Introduction

Shopping bags represent cycles of invention, patent, and fashion, and their history is rich with contributions from printers, artists, engineers, and designers. By the late twentieth century, shopping bags were ubiquitous and symbolic of consumer culture; they were carriers for purchased goods and canvases for design messages. Art collectors recognized the artistic value of shopping bags by the 1960s, when Pop artists such as Andy Warhol (1928–1987), Roy Lichtenstein (1923–1997), and Robert Indiana (1928–2018) began producing designs printed on shopping bags and sold for accessible prices, and shopping bags have since been collected alongside traditional art objects1,2. Curated exhibitions began examining shopping bags with artists’ designs as early as 1987 in Art To Go: An Exhibition of Shopping Bags, curated by Fred B. Adelson at Glassboro State College3 and as recently as 2017 in Carry On! Selections from the J. Scott Patnode Shopping Bag Collection at the Cary Graphic Arts Collection at the Rochester Institute of Technology4.

In 1991, the Nevada Museum of Art presented It’s on the Bag, an exhibition of shopping bags belonging to the curator and art historian J. Scott Patnode that aimed to bridge the gap between functional and fine art by celebrating the shopping bag as an accessible reflection of consumerism and pop culture at the close of the twentieth century5. Many of the bags in the exhibition were printed with reproduced or commissioned imagery by popular contemporary artists. One such artist was the American painter, printmaker, and sculptor Frank Stella (1936–2024). In the early 1980s, Stella was commissioned to design a composition that would be subsequently printed as a limited edition of 35,000 shopping bags and distributed by the Minnesota-based chain of Dayton’s department stores to promote the 1984 expansion of the Walker Art Center in Minneapolis, Minnesota5. The officially untitled shopping bag is sometimes referred to as the Dayton’s Shopping Bag. However, for the Nevada Museum of Art exhibition, Stella’s shopping bag was named Tyler Graphics Bag (Fig. 1) after Tyler Graphics Ltd., the experimental fine art printing studio with which Stella frequently collaborated.

Fig. 1: Frank Stella, Tyler Graphics Bag (1984).
figure 1

White Kraft paper with energy-cured coating; pulp-dyed twisted paper handle (47.3 × 40.6 cm). Image showing both sides of the printed design. Images acquired by Carlsmith and Lee. Image rights: Tyler Graphics Bag © 2023 Artists Rights Society (ARS), New York. All reproductions of the work(s) are excluded from the CC-BY License.

Full size image

Concurrent with the production of the bag and advertised on its narrow side panels, the premiere exhibition in the Walker Museum’s new gallery space was a selection of printed works produced at Tyler Graphics by Stella’s long-time collaborator, American master printer Kenneth Tyler (b. 1931). Much of Stella’s printing was facilitated by Tyler, who frequently collaborated with artists including Stella, Lichtenstein, Robert Rauschenberg, David Hockney, Helen Frankenthaler, and Ellsworth Kelly from the early 1970s until he closed his studio in 20016. Tyler pioneered many new printing systems, even designing new presses during his career, and was at the forefront of experimental techniques, especially in the last quarter of the twentieth century7.

Throughout his long working relationship with Tyler Graphics (Fig. 2), Stella employed various printing techniques and materials, including etching, aquatint, engraving, relief, screen printing, and lithography, often in combination. Reportedly, “Tyler and Stella [would] confer frequently as to various print media used to transpose the various elements of collage into the final editioned prints. ‘The collage is what it is,’ Stella remarked… ‘but the print is, in a way, better. It’s the finalized version…. There’s a record of change in the collage and the print is a record of completion’”8. Stella also often reused studio debris from his painting and sculpture practices as printing matrices and iterated on compositions he had previously explored. The imagery on the Tyler Graphics Bag is related to Stella’s Circuits series (1982–1984), particularly the Imola Three prints, which Tyler Graphics also produced. Prints in the Circuits series included techniques such as woodcut, stencil, and screenprint in dizzying layers of experimentation that varied among works and would come to characterize Stella’s printing oeuvre.

Fig. 2: Frank Stella and Kenneth Tyler working on a collage, Tyler Graphics Ltd., Mount Kisco, New York, 1988.
figure 2

Photo by Marabeth Cohen-Tyler. Image courtesy of National Gallery of Australia. All reproductions are excluded from the CC-BY License.

