Galleries, Inc. Container 35 (1917. Travel THRU 1917: ACCTS: Robertson, A. R. 1977. “The CIE 1976 Color-Differ- Govt. Prop. Sold) Folder titled “ACCTS: COL- ence Formulae.” Color Research and Applications ORS” U. S. National Archives and Records 2(1): 7-11. Administration (NARA). Świt, P., et al. 2021. “Beam characterization of a mi- USPS (2022). “Rates for Domestic Letter Since crofading tester: evaluation of several methods.” 1863.” Retrieved 12/20/22, 2022, from https:// Heritage Science 9. about.usps.com/who/profile/history/ Thomson, G. 1978. The Museum Environment. Lon- pdf/domestic-letter-rates-since-1863.pdf. don, Butterworth-Heinemann. Whitmore, P. M., et al. 1999. “Predicting the Fading Letter from Treasury Department to Speaker of of Objects: Identifcation of Fugitive Colorants the House of Representatives. 26 June 1917. RG Through Direct Nondestructive Lightfastness 0318 (Bureau of Engraving and Printing) ) A1 Measurements.” Journal of American Institute of 12 (Central Correspondence Files 1881 - 1949). Conservators: 395-409. Glossary list of terms:(in order of appearance in the text) Candela steradian/square meter – is the SI (Inter- al) was used. national System of Units) unit of illuminance and Full width at half the maximum (FWHM) – a figure luminous emittance in a specific direction and takes of merit for spectral resolution, in this case a one into account the angular distribution of the light. dimensional spatial measure of intensity across a fo- Footcandles –a non-SI unit defined as illuminance cused spot of white light. From such a plot the width on a one square foot surface from a uniform source at half the maximum intensity is defined as the full of light. width at half max. Lux – the SI unit of measure for illuminance, i.e. the ISO Blue wool standards – International Organiza- amount of light falling on a surface per unit area. tion for Standardization (ISO) blue wool standards Region of interest – an area selected for analysis are a series of eight blue dyed woven textiles ranging Microfade testing (MFT) – MFT is a non-contact from blue wool 1 to blue wool 8 that have docu- and “virtually non-destructive” method for deter- mented light sensitivities originally developed for mining small increments of reflectance color shift on the textile industry. The lower number blue wools cultural heritage materials (Ford and Druzik 2013). are the most light sensitive and increasing blue wool The technique uses a light source (originally xenon, numbers require increasing light dosage to fade but now LED systems have been developed), to gen- (Technology). In making MFT measurements from erate a spectral distribution of visible wavelengths blue wools, the topography (height variation) of ≈ 370 nm to ≈ 760 nm (Whitmore, Pan et al. 1999); the woven wools has to be taken into consideration. which for this design was filtered through compo- Measurements must be taken on the top surface of a nents comprising a “Fading Test System” (Newport weave and not from a valley. With the neutral densi- Oriel Corporation), where the light is focused onto ty filter in place, the positioning to the top of weave an object by fiber optics. will result in the brightest spot and can be confirmed through the endoscope magnifier. Dark reference – the background current in a charge coupled device (CCD) measured periodically by use CIE-L*ab Color space coordinates (L*,a*,b*) – of a shutter, and this background is then subtracted stands for the International Commission on Illumi- from spectral measurements. nation (CIELAB) color space coordinates L*, a*, b*. CIELAB color space coordinates of L* is expressed White reference – spectrum for a material that has on the z-axes and represents the lightness to dark- little to no features across the range of wavelengths ness value, the a* is for red/green axis, and b* for the measured. For this study a 99% calibrated Spectralon® yellow/blue axis (shown in Figure 3A). diffuse reflectance standard from Labsphere, Inc., (a proprietary polytetrafluoroethylene (PTFE) materi- D70E /a bI,n DstEit7u6t –e Cfoorl oArn dailfyfteirceanlc Pe hinil atteermly s ofC ∆o𝐸𝐸lo∗𝑎𝑎r𝑎𝑎 A isn acalylcsuisl aatendd uMsiincgro tfhaed feo lTleoswtiinngg eoqfu tahteio .n . . ( CJeIEnn y US Airmail Stamp 1D9E7a6b),: D E76 – Color difference in terms of ∆Ε * ab is cal- and the RT function are all specifically designed to culated using the following equation (CIE 1976): improve the color difference equation for chromatic ∆𝐸𝐸∗ = √(𝐿𝐿∗ − 𝐿𝐿∗𝑎𝑎𝑎𝑎 0 𝑡𝑡)2 + (𝑎𝑎∗ ∗0 − 𝑎𝑎𝑡𝑡 )2 + (𝑏𝑏∗ ∗0 − 𝑏𝑏𝑡𝑡 )2 ( e q . 