Full size image

The Tyler Graphics Bag is constructed from thick white paper and includes color-printed imagery on its five exterior sides. The bag displays two abstract compositions by Stella: one predominantly red, the other predominantly blue-green. These images are printed on the two widest panels, while promotional information about the Walker Art Center’s new galleries in a printed facsimile of Stella’s hand and Dayton’s logo appear on the narrow side panels. The print is full bleed, including not just the image area but a yellow printed margin that extends to the edges of the paper. Affixed to the top of the shopping bag with metal staples are two looped handles composed of twisted pulp-dyed paper.

Prints from the Tyler Graphics Bag edition are held in the collections of institutions including the Cooper Hewitt, Smithsonian Design Museum, and the Conservation Center of the Institute of Fine Arts, New York University (NYU) Study Collection, and they are often listed for sale on the secondary market. While often described as an offset lithograph, the Tyler Graphics Bag is characterized by the Cooper Hewitt, Smithsonian Design Museum as white Kraft paper with an electron beam (EB) coating printed by AG Industry Co., Ltd.9. While the source of this information could not be confirmed, the specificity of this medium line was intriguing, and two prints from the Tyler Graphics Bag edition held in the NYU Study Collection were analyzed to characterize the support, printing techniques, and colorants used to produce the shopping bags.

Experts at the Kenneth Tyler Archives at the National Gallery of Australia (NGA) were consulted to unravel the printing processes and materials used to produce the shopping bag. Curatorial staff noted that the NGA had not encountered any evidence of Tyler using digital tools before Tyler Graphics Ltd. closed in 200110. While artists were known to use computers and digital imaging in collaboration with Tyler Graphics in the 1980s, NGA consultants clarified that “it is unclear if this was Tyler equipment or used and brought in by the artists”11,12.

The Kenneth Tyler Archives holds a receipt from Tyler Graphics for printing the Tyler Graphics Bag design (Fig. 3). This document describes the different colors used for the prints and also mentions the use of Mylar, implying that the printing process involved a photographic transfer, most likely an aluminum plate used for an offset lithographic print10. Optical microscopy revealed only five colorants on the Tyler Graphics Bag. However, according to this receipt, Tyler (or, rather, his employee Lee Funderburg, whose initials appear under the printer column) printed nine different colors: the print contained multiple green and yellow inks, even a “silver copper” ink was employed, and printing took place over a week. This receipt suggests that different design elements were printed individually, as might be done with a multi-process print of the type Stella sometimes made in collaboration with Tyler8. While perhaps the five sheets of Mylar mentioned in the supplies list could account for some version of halftone printing, it appears that DayGlo colors had also been tested. While Stella often employed daylight fluorescent pigments like DayGlo colors in his paintings, none were clearly evident in the final version of the Tyler Graphics Bag. Furthermore, the “R” column on the receipt means “right to print”; this field is blank, meaning that while all the colors on the receipt were proofed, none were confirmed for printing10. Thus, this receipt likely reflects the production of earlier multi-plate, multi-process prints upon which the halftone color-separated final Tyler Graphics Bag print was later based.

Fig. 3: Scanned receipt for print job from Tyler Graphics Ltd., “Frank Stella Dayton’s Shopping Bag/Walker Show,” April 10, 1984.
figure 3

(Left) first page with paper-clipped Polaroid of a proof attached. (Center) second page showing print dates and numbers, colors used (including “Red Split”, “Black Line”, “Yellow Green”, “Grass Green”, “Silver Copper”, “Yellow Gold”, “Yellow Border”, “Blue”, and “Violet”), printer Lee Funderburg’s initials, and no colors indicated as “right to print”. (Right) third page showing the list of materials used, including “10 Sheets Arches Cover [paper]”, “1 gal[lon] blanket wash”, “2 lbs test Day-Glo + Trans[parent] base”, “5 Sheets Mylar”, and “1 Sheet 24 ×36 plywood”. Image courtesy of National Gallery of Australia. All reproductions are excluded from the CC-BY License.

Full size image

In addition to aiding the investigation of the printing techniques used for the Tyler Graphics Bag, this receipt further exemplifies the highly iterative nature of Stella’s printing practice. The Polaroid photograph of the proof print, which is paper-clipped to the receipt (Fig. 3), reveals significant differences from the design on the final shopping bag. The colors and shapes in the predominantly red composition changed between iterations; for instance, the violet swath at the top of the composition in the Polaroid photograph does not appear on the final Tyler Graphics Bag, and the shape of the central “scribble” motif was adjusted between versions.

In a photo slide from 1984 in the Kenneth Tyler Archives, Frank Stella appears in his studio holding what seems to be the Tyler Graphics Bag (Fig. 4). Closer inspection reveals that this bag, too, is a proof, as the bleed of the image on the bag shown in the photograph is wider than in the final print, and the edges of the proof print extend beyond the paper bag itself.

Fig. 4: “Frank Stella looking at the camera as he uses a dry paintbrush on his ‘Dayton’s Shopping Bag,’ in the artist’s studio, Bedford Village, Westchester County, New York, United States of America, 1984.”
figure 4

Digital File from 35 mm color slide, gift of Kenneth Tyler 2002. All reproductions are excluded from the CC-BY License.

Full size image

Further research into the Kenneth Tyler Archives holdings at the National Gallery of Australia yielded over a dozen alternative proof prints for the Tyler Graphics Bag, with media lines of “colour offset printed proof” along with “additions” ranging from acrylic to crayon13, affirming that the final shopping bag results from extensive experimentation and multiple iterations. Among these archival materials is a carved woodblock (Fig. 5) displaying the portion of the central design that is shared by both prints in different color palettes; this woodblock, examined alongside the multitude of proof prints, supports the conclusion that the final shopping bag design is the photomechanical reproduction of a woodblock that was printed in two color palettes, embellished with other print media, finished with hand-drawn scribbles in acrylic and/or wax-based media, and later commercially reproduced. Furthermore, the Cooper Hewitt, Smithsonian Design Museum records state that the printer of the Tyler Graphics Bag was a company called AG Industry Co., Ltd. While the source of this information is unknown, it similarly suggests that Stella designed the compositions and a print was made at Tyler Graphics, but that the actual printing of the bags was outsourced to an industrial printing company that could accommodate an order of 35,000 shopping bags.

Fig. 5: Frank Stella, woodblock used to print Tyler Graphics Bag proof images (61 × 50.8 × 1 cm).
figure 5

Image rights: Frank Stella© 2023 Artists Rights Society (ARS), New York. All reproductions are excluded from the CC-BY License.

Full size image

While archival research was vital for understanding the production of the work and contextualizing scientific findings, employing advanced analytical techniques was key to unraveling Stella’s notorious experimentation and confirming the printing processes used to produce the Tyler Graphics Bag.

The first phase of the research relied on multispectral imaging (MSI), which has been demonstrated as a suitable technique for characterizing modern colorants14. Subsequent sampling for analysis with micro-Fourier transform infrared spectroscopy (µ-FTIR) was employed to determine the paper type15 and printing inks16, and Raman spectroscopy was used to characterize the colorants decisively17; both techniques were essential to fully describe the printing process used for the Tyler Graphics Bag. Additionally, large-area micro-X-ray fluorescence (XRF) scanning was used to map the diagnostic elemental signatures of the pigments identified by sampling techniques18.

Since shopping bags and other mass-produced items have demonstrably come to be considered collectors’ items and art objects, there is a need for thoughtful recommendations that combine chemical analysis with microfade testing (MFT) to maintain sufficient access to these objects while ensuring their safe and effective display. To that end, MFT was conducted to evaluate the light sensitivity of the printing inks and paper support. In recent years, MFT has become indispensable when considering the effects of light on collection objects during display19.

Methods

Optical microscopy

The Tyler Graphics Bag was examined using an Olympus SZX9 microscope with a Jenoptik Gryphax camera and software with objective magnification[s] of ×5, ×10, ×20, and ×50.

Multispectral imaging (MSI)

Images at different spectral bands were acquired with a modified Nikon D610 DSLR camera, and images were processed using Adobe Photoshop. The MSI suite includes a visible light (VIS) image captured with two Connolux White LEDs and a Midopt 550 filter; an ultraviolet-induced fluorescence (UVF) image captured with two Triple Bright II UV Lamps, a Midopt 550 filter, and a Kodak 2E Wratten gel filter; an ultraviolet reflectance (UVR) image captured with two Triple Bright II UV Lamps, an X-Nite BP1 filter, and an X-Nite 330 filter; an infrared reflectance (IRR) image captured with two Lowel Scandles Tungsten Lamps and an X-Nite 830 filter; a visible-induced luminescence (VIL) image captured with two American DJ LED Par Cans and an X-nite 830 filter; a 660 nm bandpass image captured with two Lowel Scandles Tungsten Lamps and a Midopt 660 filter; and a 735 nm bandpass image captured with two Lowel Scandles Tungsten Lamps and a Midopt 735 filter. Using Adobe Photoshop, a multiband reflectance (MBR) image was produced by subtracting the 660 nm bandpass and 735 nm bandpass images; an infrared false-color (IRR-FC) image was produced by channel mixing the VIS and UVR images; and an ultraviolet false color (UVR-FC) image was produced by channel mixing the VIS and IRR images.