1 ) d i ff e r e n c e s i n t h e b l u e r e g i o n(e (qL.1u) o, Cui et al. 2001). For eq.1, L*₀, α*₀, b*₀ are L*, a*, and b* collected at The five correction terms implemented within the * time zero, while L* , α* , b* are values after time t has ∆Ε00 formula include the L*, a*, and b* inputs and t t t Feloarp esqe.d1., 𝐿𝐿∗ , 𝑎𝑎∗0 0, 𝑏𝑏∗0 are L*, a*, and b* collected at time zero, wrehqiulei r𝐿𝐿e∗s, a𝑎𝑎 r∗i,g 𝑏𝑏o∗rous description of color; further 𝑡𝑡 𝑡𝑡 𝑡𝑡 are values after tDimEe t h–aThs eela cposelod. details regarding this color space are described by 2000 r difference by DE2000 or ∆Ε * 00 for- Luo and co-authors (Luo, Cui et al. 2001) and are DmEu2la00 i0s – (TLhueo c, oClouri deitf afel.r e2n0c0e1 )b:y DE2000 or ∆𝐸𝐸∗00 formula is (oLutos,i dCeu it heet aslc.o 2p0e0 1o)f: t his paper. (eq. 2) Light fastness – chemical stability of a color to long √( ∆𝐿𝐿 ′ 2 ) + ( ∆𝐶𝐶 ′ 2 ∆𝐻𝐻′ 2) + ( ) + 𝑅𝑅 ∆𝐶𝐶 ′ ∆𝐻𝐻′𝑇𝑇 expos ure of light.The𝑘𝑘𝐿𝐿 c𝑆𝑆o𝐿𝐿 lor diff𝑘𝑘e𝐶𝐶r𝑆𝑆e𝐶𝐶nce by𝑘𝑘 ∆𝐻𝐻Ε𝑆𝑆𝐻𝐻* 𝑘𝑘𝐶𝐶𝑆𝑆𝐶𝐶 𝑘𝑘𝐻𝐻𝑆𝑆 (eq. 2) 𝐻𝐻 00 takes into account the Standard deviation (σ) of the population - the perceptual non-uniformities of the CIELAB color Tsphaec ceo blor differen ∗ statistical spread of an average value of randomly y implemecnet binyg ∆ m𝐸𝐸0a0th teamkeast iincatol fiavcceo cuonrtr tehcet ipoenrsc eptual non-uniformities of the CIELAB co distributed data. Since the deviation can be positive to laosrp sepcatcse o bf yc oimloprl eamttreinbtuintegs m waitthheimn athtiec aflo frimveu clao:r 1re)c tions to aspects of color attributes within tchoer rfoercmtioulna :f o1r) correction for neutral colors using ∆𝐿𝐿′, ∆𝐶𝐶′, an odr ∆n𝐻𝐻eg′a tteirvme,s t; h2e) sap lrigeahdtn oesf st htoe data that results in plus neutral colors using ∆L′, ∆C′, and ∆H′ ′ 2 or min′us2 one or two standard deviations lies within dtearrmknse;s 2s )c ao rlrigechttinoens tse trom d, a(r∆k𝐿𝐿ne)ss; c3o)r ar ecchtrioomn ate cromrr,e ction termth, e( b∆𝐶𝐶𝑘𝑘 𝑆𝑆 𝑘𝑘 o𝑆𝑆u)nd;s 4 o)f ar ahnude ocmor rcehcatinocne to within 68 % (1 σ) or 𝐿𝐿 𝐿𝐿 ∆𝐻𝐻′ 2 95 % 𝐶𝐶 (2𝐶𝐶 σ) percent confidence respectively. term, (; 3) a )ch;r aonm𝑘𝑘 𝑆𝑆 d a5 c) othrree RctTi ofunn tcetriomn,, which 4is) aan h iunete ractive term betw𝐻𝐻 𝐻𝐻 Analysis oefe vna crhiaronmcea (aAndN hOuVeA ) – compares data dciofrfererecnticoens, 𝑅𝑅 ∆𝐶𝐶′ ∆𝐻𝐻′ ter𝑇𝑇m𝑘𝑘, 𝑆𝑆 𝑘𝑘 𝑆𝑆 .; a Nndo t5e) t hthaet cRhroma, which is defined p oasp upulartiitoy no fv aar ciaonlocre;s h /uset,a nwdhaicrdh deviation (σ) of the 𝐶𝐶 𝐶𝐶 𝐻𝐻 𝐻𝐻 T function, which is an interactive term between chroma and group of interests to a control group. refers to the color family, namely, a position on a color wheel; (Laudato Beltran, Pesme et al. 2021), and the 𝑅𝑅 function are all specifically designed to impHroovlem t-hSei dcoalko mr deitffheored𝑇𝑇 n/cPe veaqluuaeti –on m ethod to statisti- fhour ec hdriffomeraetnicc edsif, ferences in the .b Nluoet ere tghioatn c(hLruoom, Ca,u i et al. 20c0a1ll)y. Tcohme fpiavree c oobrrseecrtvioanti otenrsm ws ith a cutoff for dissimi- iwmhpilcehm iesn dteedfi nweidth ains pthuer i∆ty𝐸𝐸 0o∗0f afo cromluolra; hinucelu, dweh tihceh L*, a*, and lba*ri tiny paut t>s a9n5d% r ecqhuainrecse .a rigorous dreefsecrrsip ttoio tnh eo fc coololorr f;a fmuritlhy,e rn dametaeillys, rae gpaorsditiniogn t hoins cao lor space Marelu dxe∙shc r–i bise dm beyg aL luuox a hnodu cros,- which is equivalent to acuotlhoorr ws (hLeueol;, (CLuaiu edt aatlo. 2B0e0l1tr)a ann, dP aersem oeu etsti dale. t2h0e2 s1c)o, pe of this1 ,p0a0p0e,r0. 0 0 lux hours (lux∙h). Light fastness – chemical stability of a color to long exposure of light. Standard deviation () of the population - the statistical spread of an average value of randomly distributed data. Since the deviation can be positive or negative, the spread of the data that results in plus or minus one or two standard deviations lies within the bounds of random chance to within 68 % (1 ) or 95 % (2 ) percent confidence respectively. Analysis of variance (ANOVA) – compares data population variances /standard deviation () of the group of interests to a control group. Holm-Sidak method/P value – method to statistically compare observations with a cutoff for dissimilarity at > 95% chance. Mlux∙h – is mega lux hours, which is equivalent to 1,000,000 lux hours (lux∙h).