Sampling

Physical, microscopic samples were acquired with a scalpel under magnification from sites where individual inks could be most isolated. Successful sampling sites were identified along the borders of printed color fields adjacent to white, uninked areas of the paper, as the colorants are overlaid in most other parts of the compositions.

Raman spectroscopy

Samples were analyzed using a Renishaw In-via Raman system equipped with a 785 nm diode laser operated between 0.3 to 3 mW, an 830 lines/mm grating, and a Leica confocal microscope with a 50× LWD objective. Spectra were acquired for 10 s and averaged over five accumulations. Spectra were examined using the Thermo Scientific OMNIC 9.0 software package, and sample identification was achieved by comparison of the unknown spectrum to published reference spectra.

Micro-Fourier transform infrared spectroscopy (µ-FTIR)

Samples were analyzed in transmission mode using a Nicolet iS50-μ-FTIR coupled with a Thermo Nicolet Continuum infrared microscope equipped with an MCT-A detector. Spectra were collected in the 4000–600 cm−1 range with a 4 cm−1 resolution and 128 scans. Spectra were examined using the Thermo Scientific OMNIC 9.0 software package, and sample identification was achieved by comparison of the unknown spectrum to published reference spectra.

Large-area micro-X-ray fluorescence (XRF)

Elemental distribution mapping was conducted using a Bruker M6 Jetstream with a rhodium target and 30 mm2 silicon drift detector (SDD) at 50 kV and 600 μA, a 550 μm spot size, a 550 μm pixel size, and a dwell time of 10 ms/pixel under ambient conditions. The spectral maps were processed using the Bruker M6 Jetstream software. A hollow, rigid armature was constructed to fit inside the bag during scanning XRF analysis, holding it open and permitting analysis of just one printed side and minimizing the interference of elemental data from the opposite wall of the three-dimensional bag. Nevertheless, the face of the bag was not completely flat; some concavity in the lower region of the bag produced a greater distance between the bag and the scanning XRF measurement head that resulted in some unevenness along the bottom of the elemental maps.

Microfade testing (MFT)

Assessment of light sensitivity was performed using an Instytut Fotonowy automated microfade tester with a spot diameter of 0.5 mm and a 3000 K LED light source, and the current was set at 900 μA. The integration time was set at 6 ms and averaged over 3 readings. Each test occurred over 600 s for a total light exposure of 1.65 Mlx h. Data was acquired using the included Instytut Fotonowy MFT software, and color change was reported using the CIE Lab ΔE00 equation. Three representative test sites were chosen per printed area of predominantly cyan, red, yellow, and gray ink, and the uninked white paper substrate was tested as well; no areas of black were sufficiently isolated, and no MFT data is reported for that colorant. MFT of Blue Wool (BW) Standards 1, 2, and 3 was also conducted. Test sites were determined through visual examination with optical microscopy, and ΔE00 curves and spectral features were investigated using Getty Spectral Viewer v.3.0.0.62.

Results

Microscopic examination

The visual appearance of the Tyler Graphics Bag suggests that multiple techniques were used in its creation, even if not in its final offset production. To better understand the processes reflected in the Tyler Graphics Bag and to characterize the materials, the printed composition was studied using analytical techniques to complement the archival documentation.

To more accurately characterize the printing processes employed in the production of the Tyler Graphics Bag, the printed abstract compositions were first examined with optical microscopy. Unlike traditional stone lithography, offset lithography is a photomechanical and indirect printmaking process in which the image is transferred to another medium—often a rubber blanket mounted on a cylinder—before it is printed on the substrate. The process was pioneered in the late nineteenth and early twentieth centuries, and many types of offset presses exist. An offset lithographic print may be wet or waterless; the printing substrate may be a single sheet or a roll, and the plates may be metal or polyester11. Inks used in offset lithography are typically quite viscous, traditionally oil-based for tackiness, and some synthetic polymer inks may be heat-set, cold-set, or energy-curable20. Planographic prints, like lithographs, can be identified with microscopy, as the paper and ink sit on the same plane21,22.

Under magnification, it became apparent that the ink sits on the same plane as the paper, suggesting that a relief printing technique was not used to produce the imagery. Furthermore, the microscopic investigation revealed that the composition was printed with a halftone dot pattern and color separation (Fig. 6). This suggests that the Tyler Graphics Bag was produced with a photomechanical printing technique. Based on the subtractive color model, most photomechanical printing is achieved with a four-color separation of cyan, magenta, yellow, and black (CMYK). However, optical microscopy revealed that the Tyler Graphics Bag is rendered in the four standard CMYK process colors and uses a fifth color: gray. In general, additional colors used in the CYMK process, such as the gray ink found in the Tyler Graphics Bag, are called “spot colors” and are often lighter versions of the CMYK “process colors”23. Because the gray and black printed dots overlap, they could not have been printed from the same plate, and evidently, the process of printing the Tyler Graphics Bag included what appear to be five distinct inks.

Fig. 6: Photomicrograph detail of the Tyler Graphics Bag illustrating the five ink colors: cyan, magenta, yellow, black, and gray.
figure 6

Image by Carlsmith and Lee (2023). Image rights: Tyler Graphics Bag © 2023 Artists Rights Society (ARS), New York. All reproductions of the work(s) are excluded from the CC-BY License.

Full size image

Although the work was clearly printed photomechanically, the printed imagery on the Tyler Graphics Bag also has visual qualities associated with stone lithography, including the pastel-like strokes at the center (Fig. 7a). In traditional printmaking, marks such as these could have been produced by drawing on a lithographic limestone with a greasy crayon. Additionally, the images appear to bear a wood grain, which could suggest using a woodcut relief printing technique to reach the final image (Fig. 7b).

Fig. 7: Photomicrograph details of the Tyler Graphics Bag illustrating possible different printing techniques.
figure 7

In a pastel-like strokes and b wood grain pattern are visible Images acquired by Carlsmith and Lee.

Full size image

Most contemporary color separation is done digitally, but in early offset printing, plates were coated with a light-sensitive emulsion or filter, exposed to light through a halftone negative at different angles, developed, and printed11. Since analog and digital printing can employ halftones, the Tyler Graphics Bag dot matrix alone did not confirm the specific printing technique used to produce the images. Furthermore, these shopping bags were produced during a transitional period in digital printing technology, when laser printing became increasingly available to printmakers23.

Material characterization

A preliminary examination of the bags with MSI revealed some qualitative characteristics of the colorants; however, since the images on the bags are rendered in color-separated halftones, the overlapping dot matrices complicated the complete isolation of individual colorants. Ultimately, spectroscopic analysis of samples was necessary to identify both the printing process and inks used conclusively, and a multi-analytical investigation that included Raman spectroscopy, µ-FTIR, and XRF scanning was conducted.

The printed passages of the Tyler Graphics Bag that appear dark gray and black— or are a darker shade of cyan, yellow, and magenta— in standard visible light (Fig. 8a) also exhibit strong absorbance in the IRR image (Fig. 8b); this response suggests the presence of a carbon-based black colorant, likely Pigment Black 7 (PB 7), the most common type of black pigment in commercial printing inks20. The printed color fields containing cyan dots appear red in the infrared false-color image (Fig. 8c), suggesting the presence of IR-reflective copper phthalocyanine blue (Pigment Blue 15; PB15)14. This blue pigment remains the dominant blue colorant in CYMK printing today20,24. Identification of the yellow and magenta pigments was not as direct with MSI. Further analysis with Raman spectroscopy (Fig. 9 and Table 1) was required to identify them as the diarylide Pigment Yellow 12 (PY12) and the monoazo lake Pigment Red 52 (PR52)17. These pigments are used appreciably in various industrial printing methods and are not characteristic of any particular formulation. While the absence of fluorescence under ultraviolet radiation (Fig. 8d) strongly indicates that DayGlo inks are not observed in the printed imagery, instrumental analysis pointed against to their use in the final printing25.

Fig. 8: Technical Imaging of the Tyler Graphics Bag with MSI.
figure 8

Each image illustrates a different modality: a Standard visible light (VIS), b infrared reflectance (IRR), c infrared reflectance false-color (IRR-FC), and d ultraviolet fluorescence (UVF). Images by Carlsmith and Lee (2023). Image rights for a: Tyler Graphics Bag © 2023 Artists Rights Society (ARS), New York. All reproductions of the work(s) are excluded from the CC-BY License.

Full size image
Fig. 9: Raman spectra (from bottom) of cyan, yellow, and magenta samples from the Tyler Graphics Bag.
figure 9

The spectra confirm the use of printing inks that contain PB15, PY12 and PR52:1, respectively (peak wavenumbers in cm−1 listed in Table 1).

Full size image
Table 1 µ-FTIR and Raman peaks for the support, ink, and pigments identified in the Tyler Graphics Bag
Full size table

Examining the printed surface of the Tyler Graphics Bag with XRF scanning revealed the distribution of elements detectable by XRF: copper in PB15, chlorine in PY12, and likely calcium in PR52 (Fig. 10); copper in PB15 originates from the central copper(II) ion (Cu²+) in copper phthalocyanine blue, chlorine in PY12 from the dichlorobenzidine skeleton of the pigment, and calcium in PR52 indicates the presence of a calcium-based lake. The relative intensity of each element can roughly correspond to the relative overall concentration of each of these pigments within the composition and is a helpful guide to survey the pigment density and extent of layering in the composition resulting from the halftone printing process. The gray color fields in the composition display particularly strong chlorine signals and indicate a higher density of PY12 overlap over the gray spot color. The elemental maps display an abundance of calcium in the printed color fields containing magenta dots, allowing the magenta pigment to be further identified as the calcium-based PR 52:1 instead of the manganese-based PR 52:226. Together, optical microscopy, MSI, Raman spectroscopy, and XRF scanning provided a clear image of the overall composition of the printed imagery on the Tyler Graphics Bag.

Fig. 10: Elemental distribution maps acquired with XRF scanning of the green-blue abstract composition on the Tyler Graphics Bag.
figure 10

The maps illustrate the location of the different pigments found in the CYMK printing inks: copper (Cu) in PB15, chlorine (Cl) in PY12, and calcium (Ca) in PR52:1. Image rights: Tyler Graphics Bag © 2023 Artists Rights Society (ARS), New York. All reproductions of the work(s) are excluded from the CC-BY License.

Full size image

Spectroscopic analysis with µ-FTIR of the same samples used for Raman spectroscopy determined that the paper fibers are composed mainly of cellulose (Fig. 11 and Table 1)27. However, some diagnostic bands for lignin (C−H stretching modes in methyl, methylene, and methoxyl groups at 2922 cm−1 and 2852 cm−128 and −CH bending mode at 1317 cm−127) indicate that the bag is likely a Kraft paper derived from wood pulp, consistent with the medium of the shopping bag as described by the Cooper Hewitt, Smithsonian Design Museum. The Kraft chemical pulping process removes approximately 80% of the lignin from the raw pulp and other undesirable components and produces stronger paper widely used for packaging29. The bright white color of the uninked paper also suggests that the paper pulp was bleached30. Additionally, in the UVF image of the Tyler Graphics Bag (Fig. 8d), an overall blue-white fluorescence can be observed that is particularly pronounced in the uninked and sparsely colored portions of the composition and is likely related to the use of fluorescent optical brighteners in the white paper substrate31. The presence of titanium was confirmed across the paper substrate by XRF scanning, suggesting that titanium white (Pigment White 6; PW6) was used as a filler to opacify and brighten the bleached paper pulp32. Bleached Kraft paper pulp allows for very lightweight papers with excellent stiffness and favorable attributes for shopping bags30.

Fig. 11: μ-FTIR spectra obtained from cyan, yellow, and magenta samples taken from the Tyler Graphics Bag.
figure 11

The diagnostic bands for epoxy acrylate highlighted in gray (peak wavenumbers in cm−1 listed in Table 1).

Full size image

The µ-FTIR spectra further reinforce the conclusion that the Tyler Graphics Bag was mass-produced: several diagnostic peaks for an energy-cured ink based on an epoxy acrylate polymer (Fig. 11 and Table 1) were identified alongside cellulose and lignin bands. Some characteristic bands for the cyan (PB15) and magenta pigments (PR52) were identified in the µ-FTIR spectra as well, confirming that the samples were indeed representative of the printing ink16. The µ-FTIR spectrum of an epoxy acrylate is characterized by C−H stretching modes at 2960 cm−1 and 2871 cm−1, a C=O stretching mode at 1730 cm−1, a CH2=C stretching mode at 1635 cm−1, aromatic breathing modes at 1610 cm−1 and 1507 cm−1, an in-plane CH2=CH bending mode at 1408 cm−1, C−O−C stretching modes at 1295 cm−1 and 1248 cm−1, an out-of-plane CH=C−H bending mode at 810 cm−1, and aromatic C−H bending modes at 762 cm−1 and 752 cm−112,33,34. While the mode at 830 cm−1 could be indicative of oxirane deformations from incomplete epoxy ring opening during polymerization with acrylic acid, the absence of an additional oxirane stretching mode ca. 910–920 cm−1 also confirms the successful synthesis of the polymer35,36; therefore, the mode at 830 cm−1 likely is an additional aromatic C−H bending mode. Furthermore, the severe attenuation of the vinylic acrylate modes at 1635 cm−1, 1408 cm−1, and 810 cm−1 due to C=C groups reacting with each during irradiation also points to complete curing at the time of printing37,38. Inks based on epoxy acrylates are used readily in energy-curable processes, which are favored techniques for industrial purposes due to their facilitation of fast, high-volume printing with great accuracy and consistency. However, in practice, there are not many outstanding distinctions between EB and UV performance for most of the popular resins in use, and selection is probably best based on the end-use; UV-curable, sheet-fed lithographic printing is particularly well-suited for the coating of paper and board substrates20.

Evaluation of lightfastness

Though the Tyler Graphics Bag was likely industrially produced in a manner consistent with printed works designed to be disposable, it has entered institutional collections alongside what are more traditionally described as “fine art” prints. The Tyler Graphics Bag and more bespoke printed works on paper have dissimilar origins but are in a position today to share gallery spaces; thus, it was determined that the light sensitivity of the shopping bag should be evaluated to ascertain whether current museum lighting recommendations for works on paper are suitable for the storage and display of mass-produced works on paper such as the Tyler Graphics Bag.

MFT was performed to classify the light sensitivity of the printed images in terms of ΔE00. Densely printed color fields with a predominantly isolated colorant were chosen as test sites to obtain the most representative MFT data. The cyan, yellow, and magenta inks and the uninked white paper substrate were analyzed separately. No predominantly gray or black areas were suitable for MFT due to the significant overlap of multiple colorants in each gray and black color field. The results were compared to Blue Wool (BW) 1, 2, and 3 Standards to classify the overall lightfastness of each colorant. The overall change curves characterized by ΔE00 (Fig. 12) reveal that the yellow, magenta, and cyan inks rank from most to least lightfast in that order. The lightfastness of the yellow ink can be classified as more lightfast than that of BW3, whereas the cyan and magenta inks fall between BW2 and BW3, though both are much closer to BW3. None of the colorants, however, surpassed the just noticeable difference (JND) threshold of ΔE00 = 1.5 during the experiment39.

Fig. 12: MFT-induced color change (ΔE00) curves with an inset illustrated for clarity.
figure 12

The shaded area around each curve indicates the standard deviation, and BW1 was not included.

Full size image

In the CIE L*a*b* color space, shifts along the L*, a*, b*, and C* axes can better qualify possible changes, where ΔL* refers to light-dark shifts, Δa refers to green-red shifts, Δb* refers to blue-yellow shifts, and ΔC* refers to shifts in the saturation of the measured color as a function of a* and b* (Fig. 13). All colorants on the Tyler Graphics Bag, except for the cyan ink, demonstrated negative Δa* and Δb* values, shifting to cooler hues. The combined positive Δa* and negative Δb* for the cyan colorant corresponds to a deepening of the blue shade overall, potentially shifting towards a more purple hue. Whereas the cyan colorant shifted toward red, all other colorants on the Tyler Graphics Bag shifted slightly toward green. All colorants and the uninked paper substrate demonstrated negative ΔL* and ΔC* values, indicating overall desaturation and loss of vibrancy.

Fig. 13: ΔL*, Δa*, Δb*, and ΔC* for the cyan, magenta, and yellow inks, in addition to the paper support.
figure 13

The CIE L*a*b* color space is represented for reference in the lower right corner.

Full size image

The MFT data may be extrapolated to predict the desaturation, darkening, and cooling of the printed imagery on the Tyler Graphics Bag as it experiences light exposure. Type II fading shows rapid initial change followed by a slower, constant rate of change with continued exposure, whereas Type III change is characterized by a linear or constant rate of change commonly observed with lightfast pigments that form larger aggregates inside the fiber40. The color change can plateau over time, as demonstrated by the Type II change observed for the red and yellow colorants; the cyan passages, however, appear to retain a linear slope akin to Type III change after an initial rapid change, requiring further investigation and longer exposure times.

The Tyler Graphics Bag prints in the NYU Study Collection were acquired from a private collection, and their previous storage and display histories are unknown. Prior light exposure can significantly affect the lightfastness and, subsequently, the results from MFT analysis of the colorants used in CYMK printing41 or otherwise42. Even though PB15 is rated as highly lightfast, the Tyler Graphics Bag prints in the NYU Study Collection may have experienced their most dramatic light-induced color change before they were subjected to accelerated aging, including possibly being exposed to illumination sources emitting UV radiation. Such previous light exposure could be why magenta, the pigments of which are generally less lightfast43, exhibited less change in ΔE00 relative to the PB15-based cyan ink.

Cyan inks have shown variable responses to light exposure that could depend on the substrate44,45 or on temperature and humidity46. However, these studies have not been carried out with MFT. Most directly related was the analysis carried out by Ford and Smith on two commercial CMYK printed posters of the Coulter Panorama (1911, National Museum of Australia), one in pristine condition and another print that had experienced substantial prior light exposure, with trace amounts remaining of magenta ink, very little cyan ink, and unaffected passages with yellow ink41. In the analysis of both pristine and exposed prints, cyan also appeared to be light sensitive, on the order of BW1 and BW2, respectively; the yellow colorant was the least sensitive and not affected dramatically by prior exposure. The results from Ford and Smith mirrored those seen in the MFT of the Tyler Graphics Bag, and although the specific pigments are unknown, the cyan ink is likely made with PB15, and the yellow ink contains a mono- or diazo arylide pigment, given common industrial formulations20. It is worth investigating in future work if the increase in b* for the cyan ink may result from the bleaching of the epoxy medium and/or paper support rather than changes to pigment itself, considering the high light stability of PB15.

Nevertheless, these results, and the shopping bags themselves, challenge notions of how works on paper are considered in museums and how they may be safely displayed. MFT data can be used practically to establish lighting recommendations for display and resting periods and to anticipate objects’ lifetimes, given that the principle of reciprocity is upheld39. During an MFT measurement, each test site experienced a light dosage of 1.65 megalux hours (Mlx h). Considering that the current recommended light level for displaying works on paper is 50 lux47, and most museums and galleries are open and lit for an average of fewer than 3000 h annually, the Tyler Graphics Bag would, therefore, need to be placed on uninterrupted display for 11 years to undergo a color change equivalent to 1.65 Mlx h hours, still less than required to surpass JND of ΔE00 = 1.5.

Overall, the MFT results suggest that the Tyler Graphics Bag and its colorants are fairly light-stable under gallery lighting that filters both ultraviolet and infrared radiation. Therefore, it could be appropriate to exhibit the shopping bag under brighter light levels that facilitate better viewing for shorter periods without reaching JND, especially if the work is part of a regularly rotated collection, or it could also be displayed under lower lighting for more extended periods48; a combination of both approaches is certainly feasible as well. Ultimately, the MFT data demonstrates flexibility when displaying the Tyler Graphics Bag. Moreover, most commercial shopping bags are also three-dimensional and printed on both sides, providing an inherent opportunity to rotate the exposed face during ongoing display.

Discussion

Frank Stella initially saw printing as a means of reproducing his paintings as multiples. However, he soon discovered the versatility of print media, which opened up new modes of visual experimentation in his practice. The research described herein is a window into the many forms of making buried in the palimpsests of his prints, both rare and common. These findings also support the conclusion that while the imagery printed on the Tyler Graphics Bag was designed by Frank Stella and produced to some extent with the help of Kenneth Tyler and his team at Tyler Graphics Ltd., the manufacture of the shopping bag itself was almost certainly outsourced to an industrial printing company. By the 1980s, more than 100 organic pigments were used in commercial printing inks, and the colorants identified on the Tyler Graphics Bag through multi-technique analysis were among the most common. The inks were identified as energy-cured epoxy acrylates, a choice consistent with the era’s high-volume, fast-turnaround industrial printing processes. Furthermore, the support material was determined to be bleached Kraft paper with titanium white filler, enhanced with optical brighteners—materials typical of commercial packaging rather than fine art printmaking. This distinction emphasizes the industrial nature of the bag’s production, aligning with the mass-produced intent of a shopping bag while simultaneously carrying the artistic significance of Stella’s design.

Furthermore, evaluating the bag’s lightfastness through microfade testing further expanded the understanding of its longevity. While designed as a disposable item, the study demonstrated that the bag’s materials, including the commercial pigments and binder, exhibit relatively high stability when tested with accelerated aging that simulates gallery lighting conditions. An industrial printing company may have elected to employ these colorants in producing 35,000 Tyler Graphics Bag prints because of the lightfastness and durability of these pigment-based inks for commercial printing purposes. This finding challenges the assumption that mass-produced objects lack the durability of bespoke works and offers flexibility for their display and rotation in institutional settings.

The convergence of these findings clarifies the production history of the Tyler Graphics Bag and prompts a reevaluation of the boundaries between fine art and industrial manufacture. Through printmaking, Stella’s works developed conversations with one another as materials and compositional elements were recycled and modified from work to work. Stella’s willingness to engage with commercial processes expands the dialogue around artistic intent, and the bag embodies a fusion of high art and consumer culture, reflecting the shifting landscapes of artistic production and distribution during the 1980s. The many compositional permutations of the Tyler Graphics Bag perfectly illustrate this iterative practice and prove that outsourcing the Tyler Graphics Bag production does not further distance the artist from the final commercial product. The Tyler Graphics Bag is the finalized version of Stella’s vision, and it holds traces of changes through iteration. Ultimately, whatever else it may be, it is a record of completion.